first commit

.gitattributes

@@ -0,0 +1,5 @@
+lightning_logs/version_0/checkpoints/frn-epoch=65-val_loss=0.2290.ckpt filter=lfs diff=lfs merge=lfs -text
+lightning_logs/version_0/checkpoints/frn.onnx filter=lfs diff=lfs merge=lfs -text
+lightning_logs/predictor/checkpoints/predictor.ckpt filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ckpt filter=lfs diff=lfs merge=lfs -text

LICENSE

@@ -0,0 +1,400 @@
+
+Attribution-NonCommercial 4.0 International
+
+=======================================================================
+
+Creative Commons Corporation ("Creative Commons") is not a law firm and
+does not provide legal services or legal advice. Distribution of
+Creative Commons public licenses does not create a lawyer-client or
+other relationship. Creative Commons makes its licenses and related
+information available on an "as-is" basis. Creative Commons gives no
+warranties regarding its licenses, any material licensed under their
+terms and conditions, or any related information. Creative Commons
+disclaims all liability for damages resulting from their use to the
+fullest extent possible.
+
+Using Creative Commons Public Licenses
+
+Creative Commons public licenses provide a standard set of terms and
+conditions that creators and other rights holders may use to share
+original works of authorship and other material subject to copyright
+and certain other rights specified in the public license below. The
+following considerations are for informational purposes only, are not
+exhaustive, and do not form part of our licenses.
+
+     Considerations for licensors: Our public licenses are
+     intended for use by those authorized to give the public
+     permission to use material in ways otherwise restricted by
+     copyright and certain other rights. Our licenses are
+     irrevocable. Licensors should read and understand the terms
+     and conditions of the license they choose before applying it.
+     Licensors should also secure all rights necessary before
+     applying our licenses so that the public can reuse the
+     material as expected. Licensors should clearly mark any
+     material not subject to the license. This includes other CC-
+     licensed material, or material used under an exception or
+     limitation to copyright. More considerations for licensors:
+	wiki.creativecommons.org/Considerations_for_licensors
+
+     Considerations for the public: By using one of our public
+     licenses, a licensor grants the public permission to use the
+     licensed material under specified terms and conditions. If
+     the licensor's permission is not necessary for any reason--for
+     example, because of any applicable exception or limitation to
+     copyright--then that use is not regulated by the license. Our
+     licenses grant only permissions under copyright and certain
+     other rights that a licensor has authority to grant. Use of
+     the licensed material may still be restricted for other
+     reasons, including because others have copyright or other
+     rights in the material. A licensor may make special requests,
+     such as asking that all changes be marked or described.
+     Although not required by our licenses, you are encouraged to
+     respect those requests where reasonable. More_considerations
+     for the public:
+	wiki.creativecommons.org/Considerations_for_licensees
+
+=======================================================================
+
+Creative Commons Attribution-NonCommercial 4.0 International Public
+License
+
+By exercising the Licensed Rights (defined below), You accept and agree
+to be bound by the terms and conditions of this Creative Commons
+Attribution-NonCommercial 4.0 International Public License ("Public
+License"). To the extent this Public License may be interpreted as a
+contract, You are granted the Licensed Rights in consideration of Your
+acceptance of these terms and conditions, and the Licensor grants You
+such rights in consideration of benefits the Licensor receives from
+making the Licensed Material available under these terms and
+conditions.
+
+Section 1 -- Definitions.
+
+  a. Adapted Material means material subject to Copyright and Similar
+     Rights that is derived from or based upon the Licensed Material
+     and in which the Licensed Material is translated, altered,
+     arranged, transformed, or otherwise modified in a manner requiring
+     permission under the Copyright and Similar Rights held by the
+     Licensor. For purposes of this Public License, where the Licensed
+     Material is a musical work, performance, or sound recording,
+     Adapted Material is always produced where the Licensed Material is
+     synched in timed relation with a moving image.
+
+  b. Adapter's License means the license You apply to Your Copyright
+     and Similar Rights in Your contributions to Adapted Material in
+     accordance with the terms and conditions of this Public License.
+
+  c. Copyright and Similar Rights means copyright and/or similar rights
+     closely related to copyright including, without limitation,
+     performance, broadcast, sound recording, and Sui Generis Database
+     Rights, without regard to how the rights are labeled or
+     categorized. For purposes of this Public License, the rights
+     specified in Section 2(b)(1)-(2) are not Copyright and Similar
+     Rights.
+  d. Effective Technological Measures means those measures that, in the
+     absence of proper authority, may not be circumvented under laws
+     fulfilling obligations under Article 11 of the WIPO Copyright
+     Treaty adopted on December 20, 1996, and/or similar international
+     agreements.
+
+  e. Exceptions and Limitations means fair use, fair dealing, and/or
+     any other exception or limitation to Copyright and Similar Rights
+     that applies to Your use of the Licensed Material.
+
+  f. Licensed Material means the artistic or literary work, database,
+     or other material to which the Licensor applied this Public
+     License.
+
+  g. Licensed Rights means the rights granted to You subject to the
+     terms and conditions of this Public License, which are limited to
+     all Copyright and Similar Rights that apply to Your use of the
+     Licensed Material and that the Licensor has authority to license.
+
+  h. Licensor means the individual(s) or entity(ies) granting rights
+     under this Public License.
+
+  i. NonCommercial means not primarily intended for or directed towards
+     commercial advantage or monetary compensation. For purposes of
+     this Public License, the exchange of the Licensed Material for
+     other material subject to Copyright and Similar Rights by digital
+     file-sharing or similar means is NonCommercial provided there is
+     no payment of monetary compensation in connection with the
+     exchange.
+
+  j. Share means to provide material to the public by any means or
+     process that requires permission under the Licensed Rights, such
+     as reproduction, public display, public performance, distribution,
+     dissemination, communication, or importation, and to make material
+     available to the public including in ways that members of the
+     public may access the material from a place and at a time
+     individually chosen by them.
+
+  k. Sui Generis Database Rights means rights other than copyright
+     resulting from Directive 96/9/EC of the European Parliament and of
+     the Council of 11 March 1996 on the legal protection of databases,
+     as amended and/or succeeded, as well as other essentially
+     equivalent rights anywhere in the world.
+
+  l. You means the individual or entity exercising the Licensed Rights
+     under this Public License. Your has a corresponding meaning.
+
+Section 2 -- Scope.
+
+  a. License grant.
+
+       1. Subject to the terms and conditions of this Public License,
+          the Licensor hereby grants You a worldwide, royalty-free,
+          non-sublicensable, non-exclusive, irrevocable license to
+          exercise the Licensed Rights in the Licensed Material to:
+
+            a. reproduce and Share the Licensed Material, in whole or
+               in part, for NonCommercial purposes only; and
+
+            b. produce, reproduce, and Share Adapted Material for
+               NonCommercial purposes only.
+
+       2. Exceptions and Limitations. For the avoidance of doubt, where
+          Exceptions and Limitations apply to Your use, this Public
+          License does not apply, and You do not need to comply with
+          its terms and conditions.
+
+       3. Term. The term of this Public License is specified in Section
+          6(a).
+
+       4. Media and formats; technical modifications allowed. The
+          Licensor authorizes You to exercise the Licensed Rights in
+          all media and formats whether now known or hereafter created,
+          and to make technical modifications necessary to do so. The
+          Licensor waives and/or agrees not to assert any right or
+          authority to forbid You from making technical modifications
+          necessary to exercise the Licensed Rights, including
+          technical modifications necessary to circumvent Effective
+          Technological Measures. For purposes of this Public License,
+          simply making modifications authorized by this Section 2(a)
+          (4) never produces Adapted Material.
+
+       5. Downstream recipients.
+
+            a. Offer from the Licensor -- Licensed Material. Every
+               recipient of the Licensed Material automatically
+               receives an offer from the Licensor to exercise the
+               Licensed Rights under the terms and conditions of this
+               Public License.
+
+            b. No downstream restrictions. You may not offer or impose
+               any additional or different terms or conditions on, or
+               apply any Effective Technological Measures to, the
+               Licensed Material if doing so restricts exercise of the
+               Licensed Rights by any recipient of the Licensed
+               Material.
+
+       6. No endorsement. Nothing in this Public License constitutes or
+          may be construed as permission to assert or imply that You
+          are, or that Your use of the Licensed Material is, connected
+          with, or sponsored, endorsed, or granted official status by,
+          the Licensor or others designated to receive attribution as
+          provided in Section 3(a)(1)(A)(i).
+
+  b. Other rights.
+
+       1. Moral rights, such as the right of integrity, are not
+          licensed under this Public License, nor are publicity,
+          privacy, and/or other similar personality rights; however, to
+          the extent possible, the Licensor waives and/or agrees not to
+          assert any such rights held by the Licensor to the limited
+          extent necessary to allow You to exercise the Licensed
+          Rights, but not otherwise.
+
+       2. Patent and trademark rights are not licensed under this
+          Public License.
+
+       3. To the extent possible, the Licensor waives any right to
+          collect royalties from You for the exercise of the Licensed
+          Rights, whether directly or through a collecting society
+          under any voluntary or waivable statutory or compulsory
+          licensing scheme. In all other cases the Licensor expressly
+          reserves any right to collect such royalties, including when
+          the Licensed Material is used other than for NonCommercial
+          purposes.
+
+Section 3 -- License Conditions.
+
+Your exercise of the Licensed Rights is expressly made subject to the
+following conditions.
+
+  a. Attribution.
+
+       1. If You Share the Licensed Material (including in modified
+          form), You must:
+
+            a. retain the following if it is supplied by the Licensor
+               with the Licensed Material:
+
+                 i. identification of the creator(s) of the Licensed
+                    Material and any others designated to receive
+                    attribution, in any reasonable manner requested by
+                    the Licensor (including by pseudonym if
+                    designated);
+
+                ii. a copyright notice;
+
+               iii. a notice that refers to this Public License;
+
+                iv. a notice that refers to the disclaimer of
+                    warranties;
+
+                 v. a URI or hyperlink to the Licensed Material to the
+                    extent reasonably practicable;
+
+            b. indicate if You modified the Licensed Material and
+               retain an indication of any previous modifications; and
+
+            c. indicate the Licensed Material is licensed under this
+               Public License, and include the text of, or the URI or
+               hyperlink to, this Public License.
+
+       2. You may satisfy the conditions in Section 3(a)(1) in any
+          reasonable manner based on the medium, means, and context in
+          which You Share the Licensed Material. For example, it may be
+          reasonable to satisfy the conditions by providing a URI or
+          hyperlink to a resource that includes the required
+          information.
+
+       3. If requested by the Licensor, You must remove any of the
+          information required by Section 3(a)(1)(A) to the extent
+          reasonably practicable.
+
+       4. If You Share Adapted Material You produce, the Adapter's
+          License You apply must not prevent recipients of the Adapted
+          Material from complying with this Public License.
+
+Section 4 -- Sui Generis Database Rights.
+
+Where the Licensed Rights include Sui Generis Database Rights that
+apply to Your use of the Licensed Material:
+
+  a. for the avoidance of doubt, Section 2(a)(1) grants You the right
+     to extract, reuse, reproduce, and Share all or a substantial
+     portion of the contents of the database for NonCommercial purposes
+     only;
+
+  b. if You include all or a substantial portion of the database
+     contents in a database in which You have Sui Generis Database
+     Rights, then the database in which You have Sui Generis Database
+     Rights (but not its individual contents) is Adapted Material; and
+
+  c. You must comply with the conditions in Section 3(a) if You Share
+     all or a substantial portion of the contents of the database.
+
+For the avoidance of doubt, this Section 4 supplements and does not
+replace Your obligations under this Public License where the Licensed
+Rights include other Copyright and Similar Rights.
+
+Section 5 -- Disclaimer of Warranties and Limitation of Liability.
+
+  a. UNLESS OTHERWISE SEPARATELY UNDERTAKEN BY THE LICENSOR, TO THE
+     EXTENT POSSIBLE, THE LICENSOR OFFERS THE LICENSED MATERIAL AS-IS
+     AND AS-AVAILABLE, AND MAKES NO REPRESENTATIONS OR WARRANTIES OF
+     ANY KIND CONCERNING THE LICENSED MATERIAL, WHETHER EXPRESS,
+     IMPLIED, STATUTORY, OR OTHER. THIS INCLUDES, WITHOUT LIMITATION,
+     WARRANTIES OF TITLE, MERCHANTABILITY, FITNESS FOR A PARTICULAR
+     PURPOSE, NON-INFRINGEMENT, ABSENCE OF LATENT OR OTHER DEFECTS,
+     ACCURACY, OR THE PRESENCE OR ABSENCE OF ERRORS, WHETHER OR NOT
+     KNOWN OR DISCOVERABLE. WHERE DISCLAIMERS OF WARRANTIES ARE NOT
+     ALLOWED IN FULL OR IN PART, THIS DISCLAIMER MAY NOT APPLY TO YOU.
+
+  b. TO THE EXTENT POSSIBLE, IN NO EVENT WILL THE LICENSOR BE LIABLE
+     TO YOU ON ANY LEGAL THEORY (INCLUDING, WITHOUT LIMITATION,
+     NEGLIGENCE) OR OTHERWISE FOR ANY DIRECT, SPECIAL, INDIRECT,
+     INCIDENTAL, CONSEQUENTIAL, PUNITIVE, EXEMPLARY, OR OTHER LOSSES,
+     COSTS, EXPENSES, OR DAMAGES ARISING OUT OF THIS PUBLIC LICENSE OR
+     USE OF THE LICENSED MATERIAL, EVEN IF THE LICENSOR HAS BEEN
+     ADVISED OF THE POSSIBILITY OF SUCH LOSSES, COSTS, EXPENSES, OR
+     DAMAGES. WHERE A LIMITATION OF LIABILITY IS NOT ALLOWED IN FULL OR
+     IN PART, THIS LIMITATION MAY NOT APPLY TO YOU.
+
+  c. The disclaimer of warranties and limitation of liability provided
+     above shall be interpreted in a manner that, to the extent
+     possible, most closely approximates an absolute disclaimer and
+     waiver of all liability.
+
+Section 6 -- Term and Termination.
+
+  a. This Public License applies for the term of the Copyright and
+     Similar Rights licensed here. However, if You fail to comply with
+     this Public License, then Your rights under this Public License
+     terminate automatically.
+
+  b. Where Your right to use the Licensed Material has terminated under
+     Section 6(a), it reinstates:
+
+       1. automatically as of the date the violation is cured, provided
+          it is cured within 30 days of Your discovery of the
+          violation; or
+
+       2. upon express reinstatement by the Licensor.
+
+     For the avoidance of doubt, this Section 6(b) does not affect any
+     right the Licensor may have to seek remedies for Your violations
+     of this Public License.
+
+  c. For the avoidance of doubt, the Licensor may also offer the
+     Licensed Material under separate terms or conditions or stop
+     distributing the Licensed Material at any time; however, doing so
+     will not terminate this Public License.
+
+  d. Sections 1, 5, 6, 7, and 8 survive termination of this Public
+     License.
+
+Section 7 -- Other Terms and Conditions.
+
+  a. The Licensor shall not be bound by any additional or different
+     terms or conditions communicated by You unless expressly agreed.
+
+  b. Any arrangements, understandings, or agreements regarding the
+     Licensed Material not stated herein are separate from and
+     independent of the terms and conditions of this Public License.
+
+Section 8 -- Interpretation.
+
+  a. For the avoidance of doubt, this Public License does not, and
+     shall not be interpreted to, reduce, limit, restrict, or impose
+     conditions on any use of the Licensed Material that could lawfully
+     be made without permission under this Public License.
+
+  b. To the extent possible, if any provision of this Public License is
+     deemed unenforceable, it shall be automatically reformed to the
+     minimum extent necessary to make it enforceable. If the provision
+     cannot be reformed, it shall be severed from this Public License
+     without affecting the enforceability of the remaining terms and
+     conditions.
+
+  c. No term or condition of this Public License will be waived and no
+     failure to comply consented to unless expressly agreed to by the
+     Licensor.
+
+  d. Nothing in this Public License constitutes or may be interpreted
+     as a limitation upon, or waiver of, any privileges and immunities
+     that apply to the Licensor or You, including from the legal
+     processes of any jurisdiction or authority.
+
+=======================================================================
+
+Creative Commons is not a party to its public
+licenses. Notwithstanding, Creative Commons may elect to apply one of
+its public licenses to material it publishes and in those instances
+will be considered the “Licensor.” The text of the Creative Commons
+public licenses is dedicated to the public domain under the CC0 Public
+Domain Dedication. Except for the limited purpose of indicating that
+material is shared under a Creative Commons public license or as
+otherwise permitted by the Creative Commons policies published at
+creativecommons.org/policies, Creative Commons does not authorize the
+use of the trademark "Creative Commons" or any other trademark or logo
+of Creative Commons without its prior written consent including,
+without limitation, in connection with any unauthorized modifications
+to any of its public licenses or any other arrangements,
+understandings, or agreements concerning use of licensed material. For
+the avoidance of doubt, this paragraph does not form part of the
+public licenses.
+
+Creative Commons may be contacted at creativecommons.org.

PLCMOS/models/plcmos_v0.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:34ca776550b1ef611eca849463ad9de52e4f27b2a26632969ae44b7530d8f934
+size 690971

PLCMOS/models/plcmos_v1_intrusive.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:748d2c08694d46aff31668c00c5182b0338d5441edafb1092cb8ecdfd71d286e
+size 279978

PLCMOS/models/plcmos_v1_nonintrusive.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9aca9f5ddd998f0a304cc063b6620832c37cf65627c242accffb8dc38acb8c79
+size 128653

PLCMOS/plc_mos.py

@@ -0,0 +1,247 @@
+import math
+import os
+
+import librosa
+import numpy as np
+import onnxruntime as ort
+from numpy.fft import rfft
+from numpy.lib.stride_tricks import as_strided
+
+from utils.utils import LSD
+
+
+class PLCMOSEstimator():
+    def __init__(self, model_version=1):
+        """
+        Initialize a PLC-MOS model of a given version. There are currently three models available, v0 (intrusive)
+        and v1 (both non-intrusive and intrusive available). The default is to use the v1 models.
+        """
+
+        self.model_version = model_version
+        model_paths = [
+            # v0 model:
+            [("models/plcmos_v0.onnx", 999999999999), (None, 0)],
+
+            # v1 models:
+            [("models/plcmos_v1_intrusive.onnx", 768),
+             ("models/plcmos_v1_nonintrusive.onnx", 999999999999)],
+        ]
+        self.sessions = []
+        self.max_lens = []
+        options = ort.SessionOptions()
+        options.intra_op_num_threads = 8
+        options.graph_optimization_level = ort.GraphOptimizationLevel.ORT_ENABLE_ALL
+        for path, max_len in model_paths[model_version]:
+            if not path is None:
+                file_dir = os.path.dirname(os.path.realpath(__file__))
+                self.sessions.append(ort.InferenceSession(
+                    os.path.join(file_dir, path), options))
+                self.max_lens.append(max_len)
+            else:
+                self.sessions.append(None)
+                self.max_lens.append(0)
+
+    def logpow_dns(self, sig, floor=-30.):
+        """
+        Compute log power of complex spectrum.
+
+        Floor any -`np.inf` value to (nonzero minimum + `floor`) dB.
+        If all values are 0s, floor all values to -80 dB.
+        """
+        log10e = np.log10(np.e)
+        pspec = sig.real ** 2 + sig.imag ** 2
+        zeros = pspec == 0
+        logp = np.empty_like(pspec)
+        if np.any(~zeros):
+            logp[~zeros] = np.log(pspec[~zeros])
+            logp[zeros] = np.log(pspec[~zeros].min()) + floor / 10 / log10e
+        else:
+            logp.fill(-80 / 10 / log10e)
+
+        return logp
+
+    def hop2hsize(self, wind, hop):
+        """
+        Convert hop fraction to integer size if necessary.
+        """
+        if hop >= 1:
+            assert type(hop) == int, "Hop size must be integer!"
+            return hop
+        else:
+            assert 0 < hop < 1, "Hop fraction has to be in range (0,1)!"
+            return int(len(wind) * hop)
+
+    def stana(self, sig, sr, wind, hop, synth=False, center=False):
+        """
+        Short term analysis by windowing
+        """
+        ssize = len(sig)
+        fsize = len(wind)
+        hsize = self.hop2hsize(wind, hop)
+        if synth:
+            sstart = hsize - fsize  # int(-fsize * (1-hfrac))
+        elif center:
+            sstart = -int(len(wind) / 2)  # odd window centered at exactly n=0
+        else:
+            sstart = 0
+        send = ssize
+
+        nframe = math.ceil((send - sstart) / hsize)
+
+        # Calculate zero-padding sizes
+        zpleft = -sstart
+        zpright = (nframe - 1) * hsize + fsize - zpleft - ssize
+        if zpleft > 0 or zpright > 0:
+            sigpad = np.zeros(ssize + zpleft + zpright, dtype=sig.dtype)
+            sigpad[zpleft:len(sigpad) - zpright] = sig
+        else:
+            sigpad = sig
+
+        return as_strided(sigpad, shape=(nframe, fsize),
+                          strides=(sig.itemsize * hsize, sig.itemsize)) * wind
+
+    def stft(self, sig, sr, wind, hop, nfft):
+        """
+        Compute STFT: window + rfft
+        """
+        frames = self.stana(sig, sr, wind, hop, synth=True)
+        return rfft(frames, n=nfft)
+
+    def stft_transform(self, audio, dft_size=512, hop_fraction=0.5, sr=16000):
+        """
+        Compute STFT parameters, then compute STFT
+        """
+        window = np.hamming(dft_size + 1)
+        window = window[:-1]
+        amp = np.abs(self.stft(audio, sr, window, hop_fraction, dft_size))
+        feat = self.logpow_dns(amp, floor=-120.)
+        return feat / 20.
+
+    def run(self, audio_degraded, audio_clean=None, combined=False):
+        """
+        Run the PLCMOS model and return the MOS for the given audio. If a clean audio file is passed and the
+        selected model version has an intrusive version, that version will be used, otherwise, the nonintrusive
+        model will be used. If combined is set to true (default), the mean of intrusive and nonintrusive models
+        results will be returned, when both are available
+
+        For intrusive models, the clean reference should be the unprocessed audio file the degraded audio is
+        based on. It is not required to be aligned with the degraded audio.
+
+        Audio data should be 16kHz, mono, [-1, 1] range.
+        """
+        audio_features_degraded = np.float32(self.stft_transform(audio_degraded))[
+            np.newaxis, np.newaxis, ...]
+        assert len(
+            audio_features_degraded) <= self.max_lens[0], "Maximum input length exceeded"
+
+        if audio_clean is None:
+            combined = False
+
+        mos = 0
+
+        session = self.sessions[0]
+        assert not session is None, "Intrusive model not available for this model version."
+        audio_features_clean = np.float32(self.stft_transform(audio_clean))[
+            np.newaxis, np.newaxis, ...]
+        assert len(
+            audio_features_clean) <= self.max_lens[0], "Maximum input length exceeded"
+        onnx_inputs = {"degraded_audio": audio_features_degraded,
+                       "clean_audio": audio_features_clean}
+        mos = float(session.run(None, onnx_inputs)[0])
+
+        session = self.sessions[1]
+        assert not session is None, "Nonintrusive model not available for this model version."
+        onnx_inputs = {"degraded_audio": audio_features_degraded}
+        mos_2 = float(session.run(None, onnx_inputs)[0])
+        mos = [mos, mos_2]
+        return mos
+
+
+def run_with_defaults(degraded, clean, allow_set_size_difference=False, progress=False, model_ver=1):
+    import soundfile as sf
+    import glob
+    import tqdm
+    import pandas as pd
+
+    if os.path.isfile(degraded):
+        degraded = [degraded]
+    else:
+        degraded = list(glob.glob(os.path.join(degraded, "*.wav")))
+
+    if os.path.isfile(clean):
+        clean = [clean] * len(degraded)
+    else:
+        clean = list(glob.glob(os.path.join(clean, "*.wav")))
+
+    degraded = list(sorted(degraded))
+    clean = list(sorted(clean))
+
+    if not allow_set_size_difference:
+        assert len(degraded) == len(clean)
+
+    clean_dict = {os.path.basename(x): x for x in clean}
+    clean = []
+    for degraded_name in degraded:
+        clean.append(clean_dict[os.path.basename(degraded_name)])
+    assert len(degraded) == len(clean)
+
+    iter = zip(degraded, clean)
+    if progress:
+        iter = tqdm.tqdm(iter, total=len(degraded))
+    results = []
+
+    estimator = PLCMOSEstimator(model_version=model_ver)
+    intr = []
+    nonintr = []
+    lsds = []
+    sisdrs = []
+    for degraded_name, clean_name in iter:
+        audio_degraded, sr_degraded = sf.read(degraded_name)
+        audio_clean, sr_clean = sf.read(clean_name)
+        lsd = LSD(audio_clean, audio_degraded)
+        audio_degraded = librosa.resample(audio_degraded, 48000, 16000, res_type='kaiser_fast')
+        audio_clean = librosa.resample(audio_clean, 48000, 16000, res_type='kaiser_fast')
+
+        score = estimator.run(audio_degraded, audio_clean)
+        results.append(
+            {
+                "filename_degraded": degraded_name,
+                "filename_clean": clean_name,
+                "intrusive" + str(model_ver): score[0],
+                "non-intrusive" + str(model_ver): score[1],
+
+            }
+        )
+        lsds.append(lsd)
+        intr.append(score[0])
+        nonintr.append(score[1])
+        iter.set_description("Intru {}, Non-Intr {}, LSD {}, SISDR {}".format(sum(intr) / len(intr),
+                                                                              sum(nonintr) / len(nonintr),
+                                                                              sum(lsds) / len(lsds),
+                                                                              sum(sisdrs) / len(sisdrs)))
+
+    return pd.DataFrame(results)
+
+
+if __name__ == "__main__":
+    import argparse
+
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--degraded", type=str, required=True, help="Path to folder with degraded audio files")
+    parser.add_argument("--clean", type=str, required=True, help="Path to folder with clean audio files")
+    parser.add_argument("--model-ver", type=int, default=1, help="Model version to use")
+    parser.add_argument("--out-csv", type=str, default=None, help="Path to output CSV file, if CSV output is desired")
+    parser.add_argument("--allow-set-size-difference", type=bool, default=True,
+                        help="Set to true to allow the number of degraded and clean audio files to be different")
+    args = parser.parse_args()
+
+    results = run_with_defaults(args.degraded, args.clean, args.allow_set_size_difference, True, args.model_ver)
+
+    if args.out_csv is not None:
+        results.to_csv(args.out_csv)
+    else:
+        import pandas as pd
+
+        pd.set_option("display.max_rows", None)
+        # print(results)
+        print("")

README.md

@@ -0,0 +1,193 @@
+---
+title: FRN
+emoji: 📉
+colorFrom: gray
+colorTo: red
+sdk: static
+pinned: false
+---
+
+# FRN - Full-band Recurrent Network Official Implementation
+
+**Improving performance of real-time full-band blind packet-loss concealment with predictive network - submitted to ICASSP 2023**
+
+[![Generic badge](https://img.shields.io/badge/arXiv-2211.04071-brightgreen.svg?style=flat-square)](https://arxiv.org/abs/2211.04071)
+[![Generic badge](https://img.shields.io/github/stars/Crystalsound/FRN?color=yellow&label=FRN&logo=github&style=flat-square)](https://github.com/Crystalsound/FRN/)
+[![Generic badge](https://img.shields.io/github/last-commit/Crystalsound/FRN?color=blue&label=last%20commit&style=flat-square)](https://github.com/Crystalsound/FRN/commits)
+
+## License and citation
+
+This repository is released under the CC-BY-NC 4.0. license as found in the LICENSE file.
+
+If you use our software, please cite as below.
+For future queries, please contact [anh.nguyen@namitech.io](mailto:anh.nguyen@namitech.io).
+
+Copyright © 2022 NAMI TECHNOLOGY JSC, Inc. All rights reserved.
+
+```
+@misc{Nguyen2022ImprovingPO,
+  title={Improving performance of real-time full-band blind packet-loss concealment with predictive network},
+  author={Viet-Anh Nguyen and Anh H. T. Nguyen and Andy W. H. Khong},
+  year={2022},
+  eprint={2211.04071},
+  archivePrefix={arXiv},
+  primaryClass={cs.LG}
+}
+```
+
+# 1. Results
+
+Our model achieved a significant gain over baselines. Here, we include the predicted packet loss concealment
+mean-opinion-score (PLCMOS) using Microsoft's [PLCMOS](https://github.com/microsoft/PLC-Challenge/tree/main/PLCMOS)
+service. Please refer to our paper for more benchmarks.
+
+| Model   | PLCMOS    | 
+|---------|-----------|
+| Input   | 3.517     | 
+| tPLC    | 3.463     | 
+| TFGAN   | 3.645     | 
+| **FRN** | **3.655** |
+
+We also provide several audio samples in [https://crystalsound.github.io/FRN/](https://crystalsound.github.io/FRN/) for
+comparison.
+
+# 2. Installation
+
+## Setup
+
+### Clone the repo
+
+```
+$ git clone https://github.com/Crystalsound/FRN.git
+$ cd FRN
+```
+
+### Install dependencies
+
+* Our implementation requires the `libsndfile` libraries for the Python packages `soundfile`. On Ubuntu, they can be
+  easily installed using `apt-get`:
+    ```
+    $ apt-get update && apt-get install libsndfile-dev
+    ```
+* Create a Python 3.8 environment. Conda is recommended:
+   ```
+   $ conda create -n frn python=3.8
+   $ conda activate frn
+   ```
+
+* Install the requirements:
+    ```
+    $ pip install -r requirements.txt 
+    ```
+
+# 3. Data preparation
+
+In our paper, we conduct experiments on the [VCTK](https://datashare.ed.ac.uk/handle/10283/3443) dataset.
+
+* Download and extract the datasets:
+    ```
+    $ wget http://www.udialogue.org/download/VCTK-Corpus.tar.gz -O data/vctk/VCTK-Corpus.tar.gz
+    $ tar -zxvf data/vctk/VCTK-Corpus.tar.gz -C data/vctk/ --strip-components=1
+    ```
+
+  After extracting the datasets, your `./data` directory should look like this:
+
+    ```
+    .
+    |--data
+        |--vctk
+            |--wav48
+                |--p225
+                    |--p225_001.wav
+                    ...
+            |--train.txt   
+            |--test.txt
+    ```
+* In order to load the datasets, text files that contain training and testing audio paths are required. We have
+  prepared `train.txt` and `test.txt` files in `./data/vctk` directory.
+
+# 4. Run the code
+
+## Configuration
+
+`config.py` is the most important file. Here, you can find all the configurations related to experiment setups,
+datasets, models, training, testing, etc. Although the config file has been explained thoroughly, we recommend reading
+our paper to fully understand each parameter.
+
+## Training
+
+* Adjust training hyperparameters in `config.py`. We provide the pretrained predictor in `lightning_logs/predictor` as stated in our paper. The FRN model can be trained entirely from scratch and will work as well. In this case, initiate `PLCModel(..., pred_ckpt_path=None)`.
+
+* Run `main.py`:
+    ```
+    $ python main.py --mode train
+    ```
+* Each run will create a version in `./lightning_logs`, where the model checkpoint and hyperparameters are saved. In
+  case you want to continue training from one of these versions, just set the argument `--version` of the above command
+  to your desired version number. For example:
+    ```
+    # resume from version 0
+    $ python main.py --mode train --version 0
+    ```
+* To monitor the training curves as well as inspect model output visualization, run the tensorboard:
+    ```
+    $ tensorboard --logdir=./lightning_logs --bind_all
+    ```
+  ![image.png](https://images.viblo.asia/eb2246f9-2747-43b9-8f78-d6c154144716.png)
+
+## Evaluation
+
+In our paper, we evaluated with 2 masking methods: simulation using Markov Chain and employing real traces in PLC
+Challenge.
+
+* Get the blind test set with loss traces:
+    ```
+    $ wget http://plcchallenge2022pub.blob.core.windows.net/plcchallengearchive/blind.tar.gz
+    $ tar -xvf blind.tar.gz -C test_samples
+    ```
+* Modify `config.py` to change evaluation setup if necessary.
+* Run `main.py` with a version number to be evaluated:
+    ```
+    $ python main.py --mode eval --version 0
+    ```
+  During the evaluation, several output samples are saved to `CONFIG.LOG.sample_path` for sanity testing.
+
+## Configure a new dataset
+
+Our implementation currently works with the VCTK dataset but can be easily extensible to a new one.
+
+* Firstly, you need to prepare `train.txt` and `test.txt`. See `./data/vctk/train.txt` and `./data/vctk/test.txt` for
+  example.
+* Secondly, add a new dictionary to `CONFIG.DATA.data_dir`:
+    ```
+    {
+    'root': 'path/to/data/directory',
+    'train': 'path/to/train.txt',
+    'test': 'path/to/test.txt'
+    }
+    ```
+  **Important:** Make sure each line in `train.txt` and `test.txt` joining with `'root'` is a valid path to its
+  corresponding audio file.
+
+# 5. Audio generation
+
+* In order to generate output audios, you need to modify `CONFIG.TEST.in_dir` to your input directory.
+* Run `main.py`:
+    ```
+    python main.py --mode test --version 0
+    ```
+  The generated audios are saved to `CONFIG.TEST.out_dir`.
+
+  ## ONNX inferencing
+  We provide ONNX inferencing scripts and the best ONNX model (converted from the best checkpoint)
+  at `lightning_logs/best_model.onnx`.
+    * Convert a checkpoint to an ONNX model:
+        ```
+        python main.py --mode onnx --version 0
+        ```
+      The converted ONNX model will be saved to `lightning_logs/version_0/checkpoints`.
+    * Put test audios in `test_samples` and inference with the converted ONNX model (see `inference_onnx.py` for more
+      details):
+         ```
+        python inference_onnx.py --onnx_path lightning_logs/version_0/frn.onnx
+        ```

config.py

@@ -0,0 +1,59 @@
+class CONFIG:
+    gpus = "0,1"  # List of gpu devices
+
+    class TRAIN:
+        batch_size = 90  # number of audio files per batch
+        lr = 1e-4  # learning rate
+        epochs = 150  # max training epochs
+        workers = 12  # number of dataloader workers
+        val_split = 0.1  # validation set proportion
+        clipping_val = 1.0  # gradient clipping value
+        patience = 3  # learning rate scheduler's patience
+        factor = 0.5  # learning rate reduction factor
+
+    # Model config
+    class MODEL:
+        enc_layers = 4  # number of MLP blocks in the encoder
+        enc_in_dim = 384  # dimension of the input projection layer in the encoder
+        enc_dim = 768  # dimension of the MLP blocks
+        pred_dim = 512  # dimension of the LSTM in the predictor
+        pred_layers = 1  # number of LSTM layers in the predictor
+
+    # Dataset config
+    class DATA:
+        dataset = 'vctk'  # dataset to use
+        '''
+        Dictionary that specifies paths to root directories and train/test text files of each datasets.
+        'root' is the path to the dataset and each line of the train.txt/test.txt files should contains the path to an
+        audio file from 'root'. 
+        '''
+        data_dir = {'vctk': {'root': 'data/vctk/wav48',
+                             'train': "data/vctk/train.txt",
+                             'test': "data/vctk/test.txt"},
+                    }
+
+        assert dataset in data_dir.keys(), 'Unknown dataset.'
+        sr = 48000  # audio sampling rate
+        audio_chunk_len = 122880  # size of chunk taken in each audio files
+        window_size = 960  # window size of the STFT operation, equivalent to packet size
+        stride = 480  # stride of the STFT operation
+
+        class TRAIN:
+            packet_sizes = [256, 512, 768, 960, 1024,
+                            1536]  # packet sizes for training. All sizes should be divisible by 'audio_chunk_len'
+            transition_probs = ((0.9, 0.1), (0.5, 0.1), (0.5, 0.5))  # list of trainsition probs for Markow Chain
+
+        class EVAL:
+            packet_size = 960  # 20ms
+            transition_probs = ((0.9, 0.1))  # (0.9, 0.1) ~ 10%; (0.8, 0.2) ~ 20%; (0.6, 0.4) ~ 40%
+            masking = 'gen'  # whether using simulation or real traces from Microsoft to generate masks
+            assert masking in ['gen', 'real']
+            trace_path = 'test_samples/blind/lossy_singals'  # must be clarified if masking = 'real'
+
+    class LOG:
+        log_dir = 'lightning_logs'  # checkpoint and log directory
+        sample_path = 'audio_samples'  # path to save generated audio samples in evaluation.
+
+    class TEST:
+        in_dir = 'test_samples/blind/lossy_signals'  # path to test audio inputs
+        out_dir = 'test_samples/blind/lossy_signals_out'  # path to generated outputs

css/styles.css

@@ -0,0 +1,50 @@
+body {
+    margin: 60px auto;
+    width: 70%;
+}
+
+nav ul, footer ul {
+    font-family: 'Helvetica', 'Arial', 'Sans-Serif';
+    padding: 0px;
+    list-style: none;
+    font-weight: bold;
+}
+
+nav ul li, footer ul li {
+    display: inline;
+    margin-right: 20px;
+}
+
+a {
+    text-decoration: none;
+    color: #999;
+}
+
+a:hover {
+    text-decoration: underline;
+}
+
+h1 {
+    font-size: 2em;
+    font-family: 'Helvetica', 'Arial', 'Sans-Serif';
+}
+
+p {
+    font-size: 1.2em;
+    line-height: 1.4em;
+    color: #333;
+}
+
+footer {
+    border-top: 1px solid #d5d5d5;
+    font-size: .8em;
+}
+
+ul.posts {
+    margin: 20px auto 40px;
+    font-size: 1.5em;
+}
+
+ul.posts li {
+    list-style: none;
+}

data/vctk/test.txt

@@ -0,0 +1,3552 @@
+p376/p376_150.wav
+p376/p376_247.wav
+p376/p376_181.wav
+p376/p376_008.wav
+p376/p376_074.wav
+p376/p376_073.wav
+p376/p376_157.wav
+p376/p376_291.wav
+p376/p376_240.wav
+p376/p376_186.wav
+p376/p376_125.wav
+p376/p376_232.wav
+p376/p376_159.wav
+p376/p376_188.wav
+p376/p376_001.wav
+p376/p376_249.wav
+p376/p376_006.wav
+p376/p376_122.wav
+p376/p376_235.wav
+p376/p376_033.wav
+p376/p376_200.wav
+p376/p376_117.wav
+p376/p376_207.wav
+p376/p376_110.wav
+p376/p376_048.wav
+p376/p376_099.wav
+p376/p376_034.wav
+p376/p376_209.wav
+p376/p376_046.wav
+p376/p376_097.wav
+p376/p376_275.wav
+p376/p376_162.wav
+p376/p376_272.wav
+p376/p376_165.wav
+p376/p376_119.wav
+p376/p376_041.wav
+p376/p376_090.wav
+p376/p376_174.wav
+p376/p376_263.wav
+p376/p376_081.wav
+p376/p376_050.wav
+p376/p376_108.wav
+p376/p376_086.wav
+p376/p376_057.wav
+p376/p376_218.wav
+p376/p376_173.wav
+p376/p376_264.wav
+p376/p376_088.wav
+p376/p376_059.wav
+p376/p376_101.wav
+p376/p376_216.wav
+p376/p376_025.wav
+p376/p376_022.wav
+p376/p376_106.wav
+p376/p376_211.wav
+p376/p376_017.wav
+p376/p376_258.wav
+p376/p376_289.wav
+p376/p376_224.wav
+p376/p376_133.wav
+p376/p376_223.wav
+p376/p376_134.wav
+p376/p376_010.wav
+p376/p376_199.wav
+p376/p376_148.wav
+p376/p376_251.wav
+p376/p376_197.wav
+p376/p376_146.wav
+p376/p376_280.wav
+p376/p376_019.wav
+p376/p376_256.wav
+p376/p376_190.wav
+p376/p376_141.wav
+p376/p376_287.wav
+p376/p376_065.wav
+p376/p376_273.wav
+p376/p376_164.wav
+p376/p376_118.wav
+p376/p376_040.wav
+p376/p376_091.wav
+p376/p376_208.wav
+p376/p376_096.wav
+p376/p376_274.wav
+p376/p376_163.wav
+p376/p376_206.wav
+p376/p376_111.wav
+p376/p376_049.wav
+p376/p376_098.wav
+p376/p376_035.wav
+p376/p376_032.wav
+p376/p376_201.wav
+p376/p376_116.wav
+p376/p376_248.wav
+p376/p376_007.wav
+p376/p376_123.wav
+p376/p376_234.wav
+p376/p376_124.wav
+p376/p376_233.wav
+p376/p376_158.wav
+p376/p376_189.wav
+p376/p376_072.wav
+p376/p376_156.wav
+p376/p376_290.wav
+p376/p376_241.wav
+p376/p376_187.wav
+p376/p376_151.wav
+p376/p376_246.wav
+p376/p376_180.wav
+p376/p376_009.wav
+p376/p376_075.wav
+p376/p376_018.wav
+p376/p376_257.wav
+p376/p376_191.wav
+p376/p376_140.wav
+p376/p376_286.wav
+p376/p376_064.wav
+p376/p376_063.wav
+p376/p376_250.wav
+p376/p376_196.wav
+p376/p376_147.wav
+p376/p376_222.wav
+p376/p376_135.wav
+p376/p376_011.wav
+p376/p376_198.wav
+p376/p376_149.wav
+p376/p376_016.wav
+p376/p376_259.wav
+p376/p376_288.wav
+p376/p376_225.wav
+p376/p376_132.wav
+p376/p376_023.wav
+p376/p376_107.wav
+p376/p376_210.wav
+p376/p376_089.wav
+p376/p376_058.wav
+p376/p376_100.wav
+p376/p376_217.wav
+p376/p376_024.wav
+p376/p376_087.wav
+p376/p376_056.wav
+p376/p376_219.wav
+p376/p376_172.wav
+p376/p376_265.wav
+p376/p376_175.wav
+p376/p376_262.wav
+p376/p376_080.wav
+p376/p376_051.wav
+p376/p376_109.wav
+p376/p376_060.wav
+p376/p376_138.wav
+p376/p376_195.wav
+p376/p376_253.wav
+p376/p376_282.wav
+p376/p376_144.wav
+p376/p376_192.wav
+p376/p376_254.wav
+p376/p376_285.wav
+p376/p376_143.wav
+p376/p376_067.wav
+p376/p376_228.wav
+p376/p376_015.wav
+p376/p376_069.wav
+p376/p376_226.wav
+p376/p376_131.wav
+p376/p376_221.wav
+p376/p376_136.wav
+p376/p376_012.wav
+p376/p376_103.wav
+p376/p376_214.wav
+p376/p376_027.wav
+p376/p376_268.wav
+p376/p376_020.wav
+p376/p376_178.wav
+p376/p376_104.wav
+p376/p376_213.wav
+p376/p376_176.wav
+p376/p376_261.wav
+p376/p376_083.wav
+p376/p376_052.wav
+p376/p376_084.wav
+p376/p376_055.wav
+p376/p376_029.wav
+p376/p376_171.wav
+p376/p376_266.wav
+p376/p376_044.wav
+p376/p376_095.wav
+p376/p376_277.wav
+p376/p376_160.wav
+p376/p376_038.wav
+p376/p376_270.wav
+p376/p376_167.wav
+p376/p376_043.wav
+p376/p376_092.wav
+p376/p376_169.wav
+p376/p376_031.wav
+p376/p376_202.wav
+p376/p376_115.wav
+p376/p376_205.wav
+p376/p376_112.wav
+p376/p376_279.wav
+p376/p376_036.wav
+p376/p376_127.wav
+p376/p376_230.wav
+p376/p376_003.wav
+p376/p376_004.wav
+p376/p376_120.wav
+p376/p376_237.wav
+p376/p376_078.wav
+p376/p376_294.wav
+p376/p376_152.wav
+p376/p376_183.wav
+p376/p376_245.wav
+p376/p376_239.wav
+p376/p376_076.wav
+p376/p376_129.wav
+p376/p376_071.wav
+p376/p376_293.wav
+p376/p376_155.wav
+p376/p376_184.wav
+p376/p376_242.wav
+p376/p376_085.wav
+p376/p376_054.wav
+p376/p376_028.wav
+p376/p376_170.wav
+p376/p376_267.wav
+p376/p376_177.wav
+p376/p376_260.wav
+p376/p376_082.wav
+p376/p376_053.wav
+p376/p376_021.wav
+p376/p376_179.wav
+p376/p376_105.wav
+p376/p376_212.wav
+p376/p376_102.wav
+p376/p376_215.wav
+p376/p376_026.wav
+p376/p376_269.wav
+p376/p376_220.wav
+p376/p376_137.wav
+p376/p376_013.wav
+p376/p376_014.wav
+p376/p376_068.wav
+p376/p376_227.wav
+p376/p376_130.wav
+p376/p376_193.wav
+p376/p376_255.wav
+p376/p376_284.wav
+p376/p376_142.wav
+p376/p376_066.wav
+p376/p376_229.wav
+p376/p376_061.wav
+p376/p376_139.wav
+p376/p376_194.wav
+p376/p376_252.wav
+p376/p376_283.wav
+p376/p376_145.wav
+p376/p376_128.wav
+p376/p376_070.wav
+p376/p376_292.wav
+p376/p376_154.wav
+p376/p376_185.wav
+p376/p376_243.wav
+p376/p376_295.wav
+p376/p376_153.wav
+p376/p376_182.wav
+p376/p376_244.wav
+p376/p376_238.wav
+p376/p376_077.wav
+p376/p376_005.wav
+p376/p376_121.wav
+p376/p376_236.wav
+p376/p376_079.wav
+p376/p376_126.wav
+p376/p376_231.wav
+p376/p376_002.wav
+p376/p376_204.wav
+p376/p376_113.wav
+p376/p376_278.wav
+p376/p376_037.wav
+p376/p376_168.wav
+p376/p376_030.wav
+p376/p376_203.wav
+p376/p376_114.wav
+p376/p376_271.wav
+p376/p376_166.wav
+p376/p376_042.wav
+p376/p376_093.wav
+p376/p376_045.wav
+p376/p376_094.wav
+p376/p376_276.wav
+p376/p376_161.wav
+p376/p376_039.wav
+p362/p362_196.wav
+p362/p362_250.wav
+p362/p362_281.wav
+p362/p362_147.wav
+p362/p362_308.wav
+p362/p362_402.wav
+p362/p362_063.wav
+p362/p362_374.wav
+p362/p362_405.wav
+p362/p362_064.wav
+p362/p362_373.wav
+p362/p362_191.wav
+p362/p362_257.wav
+p362/p362_286.wav
+p362/p362_140.wav
+p362/p362_018.wav
+p362/p362_225.wav
+p362/p362_132.wav
+p362/p362_259.wav
+p362/p362_288.wav
+p362/p362_016.wav
+p362/p362_301.wav
+p362/p362_198.wav
+p362/p362_149.wav
+p362/p362_011.wav
+p362/p362_306.wav
+p362/p362_222.wav
+p362/p362_135.wav
+p362/p362_333.wav
+p362/p362_024.wav
+p362/p362_100.wav
+p362/p362_217.wav
+p362/p362_089.wav
+p362/p362_058.wav
+p362/p362_107.wav
+p362/p362_210.wav
+p362/p362_348.wav
+p362/p362_399.wav
+p362/p362_334.wav
+p362/p362_023.wav
+p362/p362_109.wav
+p362/p362_346.wav
+p362/p362_080.wav
+p362/p362_051.wav
+p362/p362_397.wav
+p362/p362_175.wav
+p362/p362_262.wav
+p362/p362_172.wav
+p362/p362_265.wav
+p362/p362_219.wav
+p362/p362_341.wav
+p362/p362_087.wav
+p362/p362_056.wav
+p362/p362_390.wav
+p362/p362_274.wav
+p362/p362_163.wav
+p362/p362_047.wav
+p362/p362_381.wav
+p362/p362_350.wav
+p362/p362_096.wav
+p362/p362_208.wav
+p362/p362_040.wav
+p362/p362_421.wav
+p362/p362_386.wav
+p362/p362_357.wav
+p362/p362_091.wav
+p362/p362_118.wav
+p362/p362_273.wav
+p362/p362_164.wav
+p362/p362_388.wav
+p362/p362_359.wav
+p362/p362_201.wav
+p362/p362_116.wav
+p362/p362_032.wav
+p362/p362_325.wav
+p362/p362_035.wav
+p362/p362_322.wav
+p362/p362_049.wav
+p362/p362_098.wav
+p362/p362_206.wav
+p362/p362_111.wav
+p362/p362_317.wav
+p362/p362_158.wav
+p362/p362_189.wav
+p362/p362_124.wav
+p362/p362_233.wav
+p362/p362_123.wav
+p362/p362_234.wav
+p362/p362_310.wav
+p362/p362_007.wav
+p362/p362_299.wav
+p362/p362_248.wav
+p362/p362_362.wav
+p362/p362_414.wav
+p362/p362_075.wav
+p362/p362_009.wav
+p362/p362_297.wav
+p362/p362_151.wav
+p362/p362_180.wav
+p362/p362_246.wav
+p362/p362_319.wav
+p362/p362_290.wav
+p362/p362_156.wav
+p362/p362_187.wav
+p362/p362_241.wav
+p362/p362_365.wav
+p362/p362_413.wav
+p362/p362_072.wav
+p362/p362_173.wav
+p362/p362_264.wav
+p362/p362_218.wav
+p362/p362_340.wav
+p362/p362_086.wav
+p362/p362_057.wav
+p362/p362_391.wav
+p362/p362_108.wav
+p362/p362_347.wav
+p362/p362_081.wav
+p362/p362_050.wav
+p362/p362_396.wav
+p362/p362_174.wav
+p362/p362_263.wav
+p362/p362_106.wav
+p362/p362_211.wav
+p362/p362_349.wav
+p362/p362_398.wav
+p362/p362_335.wav
+p362/p362_022.wav
+p362/p362_332.wav
+p362/p362_025.wav
+p362/p362_101.wav
+p362/p362_216.wav
+p362/p362_088.wav
+p362/p362_059.wav
+p362/p362_199.wav
+p362/p362_148.wav
+p362/p362_010.wav
+p362/p362_307.wav
+p362/p362_223.wav
+p362/p362_134.wav
+p362/p362_224.wav
+p362/p362_133.wav
+p362/p362_258.wav
+p362/p362_289.wav
+p362/p362_017.wav
+p362/p362_300.wav
+p362/p362_404.wav
+p362/p362_065.wav
+p362/p362_372.wav
+p362/p362_190.wav
+p362/p362_256.wav
+p362/p362_287.wav
+p362/p362_141.wav
+p362/p362_019.wav
+p362/p362_197.wav
+p362/p362_251.wav
+p362/p362_280.wav
+p362/p362_146.wav
+p362/p362_309.wav
+p362/p362_403.wav
+p362/p362_062.wav
+p362/p362_375.wav
+p362/p362_318.wav
+p362/p362_291.wav
+p362/p362_157.wav
+p362/p362_186.wav
+p362/p362_240.wav
+p362/p362_364.wav
+p362/p362_412.wav
+p362/p362_073.wav
+p362/p362_363.wav
+p362/p362_415.wav
+p362/p362_074.wav
+p362/p362_008.wav
+p362/p362_296.wav
+p362/p362_150.wav
+p362/p362_181.wav
+p362/p362_247.wav
+p362/p362_122.wav
+p362/p362_235.wav
+p362/p362_311.wav
+p362/p362_006.wav
+p362/p362_298.wav
+p362/p362_249.wav
+p362/p362_316.wav
+p362/p362_001.wav
+p362/p362_159.wav
+p362/p362_188.wav
+p362/p362_125.wav
+p362/p362_232.wav
+p362/p362_034.wav
+p362/p362_323.wav
+p362/p362_048.wav
+p362/p362_099.wav
+p362/p362_207.wav
+p362/p362_110.wav
+p362/p362_389.wav
+p362/p362_358.wav
+p362/p362_200.wav
+p362/p362_117.wav
+p362/p362_033.wav
+p362/p362_324.wav
+p362/p362_041.wav
+p362/p362_387.wav
+p362/p362_420.wav
+p362/p362_356.wav
+p362/p362_090.wav
+p362/p362_119.wav
+p362/p362_272.wav
+p362/p362_165.wav
+p362/p362_275.wav
+p362/p362_162.wav
+p362/p362_046.wav
+p362/p362_380.wav
+p362/p362_351.wav
+p362/p362_097.wav
+p362/p362_209.wav
+p362/p362_360.wav
+p362/p362_077.wav
+p362/p362_416.wav
+p362/p362_238.wav
+p362/p362_153.wav
+p362/p362_295.wav
+p362/p362_244.wav
+p362/p362_182.wav
+p362/p362_154.wav
+p362/p362_292.wav
+p362/p362_243.wav
+p362/p362_185.wav
+p362/p362_367.wav
+p362/p362_070.wav
+p362/p362_411.wav
+p362/p362_128.wav
+p362/p362_315.wav
+p362/p362_002.wav
+p362/p362_369.wav
+p362/p362_126.wav
+p362/p362_231.wav
+p362/p362_079.wav
+p362/p362_418.wav
+p362/p362_121.wav
+p362/p362_236.wav
+p362/p362_312.wav
+p362/p362_005.wav
+p362/p362_203.wav
+p362/p362_114.wav
+p362/p362_030.wav
+p362/p362_327.wav
+p362/p362_168.wav
+p362/p362_037.wav
+p362/p362_320.wav
+p362/p362_278.wav
+p362/p362_204.wav
+p362/p362_113.wav
+p362/p362_039.wav
+p362/p362_276.wav
+p362/p362_161.wav
+p362/p362_424.wav
+p362/p362_383.wav
+p362/p362_045.wav
+p362/p362_094.wav
+p362/p362_352.wav
+p362/p362_384.wav
+p362/p362_423.wav
+p362/p362_042.wav
+p362/p362_093.wav
+p362/p362_355.wav
+p362/p362_329.wav
+p362/p362_271.wav
+p362/p362_166.wav
+p362/p362_082.wav
+p362/p362_344.wav
+p362/p362_395.wav
+p362/p362_053.wav
+p362/p362_177.wav
+p362/p362_260.wav
+p362/p362_338.wav
+p362/p362_170.wav
+p362/p362_267.wav
+p362/p362_028.wav
+p362/p362_085.wav
+p362/p362_343.wav
+p362/p362_392.wav
+p362/p362_054.wav
+p362/p362_269.wav
+p362/p362_331.wav
+p362/p362_026.wav
+p362/p362_102.wav
+p362/p362_215.wav
+p362/p362_105.wav
+p362/p362_212.wav
+p362/p362_179.wav
+p362/p362_336.wav
+p362/p362_021.wav
+p362/p362_227.wav
+p362/p362_130.wav
+p362/p362_068.wav
+p362/p362_409.wav
+p362/p362_014.wav
+p362/p362_303.wav
+p362/p362_013.wav
+p362/p362_304.wav
+p362/p362_220.wav
+p362/p362_137.wav
+p362/p362_378.wav
+p362/p362_252.wav
+p362/p362_194.wav
+p362/p362_145.wav
+p362/p362_283.wav
+p362/p362_139.wav
+p362/p362_061.wav
+p362/p362_400.wav
+p362/p362_376.wav
+p362/p362_229.wav
+p362/p362_066.wav
+p362/p362_407.wav
+p362/p362_371.wav
+p362/p362_255.wav
+p362/p362_193.wav
+p362/p362_142.wav
+p362/p362_284.wav
+p362/p362_422.wav
+p362/p362_385.wav
+p362/p362_043.wav
+p362/p362_092.wav
+p362/p362_354.wav
+p362/p362_328.wav
+p362/p362_270.wav
+p362/p362_167.wav
+p362/p362_038.wav
+p362/p362_277.wav
+p362/p362_160.wav
+p362/p362_382.wav
+p362/p362_044.wav
+p362/p362_095.wav
+p362/p362_353.wav
+p362/p362_036.wav
+p362/p362_321.wav
+p362/p362_279.wav
+p362/p362_205.wav
+p362/p362_112.wav
+p362/p362_115.wav
+p362/p362_031.wav
+p362/p362_326.wav
+p362/p362_169.wav
+p362/p362_078.wav
+p362/p362_419.wav
+p362/p362_120.wav
+p362/p362_237.wav
+p362/p362_313.wav
+p362/p362_004.wav
+p362/p362_314.wav
+p362/p362_003.wav
+p362/p362_368.wav
+p362/p362_127.wav
+p362/p362_230.wav
+p362/p362_155.wav
+p362/p362_293.wav
+p362/p362_242.wav
+p362/p362_184.wav
+p362/p362_366.wav
+p362/p362_071.wav
+p362/p362_410.wav
+p362/p362_129.wav
+p362/p362_361.wav
+p362/p362_076.wav
+p362/p362_417.wav
+p362/p362_239.wav
+p362/p362_152.wav
+p362/p362_294.wav
+p362/p362_245.wav
+p362/p362_183.wav
+p362/p362_228.wav
+p362/p362_067.wav
+p362/p362_406.wav
+p362/p362_370.wav
+p362/p362_254.wav
+p362/p362_192.wav
+p362/p362_143.wav
+p362/p362_285.wav
+p362/p362_253.wav
+p362/p362_195.wav
+p362/p362_144.wav
+p362/p362_282.wav
+p362/p362_138.wav
+p362/p362_060.wav
+p362/p362_401.wav
+p362/p362_377.wav
+p362/p362_012.wav
+p362/p362_305.wav
+p362/p362_221.wav
+p362/p362_136.wav
+p362/p362_379.wav
+p362/p362_226.wav
+p362/p362_131.wav
+p362/p362_069.wav
+p362/p362_408.wav
+p362/p362_015.wav
+p362/p362_302.wav
+p362/p362_104.wav
+p362/p362_213.wav
+p362/p362_178.wav
+p362/p362_337.wav
+p362/p362_020.wav
+p362/p362_268.wav
+p362/p362_330.wav
+p362/p362_027.wav
+p362/p362_103.wav
+p362/p362_214.wav
+p362/p362_171.wav
+p362/p362_266.wav
+p362/p362_029.wav
+p362/p362_084.wav
+p362/p362_342.wav
+p362/p362_393.wav
+p362/p362_055.wav
+p362/p362_083.wav
+p362/p362_345.wav
+p362/p362_394.wav
+p362/p362_052.wav
+p362/p362_176.wav
+p362/p362_261.wav
+p362/p362_339.wav
+p347/p347_109.wav
+p347/p347_397.wav
+p347/p347_051.wav
+p347/p347_080.wav
+p347/p347_346.wav
+p347/p347_262.wav
+p347/p347_175.wav
+p347/p347_265.wav
+p347/p347_172.wav
+p347/p347_219.wav
+p347/p347_390.wav
+p347/p347_056.wav
+p347/p347_087.wav
+p347/p347_341.wav
+p347/p347_024.wav
+p347/p347_333.wav
+p347/p347_217.wav
+p347/p347_100.wav
+p347/p347_058.wav
+p347/p347_089.wav
+p347/p347_210.wav
+p347/p347_107.wav
+p347/p347_399.wav
+p347/p347_348.wav
+p347/p347_023.wav
+p347/p347_334.wav
+p347/p347_132.wav
+p347/p347_225.wav
+p347/p347_288.wav
+p347/p347_259.wav
+p347/p347_301.wav
+p347/p347_016.wav
+p347/p347_149.wav
+p347/p347_198.wav
+p347/p347_306.wav
+p347/p347_011.wav
+p347/p347_135.wav
+p347/p347_222.wav
+p347/p347_147.wav
+p347/p347_281.wav
+p347/p347_250.wav
+p347/p347_196.wav
+p347/p347_308.wav
+p347/p347_374.wav
+p347/p347_063.wav
+p347/p347_402.wav
+p347/p347_373.wav
+p347/p347_064.wav
+p347/p347_405.wav
+p347/p347_140.wav
+p347/p347_286.wav
+p347/p347_257.wav
+p347/p347_191.wav
+p347/p347_018.wav
+p347/p347_075.wav
+p347/p347_414.wav
+p347/p347_362.wav
+p347/p347_009.wav
+p347/p347_246.wav
+p347/p347_180.wav
+p347/p347_151.wav
+p347/p347_297.wav
+p347/p347_319.wav
+p347/p347_241.wav
+p347/p347_187.wav
+p347/p347_156.wav
+p347/p347_290.wav
+p347/p347_072.wav
+p347/p347_413.wav
+p347/p347_365.wav
+p347/p347_317.wav
+p347/p347_189.wav
+p347/p347_158.wav
+p347/p347_233.wav
+p347/p347_124.wav
+p347/p347_234.wav
+p347/p347_123.wav
+p347/p347_007.wav
+p347/p347_310.wav
+p347/p347_248.wav
+p347/p347_299.wav
+p347/p347_359.wav
+p347/p347_388.wav
+p347/p347_116.wav
+p347/p347_201.wav
+p347/p347_325.wav
+p347/p347_032.wav
+p347/p347_322.wav
+p347/p347_035.wav
+p347/p347_098.wav
+p347/p347_049.wav
+p347/p347_111.wav
+p347/p347_206.wav
+p347/p347_163.wav
+p347/p347_274.wav
+p347/p347_096.wav
+p347/p347_350.wav
+p347/p347_381.wav
+p347/p347_047.wav
+p347/p347_208.wav
+p347/p347_091.wav
+p347/p347_357.wav
+p347/p347_421.wav
+p347/p347_386.wav
+p347/p347_040.wav
+p347/p347_118.wav
+p347/p347_164.wav
+p347/p347_273.wav
+p347/p347_372.wav
+p347/p347_065.wav
+p347/p347_404.wav
+p347/p347_141.wav
+p347/p347_287.wav
+p347/p347_256.wav
+p347/p347_190.wav
+p347/p347_019.wav
+p347/p347_146.wav
+p347/p347_280.wav
+p347/p347_251.wav
+p347/p347_197.wav
+p347/p347_309.wav
+p347/p347_375.wav
+p347/p347_062.wav
+p347/p347_403.wav
+p347/p347_148.wav
+p347/p347_199.wav
+p347/p347_307.wav
+p347/p347_010.wav
+p347/p347_134.wav
+p347/p347_223.wav
+p347/p347_133.wav
+p347/p347_224.wav
+p347/p347_289.wav
+p347/p347_258.wav
+p347/p347_300.wav
+p347/p347_017.wav
+p347/p347_211.wav
+p347/p347_106.wav
+p347/p347_398.wav
+p347/p347_349.wav
+p347/p347_022.wav
+p347/p347_335.wav
+p347/p347_025.wav
+p347/p347_332.wav
+p347/p347_216.wav
+p347/p347_101.wav
+p347/p347_059.wav
+p347/p347_088.wav
+p347/p347_264.wav
+p347/p347_173.wav
+p347/p347_218.wav
+p347/p347_391.wav
+p347/p347_057.wav
+p347/p347_086.wav
+p347/p347_340.wav
+p347/p347_108.wav
+p347/p347_396.wav
+p347/p347_050.wav
+p347/p347_081.wav
+p347/p347_347.wav
+p347/p347_263.wav
+p347/p347_174.wav
+p347/p347_090.wav
+p347/p347_356.wav
+p347/p347_387.wav
+p347/p347_420.wav
+p347/p347_041.wav
+p347/p347_119.wav
+p347/p347_165.wav
+p347/p347_272.wav
+p347/p347_162.wav
+p347/p347_275.wav
+p347/p347_097.wav
+p347/p347_351.wav
+p347/p347_380.wav
+p347/p347_046.wav
+p347/p347_209.wav
+p347/p347_323.wav
+p347/p347_034.wav
+p347/p347_099.wav
+p347/p347_048.wav
+p347/p347_110.wav
+p347/p347_207.wav
+p347/p347_358.wav
+p347/p347_389.wav
+p347/p347_117.wav
+p347/p347_200.wav
+p347/p347_324.wav
+p347/p347_033.wav
+p347/p347_235.wav
+p347/p347_122.wav
+p347/p347_006.wav
+p347/p347_311.wav
+p347/p347_249.wav
+p347/p347_298.wav
+p347/p347_001.wav
+p347/p347_316.wav
+p347/p347_188.wav
+p347/p347_159.wav
+p347/p347_232.wav
+p347/p347_125.wav
+p347/p347_318.wav
+p347/p347_240.wav
+p347/p347_186.wav
+p347/p347_157.wav
+p347/p347_291.wav
+p347/p347_073.wav
+p347/p347_412.wav
+p347/p347_364.wav
+p347/p347_074.wav
+p347/p347_415.wav
+p347/p347_363.wav
+p347/p347_008.wav
+p347/p347_247.wav
+p347/p347_181.wav
+p347/p347_150.wav
+p347/p347_296.wav
+p347/p347_039.wav
+p347/p347_161.wav
+p347/p347_276.wav
+p347/p347_352.wav
+p347/p347_094.wav
+p347/p347_045.wav
+p347/p347_424.wav
+p347/p347_383.wav
+p347/p347_355.wav
+p347/p347_093.wav
+p347/p347_384.wav
+p347/p347_423.wav
+p347/p347_329.wav
+p347/p347_166.wav
+p347/p347_271.wav
+p347/p347_114.wav
+p347/p347_203.wav
+p347/p347_327.wav
+p347/p347_030.wav
+p347/p347_168.wav
+p347/p347_320.wav
+p347/p347_037.wav
+p347/p347_278.wav
+p347/p347_113.wav
+p347/p347_204.wav
+p347/p347_002.wav
+p347/p347_315.wav
+p347/p347_369.wav
+p347/p347_231.wav
+p347/p347_126.wav
+p347/p347_418.wav
+p347/p347_079.wav
+p347/p347_236.wav
+p347/p347_121.wav
+p347/p347_005.wav
+p347/p347_312.wav
+p347/p347_416.wav
+p347/p347_077.wav
+p347/p347_360.wav
+p347/p347_238.wav
+p347/p347_182.wav
+p347/p347_244.wav
+p347/p347_295.wav
+p347/p347_153.wav
+p347/p347_185.wav
+p347/p347_243.wav
+p347/p347_292.wav
+p347/p347_154.wav
+p347/p347_411.wav
+p347/p347_070.wav
+p347/p347_367.wav
+p347/p347_128.wav
+p347/p347_283.wav
+p347/p347_145.wav
+p347/p347_194.wav
+p347/p347_252.wav
+p347/p347_139.wav
+p347/p347_376.wav
+p347/p347_400.wav
+p347/p347_061.wav
+p347/p347_229.wav
+p347/p347_371.wav
+p347/p347_407.wav
+p347/p347_066.wav
+p347/p347_284.wav
+p347/p347_142.wav
+p347/p347_193.wav
+p347/p347_255.wav
+p347/p347_130.wav
+p347/p347_227.wav
+p347/p347_409.wav
+p347/p347_068.wav
+p347/p347_303.wav
+p347/p347_014.wav
+p347/p347_304.wav
+p347/p347_013.wav
+p347/p347_137.wav
+p347/p347_220.wav
+p347/p347_378.wav
+p347/p347_269.wav
+p347/p347_026.wav
+p347/p347_331.wav
+p347/p347_215.wav
+p347/p347_102.wav
+p347/p347_212.wav
+p347/p347_105.wav
+p347/p347_179.wav
+p347/p347_021.wav
+p347/p347_336.wav
+p347/p347_053.wav
+p347/p347_395.wav
+p347/p347_344.wav
+p347/p347_082.wav
+p347/p347_260.wav
+p347/p347_177.wav
+p347/p347_338.wav
+p347/p347_267.wav
+p347/p347_170.wav
+p347/p347_028.wav
+p347/p347_054.wav
+p347/p347_392.wav
+p347/p347_343.wav
+p347/p347_085.wav
+p347/p347_184.wav
+p347/p347_242.wav
+p347/p347_293.wav
+p347/p347_155.wav
+p347/p347_410.wav
+p347/p347_071.wav
+p347/p347_366.wav
+p347/p347_129.wav
+p347/p347_417.wav
+p347/p347_076.wav
+p347/p347_361.wav
+p347/p347_239.wav
+p347/p347_183.wav
+p347/p347_245.wav
+p347/p347_294.wav
+p347/p347_152.wav
+p347/p347_419.wav
+p347/p347_078.wav
+p347/p347_237.wav
+p347/p347_120.wav
+p347/p347_004.wav
+p347/p347_313.wav
+p347/p347_003.wav
+p347/p347_314.wav
+p347/p347_368.wav
+p347/p347_230.wav
+p347/p347_127.wav
+p347/p347_321.wav
+p347/p347_036.wav
+p347/p347_279.wav
+p347/p347_112.wav
+p347/p347_205.wav
+p347/p347_115.wav
+p347/p347_202.wav
+p347/p347_326.wav
+p347/p347_031.wav
+p347/p347_169.wav
+p347/p347_354.wav
+p347/p347_092.wav
+p347/p347_043.wav
+p347/p347_422.wav
+p347/p347_385.wav
+p347/p347_328.wav
+p347/p347_167.wav
+p347/p347_270.wav
+p347/p347_038.wav
+p347/p347_160.wav
+p347/p347_277.wav
+p347/p347_353.wav
+p347/p347_095.wav
+p347/p347_044.wav
+p347/p347_382.wav
+p347/p347_266.wav
+p347/p347_171.wav
+p347/p347_029.wav
+p347/p347_055.wav
+p347/p347_393.wav
+p347/p347_342.wav
+p347/p347_084.wav
+p347/p347_052.wav
+p347/p347_394.wav
+p347/p347_345.wav
+p347/p347_083.wav
+p347/p347_261.wav
+p347/p347_176.wav
+p347/p347_339.wav
+p347/p347_213.wav
+p347/p347_104.wav
+p347/p347_178.wav
+p347/p347_020.wav
+p347/p347_337.wav
+p347/p347_268.wav
+p347/p347_027.wav
+p347/p347_330.wav
+p347/p347_214.wav
+p347/p347_103.wav
+p347/p347_305.wav
+p347/p347_012.wav
+p347/p347_136.wav
+p347/p347_221.wav
+p347/p347_379.wav
+p347/p347_131.wav
+p347/p347_226.wav
+p347/p347_408.wav
+p347/p347_069.wav
+p347/p347_302.wav
+p347/p347_015.wav
+p347/p347_228.wav
+p347/p347_370.wav
+p347/p347_406.wav
+p347/p347_067.wav
+p347/p347_285.wav
+p347/p347_143.wav
+p347/p347_192.wav
+p347/p347_254.wav
+p347/p347_282.wav
+p347/p347_144.wav
+p347/p347_195.wav
+p347/p347_253.wav
+p347/p347_138.wav
+p347/p347_377.wav
+p347/p347_401.wav
+p347/p347_060.wav
+p361/p361_205.wav
+p361/p361_112.wav
+p361/p361_036.wav
+p361/p361_321.wav
+p361/p361_279.wav
+p361/p361_031.wav
+p361/p361_326.wav
+p361/p361_169.wav
+p361/p361_202.wav
+p361/p361_115.wav
+p361/p361_328.wav
+p361/p361_270.wav
+p361/p361_167.wav
+p361/p361_043.wav
+p361/p361_385.wav
+p361/p361_422.wav
+p361/p361_354.wav
+p361/p361_092.wav
+p361/p361_044.wav
+p361/p361_382.wav
+p361/p361_353.wav
+p361/p361_095.wav
+p361/p361_038.wav
+p361/p361_277.wav
+p361/p361_160.wav
+p361/p361_366.wav
+p361/p361_410.wav
+p361/p361_071.wav
+p361/p361_129.wav
+p361/p361_293.wav
+p361/p361_155.wav
+p361/p361_184.wav
+p361/p361_242.wav
+p361/p361_294.wav
+p361/p361_152.wav
+p361/p361_183.wav
+p361/p361_245.wav
+p361/p361_361.wav
+p361/p361_417.wav
+p361/p361_076.wav
+p361/p361_239.wav
+p361/p361_313.wav
+p361/p361_004.wav
+p361/p361_419.wav
+p361/p361_078.wav
+p361/p361_120.wav
+p361/p361_237.wav
+p361/p361_368.wav
+p361/p361_127.wav
+p361/p361_230.wav
+p361/p361_314.wav
+p361/p361_003.wav
+p361/p361_221.wav
+p361/p361_136.wav
+p361/p361_379.wav
+p361/p361_012.wav
+p361/p361_305.wav
+p361/p361_015.wav
+p361/p361_302.wav
+p361/p361_226.wav
+p361/p361_408.wav
+p361/p361_069.wav
+p361/p361_192.wav
+p361/p361_254.wav
+p361/p361_285.wav
+p361/p361_143.wav
+p361/p361_228.wav
+p361/p361_406.wav
+p361/p361_067.wav
+p361/p361_370.wav
+p361/p361_138.wav
+p361/p361_401.wav
+p361/p361_060.wav
+p361/p361_377.wav
+p361/p361_195.wav
+p361/p361_253.wav
+p361/p361_282.wav
+p361/p361_144.wav
+p361/p361_342.wav
+p361/p361_084.wav
+p361/p361_055.wav
+p361/p361_393.wav
+p361/p361_171.wav
+p361/p361_266.wav
+p361/p361_029.wav
+p361/p361_176.wav
+p361/p361_261.wav
+p361/p361_339.wav
+p361/p361_345.wav
+p361/p361_083.wav
+p361/p361_052.wav
+p361/p361_394.wav
+p361/p361_178.wav
+p361/p361_337.wav
+p361/p361_020.wav
+p361/p361_104.wav
+p361/p361_213.wav
+p361/p361_103.wav
+p361/p361_214.wav
+p361/p361_268.wav
+p361/p361_330.wav
+p361/p361_027.wav
+p361/p361_369.wav
+p361/p361_126.wav
+p361/p361_231.wav
+p361/p361_315.wav
+p361/p361_002.wav
+p361/p361_312.wav
+p361/p361_005.wav
+p361/p361_418.wav
+p361/p361_079.wav
+p361/p361_121.wav
+p361/p361_236.wav
+p361/p361_295.wav
+p361/p361_153.wav
+p361/p361_182.wav
+p361/p361_244.wav
+p361/p361_360.wav
+p361/p361_416.wav
+p361/p361_077.wav
+p361/p361_238.wav
+p361/p361_367.wav
+p361/p361_411.wav
+p361/p361_070.wav
+p361/p361_128.wav
+p361/p361_292.wav
+p361/p361_154.wav
+p361/p361_185.wav
+p361/p361_243.wav
+p361/p361_045.wav
+p361/p361_383.wav
+p361/p361_424.wav
+p361/p361_352.wav
+p361/p361_094.wav
+p361/p361_039.wav
+p361/p361_276.wav
+p361/p361_161.wav
+p361/p361_329.wav
+p361/p361_271.wav
+p361/p361_166.wav
+p361/p361_042.wav
+p361/p361_423.wav
+p361/p361_384.wav
+p361/p361_355.wav
+p361/p361_093.wav
+p361/p361_030.wav
+p361/p361_327.wav
+p361/p361_168.wav
+p361/p361_203.wav
+p361/p361_114.wav
+p361/p361_204.wav
+p361/p361_113.wav
+p361/p361_037.wav
+p361/p361_320.wav
+p361/p361_278.wav
+p361/p361_102.wav
+p361/p361_215.wav
+p361/p361_269.wav
+p361/p361_331.wav
+p361/p361_026.wav
+p361/p361_179.wav
+p361/p361_336.wav
+p361/p361_021.wav
+p361/p361_105.wav
+p361/p361_212.wav
+p361/p361_177.wav
+p361/p361_260.wav
+p361/p361_338.wav
+p361/p361_344.wav
+p361/p361_082.wav
+p361/p361_053.wav
+p361/p361_395.wav
+p361/p361_343.wav
+p361/p361_085.wav
+p361/p361_054.wav
+p361/p361_392.wav
+p361/p361_170.wav
+p361/p361_267.wav
+p361/p361_028.wav
+p361/p361_139.wav
+p361/p361_400.wav
+p361/p361_061.wav
+p361/p361_376.wav
+p361/p361_194.wav
+p361/p361_252.wav
+p361/p361_283.wav
+p361/p361_145.wav
+p361/p361_193.wav
+p361/p361_255.wav
+p361/p361_284.wav
+p361/p361_142.wav
+p361/p361_229.wav
+p361/p361_407.wav
+p361/p361_066.wav
+p361/p361_371.wav
+p361/p361_014.wav
+p361/p361_303.wav
+p361/p361_227.wav
+p361/p361_130.wav
+p361/p361_409.wav
+p361/p361_068.wav
+p361/p361_220.wav
+p361/p361_137.wav
+p361/p361_378.wav
+p361/p361_013.wav
+p361/p361_304.wav
+p361/p361_335.wav
+p361/p361_022.wav
+p361/p361_106.wav
+p361/p361_211.wav
+p361/p361_349.wav
+p361/p361_398.wav
+p361/p361_101.wav
+p361/p361_216.wav
+p361/p361_088.wav
+p361/p361_059.wav
+p361/p361_332.wav
+p361/p361_025.wav
+p361/p361_218.wav
+p361/p361_086.wav
+p361/p361_340.wav
+p361/p361_391.wav
+p361/p361_057.wav
+p361/p361_173.wav
+p361/p361_264.wav
+p361/p361_174.wav
+p361/p361_263.wav
+p361/p361_108.wav
+p361/p361_081.wav
+p361/p361_347.wav
+p361/p361_396.wav
+p361/p361_050.wav
+p361/p361_256.wav
+p361/p361_190.wav
+p361/p361_141.wav
+p361/p361_287.wav
+p361/p361_019.wav
+p361/p361_065.wav
+p361/p361_404.wav
+p361/p361_372.wav
+p361/p361_062.wav
+p361/p361_403.wav
+p361/p361_375.wav
+p361/p361_251.wav
+p361/p361_197.wav
+p361/p361_146.wav
+p361/p361_280.wav
+p361/p361_309.wav
+p361/p361_223.wav
+p361/p361_134.wav
+p361/p361_199.wav
+p361/p361_148.wav
+p361/p361_010.wav
+p361/p361_307.wav
+p361/p361_258.wav
+p361/p361_289.wav
+p361/p361_017.wav
+p361/p361_300.wav
+p361/p361_224.wav
+p361/p361_133.wav
+p361/p361_311.wav
+p361/p361_006.wav
+p361/p361_298.wav
+p361/p361_249.wav
+p361/p361_122.wav
+p361/p361_235.wav
+p361/p361_125.wav
+p361/p361_232.wav
+p361/p361_316.wav
+p361/p361_001.wav
+p361/p361_159.wav
+p361/p361_188.wav
+p361/p361_364.wav
+p361/p361_073.wav
+p361/p361_412.wav
+p361/p361_318.wav
+p361/p361_157.wav
+p361/p361_291.wav
+p361/p361_240.wav
+p361/p361_186.wav
+p361/p361_008.wav
+p361/p361_150.wav
+p361/p361_296.wav
+p361/p361_247.wav
+p361/p361_181.wav
+p361/p361_363.wav
+p361/p361_074.wav
+p361/p361_415.wav
+p361/p361_272.wav
+p361/p361_165.wav
+p361/p361_420.wav
+p361/p361_387.wav
+p361/p361_041.wav
+p361/p361_090.wav
+p361/p361_356.wav
+p361/p361_119.wav
+p361/p361_380.wav
+p361/p361_046.wav
+p361/p361_097.wav
+p361/p361_351.wav
+p361/p361_209.wav
+p361/p361_275.wav
+p361/p361_162.wav
+p361/p361_048.wav
+p361/p361_099.wav
+p361/p361_207.wav
+p361/p361_110.wav
+p361/p361_034.wav
+p361/p361_323.wav
+p361/p361_033.wav
+p361/p361_324.wav
+p361/p361_389.wav
+p361/p361_358.wav
+p361/p361_200.wav
+p361/p361_117.wav
+p361/p361_259.wav
+p361/p361_288.wav
+p361/p361_016.wav
+p361/p361_301.wav
+p361/p361_225.wav
+p361/p361_132.wav
+p361/p361_222.wav
+p361/p361_135.wav
+p361/p361_198.wav
+p361/p361_149.wav
+p361/p361_011.wav
+p361/p361_306.wav
+p361/p361_063.wav
+p361/p361_402.wav
+p361/p361_374.wav
+p361/p361_250.wav
+p361/p361_196.wav
+p361/p361_147.wav
+p361/p361_281.wav
+p361/p361_308.wav
+p361/p361_257.wav
+p361/p361_191.wav
+p361/p361_140.wav
+p361/p361_286.wav
+p361/p361_018.wav
+p361/p361_064.wav
+p361/p361_405.wav
+p361/p361_373.wav
+p361/p361_175.wav
+p361/p361_262.wav
+p361/p361_109.wav
+p361/p361_080.wav
+p361/p361_346.wav
+p361/p361_397.wav
+p361/p361_051.wav
+p361/p361_219.wav
+p361/p361_087.wav
+p361/p361_341.wav
+p361/p361_390.wav
+p361/p361_056.wav
+p361/p361_172.wav
+p361/p361_265.wav
+p361/p361_100.wav
+p361/p361_217.wav
+p361/p361_089.wav
+p361/p361_058.wav
+p361/p361_333.wav
+p361/p361_024.wav
+p361/p361_334.wav
+p361/p361_023.wav
+p361/p361_107.wav
+p361/p361_210.wav
+p361/p361_348.wav
+p361/p361_399.wav
+p361/p361_032.wav
+p361/p361_325.wav
+p361/p361_388.wav
+p361/p361_359.wav
+p361/p361_201.wav
+p361/p361_116.wav
+p361/p361_049.wav
+p361/p361_098.wav
+p361/p361_206.wav
+p361/p361_111.wav
+p361/p361_035.wav
+p361/p361_322.wav
+p361/p361_381.wav
+p361/p361_047.wav
+p361/p361_096.wav
+p361/p361_350.wav
+p361/p361_208.wav
+p361/p361_274.wav
+p361/p361_163.wav
+p361/p361_273.wav
+p361/p361_164.wav
+p361/p361_386.wav
+p361/p361_421.wav
+p361/p361_040.wav
+p361/p361_091.wav
+p361/p361_357.wav
+p361/p361_118.wav
+p361/p361_009.wav
+p361/p361_151.wav
+p361/p361_297.wav
+p361/p361_246.wav
+p361/p361_180.wav
+p361/p361_362.wav
+p361/p361_075.wav
+p361/p361_414.wav
+p361/p361_365.wav
+p361/p361_072.wav
+p361/p361_413.wav
+p361/p361_319.wav
+p361/p361_156.wav
+p361/p361_290.wav
+p361/p361_241.wav
+p361/p361_187.wav
+p361/p361_124.wav
+p361/p361_233.wav
+p361/p361_317.wav
+p361/p361_158.wav
+p361/p361_189.wav
+p361/p361_310.wav
+p361/p361_007.wav
+p361/p361_299.wav
+p361/p361_248.wav
+p361/p361_123.wav
+p361/p361_234.wav
+p363/p363_240.wav
+p363/p363_186.wav
+p363/p363_157.wav
+p363/p363_291.wav
+p363/p363_318.wav
+p363/p363_073.wav
+p363/p363_412.wav
+p363/p363_364.wav
+p363/p363_074.wav
+p363/p363_415.wav
+p363/p363_363.wav
+p363/p363_247.wav
+p363/p363_181.wav
+p363/p363_150.wav
+p363/p363_296.wav
+p363/p363_008.wav
+p363/p363_235.wav
+p363/p363_122.wav
+p363/p363_249.wav
+p363/p363_298.wav
+p363/p363_006.wav
+p363/p363_311.wav
+p363/p363_188.wav
+p363/p363_159.wav
+p363/p363_001.wav
+p363/p363_316.wav
+p363/p363_232.wav
+p363/p363_125.wav
+p363/p363_323.wav
+p363/p363_034.wav
+p363/p363_110.wav
+p363/p363_207.wav
+p363/p363_099.wav
+p363/p363_048.wav
+p363/p363_117.wav
+p363/p363_200.wav
+p363/p363_358.wav
+p363/p363_389.wav
+p363/p363_324.wav
+p363/p363_033.wav
+p363/p363_119.wav
+p363/p363_090.wav
+p363/p363_356.wav
+p363/p363_420.wav
+p363/p363_387.wav
+p363/p363_041.wav
+p363/p363_165.wav
+p363/p363_272.wav
+p363/p363_162.wav
+p363/p363_275.wav
+p363/p363_209.wav
+p363/p363_097.wav
+p363/p363_351.wav
+p363/p363_380.wav
+p363/p363_046.wav
+p363/p363_264.wav
+p363/p363_173.wav
+p363/p363_391.wav
+p363/p363_057.wav
+p363/p363_086.wav
+p363/p363_218.wav
+p363/p363_396.wav
+p363/p363_050.wav
+p363/p363_081.wav
+p363/p363_347.wav
+p363/p363_108.wav
+p363/p363_263.wav
+p363/p363_174.wav
+p363/p363_398.wav
+p363/p363_349.wav
+p363/p363_211.wav
+p363/p363_106.wav
+p363/p363_022.wav
+p363/p363_335.wav
+p363/p363_025.wav
+p363/p363_332.wav
+p363/p363_059.wav
+p363/p363_088.wav
+p363/p363_216.wav
+p363/p363_101.wav
+p363/p363_307.wav
+p363/p363_010.wav
+p363/p363_148.wav
+p363/p363_199.wav
+p363/p363_134.wav
+p363/p363_223.wav
+p363/p363_133.wav
+p363/p363_224.wav
+p363/p363_300.wav
+p363/p363_017.wav
+p363/p363_289.wav
+p363/p363_258.wav
+p363/p363_372.wav
+p363/p363_065.wav
+p363/p363_404.wav
+p363/p363_019.wav
+p363/p363_141.wav
+p363/p363_287.wav
+p363/p363_256.wav
+p363/p363_190.wav
+p363/p363_309.wav
+p363/p363_146.wav
+p363/p363_280.wav
+p363/p363_251.wav
+p363/p363_197.wav
+p363/p363_375.wav
+p363/p363_062.wav
+p363/p363_403.wav
+p363/p363_163.wav
+p363/p363_274.wav
+p363/p363_208.wav
+p363/p363_096.wav
+p363/p363_350.wav
+p363/p363_381.wav
+p363/p363_047.wav
+p363/p363_118.wav
+p363/p363_091.wav
+p363/p363_357.wav
+p363/p363_386.wav
+p363/p363_421.wav
+p363/p363_040.wav
+p363/p363_164.wav
+p363/p363_273.wav
+p363/p363_116.wav
+p363/p363_201.wav
+p363/p363_359.wav
+p363/p363_388.wav
+p363/p363_325.wav
+p363/p363_032.wav
+p363/p363_322.wav
+p363/p363_035.wav
+p363/p363_111.wav
+p363/p363_206.wav
+p363/p363_098.wav
+p363/p363_049.wav
+p363/p363_189.wav
+p363/p363_158.wav
+p363/p363_317.wav
+p363/p363_233.wav
+p363/p363_124.wav
+p363/p363_234.wav
+p363/p363_123.wav
+p363/p363_248.wav
+p363/p363_299.wav
+p363/p363_007.wav
+p363/p363_310.wav
+p363/p363_075.wav
+p363/p363_414.wav
+p363/p363_362.wav
+p363/p363_246.wav
+p363/p363_180.wav
+p363/p363_151.wav
+p363/p363_297.wav
+p363/p363_009.wav
+p363/p363_241.wav
+p363/p363_187.wav
+p363/p363_156.wav
+p363/p363_290.wav
+p363/p363_319.wav
+p363/p363_072.wav
+p363/p363_413.wav
+p363/p363_365.wav
+p363/p363_308.wav
+p363/p363_147.wav
+p363/p363_281.wav
+p363/p363_250.wav
+p363/p363_196.wav
+p363/p363_374.wav
+p363/p363_063.wav
+p363/p363_402.wav
+p363/p363_373.wav
+p363/p363_064.wav
+p363/p363_405.wav
+p363/p363_018.wav
+p363/p363_140.wav
+p363/p363_286.wav
+p363/p363_257.wav
+p363/p363_191.wav
+p363/p363_132.wav
+p363/p363_225.wav
+p363/p363_301.wav
+p363/p363_016.wav
+p363/p363_288.wav
+p363/p363_259.wav
+p363/p363_306.wav
+p363/p363_011.wav
+p363/p363_149.wav
+p363/p363_198.wav
+p363/p363_135.wav
+p363/p363_222.wav
+p363/p363_024.wav
+p363/p363_333.wav
+p363/p363_058.wav
+p363/p363_089.wav
+p363/p363_217.wav
+p363/p363_100.wav
+p363/p363_399.wav
+p363/p363_348.wav
+p363/p363_210.wav
+p363/p363_107.wav
+p363/p363_023.wav
+p363/p363_334.wav
+p363/p363_397.wav
+p363/p363_051.wav
+p363/p363_080.wav
+p363/p363_346.wav
+p363/p363_109.wav
+p363/p363_262.wav
+p363/p363_175.wav
+p363/p363_265.wav
+p363/p363_172.wav
+p363/p363_390.wav
+p363/p363_056.wav
+p363/p363_087.wav
+p363/p363_341.wav
+p363/p363_219.wav
+p363/p363_370.wav
+p363/p363_406.wav
+p363/p363_067.wav
+p363/p363_228.wav
+p363/p363_285.wav
+p363/p363_143.wav
+p363/p363_192.wav
+p363/p363_254.wav
+p363/p363_282.wav
+p363/p363_144.wav
+p363/p363_195.wav
+p363/p363_253.wav
+p363/p363_377.wav
+p363/p363_401.wav
+p363/p363_060.wav
+p363/p363_138.wav
+p363/p363_305.wav
+p363/p363_012.wav
+p363/p363_379.wav
+p363/p363_136.wav
+p363/p363_221.wav
+p363/p363_408.wav
+p363/p363_069.wav
+p363/p363_131.wav
+p363/p363_226.wav
+p363/p363_302.wav
+p363/p363_015.wav
+p363/p363_213.wav
+p363/p363_104.wav
+p363/p363_020.wav
+p363/p363_337.wav
+p363/p363_178.wav
+p363/p363_027.wav
+p363/p363_330.wav
+p363/p363_268.wav
+p363/p363_214.wav
+p363/p363_103.wav
+p363/p363_029.wav
+p363/p363_266.wav
+p363/p363_171.wav
+p363/p363_055.wav
+p363/p363_393.wav
+p363/p363_084.wav
+p363/p363_052.wav
+p363/p363_394.wav
+p363/p363_345.wav
+p363/p363_083.wav
+p363/p363_339.wav
+p363/p363_261.wav
+p363/p363_176.wav
+p363/p363_354.wav
+p363/p363_092.wav
+p363/p363_043.wav
+p363/p363_385.wav
+p363/p363_422.wav
+p363/p363_167.wav
+p363/p363_270.wav
+p363/p363_328.wav
+p363/p363_160.wav
+p363/p363_277.wav
+p363/p363_038.wav
+p363/p363_353.wav
+p363/p363_095.wav
+p363/p363_044.wav
+p363/p363_382.wav
+p363/p363_321.wav
+p363/p363_036.wav
+p363/p363_112.wav
+p363/p363_205.wav
+p363/p363_115.wav
+p363/p363_202.wav
+p363/p363_169.wav
+p363/p363_326.wav
+p363/p363_031.wav
+p363/p363_237.wav
+p363/p363_120.wav
+p363/p363_419.wav
+p363/p363_078.wav
+p363/p363_004.wav
+p363/p363_313.wav
+p363/p363_003.wav
+p363/p363_314.wav
+p363/p363_230.wav
+p363/p363_127.wav
+p363/p363_368.wav
+p363/p363_184.wav
+p363/p363_242.wav
+p363/p363_293.wav
+p363/p363_155.wav
+p363/p363_129.wav
+p363/p363_410.wav
+p363/p363_071.wav
+p363/p363_366.wav
+p363/p363_239.wav
+p363/p363_417.wav
+p363/p363_076.wav
+p363/p363_361.wav
+p363/p363_183.wav
+p363/p363_245.wav
+p363/p363_294.wav
+p363/p363_152.wav
+p363/p363_053.wav
+p363/p363_395.wav
+p363/p363_344.wav
+p363/p363_082.wav
+p363/p363_338.wav
+p363/p363_260.wav
+p363/p363_177.wav
+p363/p363_028.wav
+p363/p363_267.wav
+p363/p363_170.wav
+p363/p363_054.wav
+p363/p363_392.wav
+p363/p363_343.wav
+p363/p363_085.wav
+p363/p363_026.wav
+p363/p363_331.wav
+p363/p363_269.wav
+p363/p363_215.wav
+p363/p363_102.wav
+p363/p363_212.wav
+p363/p363_105.wav
+p363/p363_021.wav
+p363/p363_336.wav
+p363/p363_179.wav
+p363/p363_409.wav
+p363/p363_068.wav
+p363/p363_130.wav
+p363/p363_227.wav
+p363/p363_303.wav
+p363/p363_014.wav
+p363/p363_304.wav
+p363/p363_013.wav
+p363/p363_378.wav
+p363/p363_137.wav
+p363/p363_220.wav
+p363/p363_283.wav
+p363/p363_145.wav
+p363/p363_194.wav
+p363/p363_252.wav
+p363/p363_376.wav
+p363/p363_061.wav
+p363/p363_139.wav
+p363/p363_371.wav
+p363/p363_407.wav
+p363/p363_066.wav
+p363/p363_229.wav
+p363/p363_284.wav
+p363/p363_142.wav
+p363/p363_193.wav
+p363/p363_255.wav
+p363/p363_238.wav
+p363/p363_416.wav
+p363/p363_077.wav
+p363/p363_360.wav
+p363/p363_182.wav
+p363/p363_244.wav
+p363/p363_295.wav
+p363/p363_153.wav
+p363/p363_185.wav
+p363/p363_243.wav
+p363/p363_292.wav
+p363/p363_154.wav
+p363/p363_128.wav
+p363/p363_411.wav
+p363/p363_070.wav
+p363/p363_367.wav
+p363/p363_002.wav
+p363/p363_315.wav
+p363/p363_231.wav
+p363/p363_126.wav
+p363/p363_369.wav
+p363/p363_236.wav
+p363/p363_121.wav
+p363/p363_418.wav
+p363/p363_079.wav
+p363/p363_005.wav
+p363/p363_312.wav
+p363/p363_114.wav
+p363/p363_203.wav
+p363/p363_168.wav
+p363/p363_327.wav
+p363/p363_030.wav
+p363/p363_278.wav
+p363/p363_320.wav
+p363/p363_037.wav
+p363/p363_113.wav
+p363/p363_204.wav
+p363/p363_161.wav
+p363/p363_276.wav
+p363/p363_039.wav
+p363/p363_352.wav
+p363/p363_094.wav
+p363/p363_045.wav
+p363/p363_383.wav
+p363/p363_355.wav
+p363/p363_093.wav
+p363/p363_042.wav
+p363/p363_423.wav
+p363/p363_384.wav
+p363/p363_166.wav
+p363/p363_271.wav
+p363/p363_329.wav
+p351/p351_399.wav
+p351/p351_348.wav
+p351/p351_210.wav
+p351/p351_107.wav
+p351/p351_023.wav
+p351/p351_334.wav
+p351/p351_024.wav
+p351/p351_333.wav
+p351/p351_058.wav
+p351/p351_089.wav
+p351/p351_217.wav
+p351/p351_100.wav
+p351/p351_265.wav
+p351/p351_172.wav
+p351/p351_390.wav
+p351/p351_056.wav
+p351/p351_087.wav
+p351/p351_341.wav
+p351/p351_219.wav
+p351/p351_397.wav
+p351/p351_051.wav
+p351/p351_080.wav
+p351/p351_346.wav
+p351/p351_109.wav
+p351/p351_262.wav
+p351/p351_175.wav
+p351/p351_373.wav
+p351/p351_064.wav
+p351/p351_405.wav
+p351/p351_018.wav
+p351/p351_140.wav
+p351/p351_286.wav
+p351/p351_257.wav
+p351/p351_191.wav
+p351/p351_308.wav
+p351/p351_147.wav
+p351/p351_281.wav
+p351/p351_250.wav
+p351/p351_196.wav
+p351/p351_374.wav
+p351/p351_063.wav
+p351/p351_402.wav
+p351/p351_306.wav
+p351/p351_011.wav
+p351/p351_149.wav
+p351/p351_198.wav
+p351/p351_135.wav
+p351/p351_222.wav
+p351/p351_132.wav
+p351/p351_225.wav
+p351/p351_301.wav
+p351/p351_016.wav
+p351/p351_288.wav
+p351/p351_259.wav
+p351/p351_234.wav
+p351/p351_123.wav
+p351/p351_248.wav
+p351/p351_299.wav
+p351/p351_007.wav
+p351/p351_310.wav
+p351/p351_189.wav
+p351/p351_158.wav
+p351/p351_317.wav
+p351/p351_233.wav
+p351/p351_124.wav
+p351/p351_241.wav
+p351/p351_187.wav
+p351/p351_156.wav
+p351/p351_290.wav
+p351/p351_319.wav
+p351/p351_072.wav
+p351/p351_413.wav
+p351/p351_365.wav
+p351/p351_075.wav
+p351/p351_414.wav
+p351/p351_362.wav
+p351/p351_246.wav
+p351/p351_180.wav
+p351/p351_151.wav
+p351/p351_297.wav
+p351/p351_009.wav
+p351/p351_118.wav
+p351/p351_091.wav
+p351/p351_357.wav
+p351/p351_386.wav
+p351/p351_421.wav
+p351/p351_040.wav
+p351/p351_164.wav
+p351/p351_273.wav
+p351/p351_163.wav
+p351/p351_274.wav
+p351/p351_208.wav
+p351/p351_096.wav
+p351/p351_350.wav
+p351/p351_381.wav
+p351/p351_047.wav
+p351/p351_322.wav
+p351/p351_035.wav
+p351/p351_111.wav
+p351/p351_206.wav
+p351/p351_098.wav
+p351/p351_049.wav
+p351/p351_116.wav
+p351/p351_201.wav
+p351/p351_359.wav
+p351/p351_388.wav
+p351/p351_325.wav
+p351/p351_032.wav
+p351/p351_133.wav
+p351/p351_224.wav
+p351/p351_300.wav
+p351/p351_017.wav
+p351/p351_289.wav
+p351/p351_258.wav
+p351/p351_307.wav
+p351/p351_010.wav
+p351/p351_148.wav
+p351/p351_199.wav
+p351/p351_134.wav
+p351/p351_223.wav
+p351/p351_309.wav
+p351/p351_146.wav
+p351/p351_280.wav
+p351/p351_251.wav
+p351/p351_197.wav
+p351/p351_375.wav
+p351/p351_062.wav
+p351/p351_403.wav
+p351/p351_372.wav
+p351/p351_065.wav
+p351/p351_404.wav
+p351/p351_019.wav
+p351/p351_141.wav
+p351/p351_287.wav
+p351/p351_256.wav
+p351/p351_190.wav
+p351/p351_396.wav
+p351/p351_050.wav
+p351/p351_081.wav
+p351/p351_347.wav
+p351/p351_108.wav
+p351/p351_263.wav
+p351/p351_174.wav
+p351/p351_264.wav
+p351/p351_173.wav
+p351/p351_391.wav
+p351/p351_057.wav
+p351/p351_086.wav
+p351/p351_340.wav
+p351/p351_218.wav
+p351/p351_025.wav
+p351/p351_332.wav
+p351/p351_059.wav
+p351/p351_088.wav
+p351/p351_216.wav
+p351/p351_101.wav
+p351/p351_398.wav
+p351/p351_349.wav
+p351/p351_211.wav
+p351/p351_106.wav
+p351/p351_022.wav
+p351/p351_335.wav
+p351/p351_117.wav
+p351/p351_200.wav
+p351/p351_358.wav
+p351/p351_389.wav
+p351/p351_324.wav
+p351/p351_033.wav
+p351/p351_323.wav
+p351/p351_034.wav
+p351/p351_110.wav
+p351/p351_207.wav
+p351/p351_099.wav
+p351/p351_048.wav
+p351/p351_162.wav
+p351/p351_275.wav
+p351/p351_209.wav
+p351/p351_097.wav
+p351/p351_351.wav
+p351/p351_380.wav
+p351/p351_046.wav
+p351/p351_119.wav
+p351/p351_090.wav
+p351/p351_356.wav
+p351/p351_420.wav
+p351/p351_387.wav
+p351/p351_041.wav
+p351/p351_165.wav
+p351/p351_272.wav
+p351/p351_074.wav
+p351/p351_415.wav
+p351/p351_363.wav
+p351/p351_247.wav
+p351/p351_181.wav
+p351/p351_150.wav
+p351/p351_296.wav
+p351/p351_008.wav
+p351/p351_240.wav
+p351/p351_186.wav
+p351/p351_157.wav
+p351/p351_291.wav
+p351/p351_318.wav
+p351/p351_073.wav
+p351/p351_412.wav
+p351/p351_364.wav
+p351/p351_188.wav
+p351/p351_159.wav
+p351/p351_001.wav
+p351/p351_316.wav
+p351/p351_232.wav
+p351/p351_125.wav
+p351/p351_235.wav
+p351/p351_122.wav
+p351/p351_249.wav
+p351/p351_298.wav
+p351/p351_006.wav
+p351/p351_311.wav
+p351/p351_278.wav
+p351/p351_320.wav
+p351/p351_037.wav
+p351/p351_113.wav
+p351/p351_204.wav
+p351/p351_114.wav
+p351/p351_203.wav
+p351/p351_168.wav
+p351/p351_327.wav
+p351/p351_030.wav
+p351/p351_355.wav
+p351/p351_093.wav
+p351/p351_042.wav
+p351/p351_423.wav
+p351/p351_384.wav
+p351/p351_166.wav
+p351/p351_271.wav
+p351/p351_329.wav
+p351/p351_161.wav
+p351/p351_276.wav
+p351/p351_039.wav
+p351/p351_352.wav
+p351/p351_094.wav
+p351/p351_045.wav
+p351/p351_383.wav
+p351/p351_424.wav
+p351/p351_185.wav
+p351/p351_243.wav
+p351/p351_292.wav
+p351/p351_154.wav
+p351/p351_128.wav
+p351/p351_411.wav
+p351/p351_070.wav
+p351/p351_367.wav
+p351/p351_238.wav
+p351/p351_416.wav
+p351/p351_077.wav
+p351/p351_360.wav
+p351/p351_182.wav
+p351/p351_244.wav
+p351/p351_295.wav
+p351/p351_153.wav
+p351/p351_236.wav
+p351/p351_121.wav
+p351/p351_418.wav
+p351/p351_079.wav
+p351/p351_005.wav
+p351/p351_312.wav
+p351/p351_002.wav
+p351/p351_315.wav
+p351/p351_231.wav
+p351/p351_126.wav
+p351/p351_369.wav
+p351/p351_304.wav
+p351/p351_013.wav
+p351/p351_378.wav
+p351/p351_137.wav
+p351/p351_220.wav
+p351/p351_409.wav
+p351/p351_068.wav
+p351/p351_130.wav
+p351/p351_227.wav
+p351/p351_303.wav
+p351/p351_014.wav
+p351/p351_371.wav
+p351/p351_407.wav
+p351/p351_066.wav
+p351/p351_229.wav
+p351/p351_284.wav
+p351/p351_142.wav
+p351/p351_193.wav
+p351/p351_255.wav
+p351/p351_283.wav
+p351/p351_145.wav
+p351/p351_194.wav
+p351/p351_252.wav
+p351/p351_376.wav
+p351/p351_400.wav
+p351/p351_061.wav
+p351/p351_139.wav
+p351/p351_028.wav
+p351/p351_267.wav
+p351/p351_170.wav
+p351/p351_054.wav
+p351/p351_392.wav
+p351/p351_343.wav
+p351/p351_085.wav
+p351/p351_053.wav
+p351/p351_395.wav
+p351/p351_344.wav
+p351/p351_082.wav
+p351/p351_338.wav
+p351/p351_260.wav
+p351/p351_177.wav
+p351/p351_212.wav
+p351/p351_021.wav
+p351/p351_336.wav
+p351/p351_179.wav
+p351/p351_026.wav
+p351/p351_331.wav
+p351/p351_269.wav
+p351/p351_215.wav
+p351/p351_102.wav
+p351/p351_003.wav
+p351/p351_314.wav
+p351/p351_230.wav
+p351/p351_127.wav
+p351/p351_368.wav
+p351/p351_237.wav
+p351/p351_120.wav
+p351/p351_419.wav
+p351/p351_078.wav
+p351/p351_004.wav
+p351/p351_313.wav
+p351/p351_239.wav
+p351/p351_417.wav
+p351/p351_076.wav
+p351/p351_361.wav
+p351/p351_183.wav
+p351/p351_245.wav
+p351/p351_294.wav
+p351/p351_152.wav
+p351/p351_184.wav
+p351/p351_242.wav
+p351/p351_293.wav
+p351/p351_155.wav
+p351/p351_129.wav
+p351/p351_410.wav
+p351/p351_071.wav
+p351/p351_366.wav
+p351/p351_160.wav
+p351/p351_277.wav
+p351/p351_038.wav
+p351/p351_353.wav
+p351/p351_095.wav
+p351/p351_044.wav
+p351/p351_382.wav
+p351/p351_354.wav
+p351/p351_092.wav
+p351/p351_043.wav
+p351/p351_385.wav
+p351/p351_422.wav
+p351/p351_167.wav
+p351/p351_270.wav
+p351/p351_328.wav
+p351/p351_115.wav
+p351/p351_202.wav
+p351/p351_169.wav
+p351/p351_326.wav
+p351/p351_031.wav
+p351/p351_279.wav
+p351/p351_321.wav
+p351/p351_036.wav
+p351/p351_112.wav
+p351/p351_205.wav
+p351/p351_027.wav
+p351/p351_330.wav
+p351/p351_268.wav
+p351/p351_214.wav
+p351/p351_103.wav
+p351/p351_213.wav
+p351/p351_104.wav
+p351/p351_020.wav
+p351/p351_337.wav
+p351/p351_178.wav
+p351/p351_052.wav
+p351/p351_394.wav
+p351/p351_345.wav
+p351/p351_083.wav
+p351/p351_339.wav
+p351/p351_261.wav
+p351/p351_176.wav
+p351/p351_029.wav
+p351/p351_266.wav
+p351/p351_171.wav
+p351/p351_055.wav
+p351/p351_393.wav
+p351/p351_342.wav
+p351/p351_084.wav
+p351/p351_282.wav
+p351/p351_144.wav
+p351/p351_195.wav
+p351/p351_253.wav
+p351/p351_377.wav
+p351/p351_401.wav
+p351/p351_060.wav
+p351/p351_138.wav
+p351/p351_370.wav
+p351/p351_406.wav
+p351/p351_067.wav
+p351/p351_228.wav
+p351/p351_285.wav
+p351/p351_143.wav
+p351/p351_192.wav
+p351/p351_254.wav
+p351/p351_408.wav
+p351/p351_069.wav
+p351/p351_131.wav
+p351/p351_226.wav
+p351/p351_302.wav
+p351/p351_015.wav
+p351/p351_305.wav
+p351/p351_012.wav
+p351/p351_379.wav
+p351/p351_136.wav
+p351/p351_221.wav
+p364/p364_007.wav
+p364/p364_248.wav
+p364/p364_299.wav
+p364/p364_234.wav
+p364/p364_123.wav
+p364/p364_233.wav
+p364/p364_124.wav
+p364/p364_189.wav
+p364/p364_158.wav
+p364/p364_072.wav
+p364/p364_187.wav
+p364/p364_241.wav
+p364/p364_290.wav
+p364/p364_156.wav
+p364/p364_009.wav
+p364/p364_180.wav
+p364/p364_246.wav
+p364/p364_297.wav
+p364/p364_151.wav
+p364/p364_075.wav
+p364/p364_164.wav
+p364/p364_273.wav
+p364/p364_091.wav
+p364/p364_040.wav
+p364/p364_118.wav
+p364/p364_096.wav
+p364/p364_047.wav
+p364/p364_208.wav
+p364/p364_163.wav
+p364/p364_274.wav
+p364/p364_098.wav
+p364/p364_049.wav
+p364/p364_111.wav
+p364/p364_206.wav
+p364/p364_035.wav
+p364/p364_032.wav
+p364/p364_116.wav
+p364/p364_201.wav
+p364/p364_023.wav
+p364/p364_210.wav
+p364/p364_107.wav
+p364/p364_217.wav
+p364/p364_100.wav
+p364/p364_058.wav
+p364/p364_089.wav
+p364/p364_024.wav
+p364/p364_219.wav
+p364/p364_056.wav
+p364/p364_087.wav
+p364/p364_265.wav
+p364/p364_172.wav
+p364/p364_262.wav
+p364/p364_175.wav
+p364/p364_109.wav
+p364/p364_051.wav
+p364/p364_286.wav
+p364/p364_140.wav
+p364/p364_191.wav
+p364/p364_257.wav
+p364/p364_018.wav
+p364/p364_064.wav
+p364/p364_063.wav
+p364/p364_281.wav
+p364/p364_147.wav
+p364/p364_196.wav
+p364/p364_250.wav
+p364/p364_308.wav
+p364/p364_135.wav
+p364/p364_222.wav
+p364/p364_149.wav
+p364/p364_198.wav
+p364/p364_306.wav
+p364/p364_011.wav
+p364/p364_288.wav
+p364/p364_259.wav
+p364/p364_301.wav
+p364/p364_016.wav
+p364/p364_132.wav
+p364/p364_225.wav
+p364/p364_033.wav
+p364/p364_117.wav
+p364/p364_200.wav
+p364/p364_099.wav
+p364/p364_048.wav
+p364/p364_110.wav
+p364/p364_207.wav
+p364/p364_034.wav
+p364/p364_097.wav
+p364/p364_046.wav
+p364/p364_209.wav
+p364/p364_162.wav
+p364/p364_275.wav
+p364/p364_165.wav
+p364/p364_272.wav
+p364/p364_090.wav
+p364/p364_041.wav
+p364/p364_119.wav
+p364/p364_008.wav
+p364/p364_181.wav
+p364/p364_247.wav
+p364/p364_150.wav
+p364/p364_074.wav
+p364/p364_073.wav
+p364/p364_240.wav
+p364/p364_291.wav
+p364/p364_157.wav
+p364/p364_232.wav
+p364/p364_125.wav
+p364/p364_001.wav
+p364/p364_188.wav
+p364/p364_159.wav
+p364/p364_006.wav
+p364/p364_249.wav
+p364/p364_298.wav
+p364/p364_235.wav
+p364/p364_122.wav
+p364/p364_289.wav
+p364/p364_258.wav
+p364/p364_300.wav
+p364/p364_017.wav
+p364/p364_133.wav
+p364/p364_224.wav
+p364/p364_134.wav
+p364/p364_223.wav
+p364/p364_148.wav
+p364/p364_199.wav
+p364/p364_010.wav
+p364/p364_062.wav
+p364/p364_280.wav
+p364/p364_146.wav
+p364/p364_197.wav
+p364/p364_251.wav
+p364/p364_309.wav
+p364/p364_287.wav
+p364/p364_141.wav
+p364/p364_190.wav
+p364/p364_256.wav
+p364/p364_019.wav
+p364/p364_065.wav
+p364/p364_263.wav
+p364/p364_174.wav
+p364/p364_108.wav
+p364/p364_050.wav
+p364/p364_081.wav
+p364/p364_218.wav
+p364/p364_057.wav
+p364/p364_086.wav
+p364/p364_264.wav
+p364/p364_173.wav
+p364/p364_216.wav
+p364/p364_101.wav
+p364/p364_059.wav
+p364/p364_088.wav
+p364/p364_025.wav
+p364/p364_022.wav
+p364/p364_211.wav
+p364/p364_106.wav
+p364/p364_137.wav
+p364/p364_220.wav
+p364/p364_304.wav
+p364/p364_013.wav
+p364/p364_303.wav
+p364/p364_014.wav
+p364/p364_130.wav
+p364/p364_227.wav
+p364/p364_068.wav
+p364/p364_142.wav
+p364/p364_284.wav
+p364/p364_255.wav
+p364/p364_193.wav
+p364/p364_229.wav
+p364/p364_066.wav
+p364/p364_061.wav
+p364/p364_145.wav
+p364/p364_283.wav
+p364/p364_252.wav
+p364/p364_194.wav
+p364/p364_054.wav
+p364/p364_085.wav
+p364/p364_267.wav
+p364/p364_170.wav
+p364/p364_028.wav
+p364/p364_260.wav
+p364/p364_177.wav
+p364/p364_053.wav
+p364/p364_082.wav
+p364/p364_179.wav
+p364/p364_021.wav
+p364/p364_212.wav
+p364/p364_105.wav
+p364/p364_215.wav
+p364/p364_102.wav
+p364/p364_269.wav
+p364/p364_026.wav
+p364/p364_113.wav
+p364/p364_204.wav
+p364/p364_037.wav
+p364/p364_278.wav
+p364/p364_030.wav
+p364/p364_168.wav
+p364/p364_114.wav
+p364/p364_203.wav
+p364/p364_166.wav
+p364/p364_271.wav
+p364/p364_042.wav
+p364/p364_094.wav
+p364/p364_045.wav
+p364/p364_039.wav
+p364/p364_161.wav
+p364/p364_276.wav
+p364/p364_070.wav
+p364/p364_128.wav
+p364/p364_243.wav
+p364/p364_185.wav
+p364/p364_154.wav
+p364/p364_292.wav
+p364/p364_244.wav
+p364/p364_182.wav
+p364/p364_153.wav
+p364/p364_077.wav
+p364/p364_238.wav
+p364/p364_005.wav
+p364/p364_079.wav
+p364/p364_236.wav
+p364/p364_121.wav
+p364/p364_231.wav
+p364/p364_126.wav
+p364/p364_002.wav
+p364/p364_214.wav
+p364/p364_103.wav
+p364/p364_268.wav
+p364/p364_027.wav
+p364/p364_178.wav
+p364/p364_020.wav
+p364/p364_213.wav
+p364/p364_104.wav
+p364/p364_261.wav
+p364/p364_176.wav
+p364/p364_083.wav
+p364/p364_055.wav
+p364/p364_084.wav
+p364/p364_266.wav
+p364/p364_171.wav
+p364/p364_029.wav
+p364/p364_138.wav
+p364/p364_060.wav
+p364/p364_144.wav
+p364/p364_282.wav
+p364/p364_253.wav
+p364/p364_195.wav
+p364/p364_143.wav
+p364/p364_285.wav
+p364/p364_254.wav
+p364/p364_192.wav
+p364/p364_228.wav
+p364/p364_067.wav
+p364/p364_302.wav
+p364/p364_015.wav
+p364/p364_131.wav
+p364/p364_226.wav
+p364/p364_069.wav
+p364/p364_136.wav
+p364/p364_221.wav
+p364/p364_305.wav
+p364/p364_012.wav
+p364/p364_230.wav
+p364/p364_127.wav
+p364/p364_003.wav
+p364/p364_004.wav
+p364/p364_078.wav
+p364/p364_237.wav
+p364/p364_120.wav
+p364/p364_245.wav
+p364/p364_183.wav
+p364/p364_152.wav
+p364/p364_294.wav
+p364/p364_076.wav
+p364/p364_239.wav
+p364/p364_071.wav
+p364/p364_129.wav
+p364/p364_242.wav
+p364/p364_184.wav
+p364/p364_155.wav
+p364/p364_293.wav
+p364/p364_095.wav
+p364/p364_044.wav
+p364/p364_038.wav
+p364/p364_160.wav
+p364/p364_277.wav
+p364/p364_167.wav
+p364/p364_270.wav
+p364/p364_092.wav
+p364/p364_043.wav
+p364/p364_031.wav
+p364/p364_169.wav
+p364/p364_115.wav
+p364/p364_202.wav
+p364/p364_112.wav
+p364/p364_205.wav
+p364/p364_036.wav
+p364/p364_279.wav
+p360/p360_149.wav
+p360/p360_198.wav
+p360/p360_306.wav
+p360/p360_011.wav
+p360/p360_135.wav
+p360/p360_222.wav
+p360/p360_132.wav
+p360/p360_225.wav
+p360/p360_288.wav
+p360/p360_259.wav
+p360/p360_301.wav
+p360/p360_016.wav
+p360/p360_373.wav
+p360/p360_405.wav
+p360/p360_064.wav
+p360/p360_286.wav
+p360/p360_140.wav
+p360/p360_191.wav
+p360/p360_257.wav
+p360/p360_018.wav
+p360/p360_281.wav
+p360/p360_147.wav
+p360/p360_196.wav
+p360/p360_250.wav
+p360/p360_308.wav
+p360/p360_374.wav
+p360/p360_402.wav
+p360/p360_063.wav
+p360/p360_265.wav
+p360/p360_172.wav
+p360/p360_219.wav
+p360/p360_056.wav
+p360/p360_390.wav
+p360/p360_341.wav
+p360/p360_087.wav
+p360/p360_109.wav
+p360/p360_051.wav
+p360/p360_397.wav
+p360/p360_346.wav
+p360/p360_080.wav
+p360/p360_262.wav
+p360/p360_175.wav
+p360/p360_210.wav
+p360/p360_107.wav
+p360/p360_399.wav
+p360/p360_348.wav
+p360/p360_023.wav
+p360/p360_334.wav
+p360/p360_024.wav
+p360/p360_333.wav
+p360/p360_217.wav
+p360/p360_100.wav
+p360/p360_058.wav
+p360/p360_089.wav
+p360/p360_322.wav
+p360/p360_035.wav
+p360/p360_098.wav
+p360/p360_049.wav
+p360/p360_111.wav
+p360/p360_206.wav
+p360/p360_359.wav
+p360/p360_388.wav
+p360/p360_116.wav
+p360/p360_201.wav
+p360/p360_325.wav
+p360/p360_032.wav
+p360/p360_357.wav
+p360/p360_091.wav
+p360/p360_040.wav
+p360/p360_421.wav
+p360/p360_386.wav
+p360/p360_118.wav
+p360/p360_164.wav
+p360/p360_273.wav
+p360/p360_163.wav
+p360/p360_274.wav
+p360/p360_350.wav
+p360/p360_096.wav
+p360/p360_047.wav
+p360/p360_381.wav
+p360/p360_208.wav
+p360/p360_319.wav
+p360/p360_187.wav
+p360/p360_241.wav
+p360/p360_290.wav
+p360/p360_156.wav
+p360/p360_413.wav
+p360/p360_072.wav
+p360/p360_365.wav
+p360/p360_414.wav
+p360/p360_075.wav
+p360/p360_362.wav
+p360/p360_009.wav
+p360/p360_180.wav
+p360/p360_246.wav
+p360/p360_297.wav
+p360/p360_151.wav
+p360/p360_234.wav
+p360/p360_123.wav
+p360/p360_007.wav
+p360/p360_310.wav
+p360/p360_248.wav
+p360/p360_299.wav
+p360/p360_317.wav
+p360/p360_189.wav
+p360/p360_158.wav
+p360/p360_233.wav
+p360/p360_124.wav
+p360/p360_025.wav
+p360/p360_332.wav
+p360/p360_216.wav
+p360/p360_101.wav
+p360/p360_059.wav
+p360/p360_088.wav
+p360/p360_211.wav
+p360/p360_106.wav
+p360/p360_398.wav
+p360/p360_349.wav
+p360/p360_022.wav
+p360/p360_335.wav
+p360/p360_108.wav
+p360/p360_050.wav
+p360/p360_396.wav
+p360/p360_347.wav
+p360/p360_081.wav
+p360/p360_263.wav
+p360/p360_174.wav
+p360/p360_264.wav
+p360/p360_173.wav
+p360/p360_218.wav
+p360/p360_057.wav
+p360/p360_391.wav
+p360/p360_340.wav
+p360/p360_086.wav
+p360/p360_280.wav
+p360/p360_146.wav
+p360/p360_197.wav
+p360/p360_251.wav
+p360/p360_309.wav
+p360/p360_375.wav
+p360/p360_403.wav
+p360/p360_062.wav
+p360/p360_372.wav
+p360/p360_404.wav
+p360/p360_065.wav
+p360/p360_287.wav
+p360/p360_141.wav
+p360/p360_190.wav
+p360/p360_256.wav
+p360/p360_019.wav
+p360/p360_133.wav
+p360/p360_224.wav
+p360/p360_289.wav
+p360/p360_258.wav
+p360/p360_300.wav
+p360/p360_017.wav
+p360/p360_148.wav
+p360/p360_199.wav
+p360/p360_307.wav
+p360/p360_010.wav
+p360/p360_134.wav
+p360/p360_223.wav
+p360/p360_001.wav
+p360/p360_316.wav
+p360/p360_188.wav
+p360/p360_159.wav
+p360/p360_232.wav
+p360/p360_125.wav
+p360/p360_235.wav
+p360/p360_122.wav
+p360/p360_006.wav
+p360/p360_311.wav
+p360/p360_249.wav
+p360/p360_298.wav
+p360/p360_415.wav
+p360/p360_074.wav
+p360/p360_363.wav
+p360/p360_008.wav
+p360/p360_181.wav
+p360/p360_247.wav
+p360/p360_296.wav
+p360/p360_150.wav
+p360/p360_318.wav
+p360/p360_186.wav
+p360/p360_240.wav
+p360/p360_291.wav
+p360/p360_157.wav
+p360/p360_412.wav
+p360/p360_073.wav
+p360/p360_364.wav
+p360/p360_162.wav
+p360/p360_275.wav
+p360/p360_351.wav
+p360/p360_097.wav
+p360/p360_046.wav
+p360/p360_380.wav
+p360/p360_209.wav
+p360/p360_356.wav
+p360/p360_090.wav
+p360/p360_041.wav
+p360/p360_387.wav
+p360/p360_420.wav
+p360/p360_119.wav
+p360/p360_165.wav
+p360/p360_272.wav
+p360/p360_358.wav
+p360/p360_389.wav
+p360/p360_117.wav
+p360/p360_200.wav
+p360/p360_324.wav
+p360/p360_033.wav
+p360/p360_323.wav
+p360/p360_034.wav
+p360/p360_099.wav
+p360/p360_048.wav
+p360/p360_110.wav
+p360/p360_207.wav
+p360/p360_079.wav
+p360/p360_418.wav
+p360/p360_236.wav
+p360/p360_121.wav
+p360/p360_005.wav
+p360/p360_312.wav
+p360/p360_002.wav
+p360/p360_315.wav
+p360/p360_369.wav
+p360/p360_231.wav
+p360/p360_126.wav
+p360/p360_243.wav
+p360/p360_185.wav
+p360/p360_154.wav
+p360/p360_292.wav
+p360/p360_070.wav
+p360/p360_411.wav
+p360/p360_367.wav
+p360/p360_128.wav
+p360/p360_077.wav
+p360/p360_416.wav
+p360/p360_360.wav
+p360/p360_238.wav
+p360/p360_244.wav
+p360/p360_182.wav
+p360/p360_153.wav
+p360/p360_295.wav
+p360/p360_093.wav
+p360/p360_355.wav
+p360/p360_384.wav
+p360/p360_423.wav
+p360/p360_042.wav
+p360/p360_329.wav
+p360/p360_166.wav
+p360/p360_271.wav
+p360/p360_039.wav
+p360/p360_161.wav
+p360/p360_276.wav
+p360/p360_094.wav
+p360/p360_352.wav
+p360/p360_424.wav
+p360/p360_383.wav
+p360/p360_045.wav
+p360/p360_320.wav
+p360/p360_037.wav
+p360/p360_278.wav
+p360/p360_113.wav
+p360/p360_204.wav
+p360/p360_114.wav
+p360/p360_203.wav
+p360/p360_327.wav
+p360/p360_030.wav
+p360/p360_168.wav
+p360/p360_212.wav
+p360/p360_105.wav
+p360/p360_179.wav
+p360/p360_021.wav
+p360/p360_336.wav
+p360/p360_269.wav
+p360/p360_026.wav
+p360/p360_331.wav
+p360/p360_215.wav
+p360/p360_102.wav
+p360/p360_267.wav
+p360/p360_170.wav
+p360/p360_028.wav
+p360/p360_392.wav
+p360/p360_054.wav
+p360/p360_085.wav
+p360/p360_343.wav
+p360/p360_395.wav
+p360/p360_053.wav
+p360/p360_082.wav
+p360/p360_344.wav
+p360/p360_260.wav
+p360/p360_177.wav
+p360/p360_338.wav
+p360/p360_229.wav
+p360/p360_371.wav
+p360/p360_066.wav
+p360/p360_407.wav
+p360/p360_142.wav
+p360/p360_284.wav
+p360/p360_255.wav
+p360/p360_193.wav
+p360/p360_145.wav
+p360/p360_283.wav
+p360/p360_252.wav
+p360/p360_194.wav
+p360/p360_139.wav
+p360/p360_376.wav
+p360/p360_061.wav
+p360/p360_400.wav
+p360/p360_304.wav
+p360/p360_013.wav
+p360/p360_137.wav
+p360/p360_220.wav
+p360/p360_378.wav
+p360/p360_130.wav
+p360/p360_227.wav
+p360/p360_068.wav
+p360/p360_409.wav
+p360/p360_303.wav
+p360/p360_014.wav
+p360/p360_115.wav
+p360/p360_202.wav
+p360/p360_326.wav
+p360/p360_031.wav
+p360/p360_169.wav
+p360/p360_321.wav
+p360/p360_036.wav
+p360/p360_279.wav
+p360/p360_112.wav
+p360/p360_205.wav
+p360/p360_038.wav
+p360/p360_160.wav
+p360/p360_277.wav
+p360/p360_095.wav
+p360/p360_353.wav
+p360/p360_382.wav
+p360/p360_044.wav
+p360/p360_092.wav
+p360/p360_354.wav
+p360/p360_422.wav
+p360/p360_385.wav
+p360/p360_043.wav
+p360/p360_328.wav
+p360/p360_167.wav
+p360/p360_270.wav
+p360/p360_076.wav
+p360/p360_417.wav
+p360/p360_361.wav
+p360/p360_239.wav
+p360/p360_245.wav
+p360/p360_183.wav
+p360/p360_152.wav
+p360/p360_294.wav
+p360/p360_242.wav
+p360/p360_184.wav
+p360/p360_155.wav
+p360/p360_293.wav
+p360/p360_071.wav
+p360/p360_410.wav
+p360/p360_366.wav
+p360/p360_129.wav
+p360/p360_003.wav
+p360/p360_314.wav
+p360/p360_368.wav
+p360/p360_230.wav
+p360/p360_127.wav
+p360/p360_078.wav
+p360/p360_419.wav
+p360/p360_237.wav
+p360/p360_120.wav
+p360/p360_004.wav
+p360/p360_313.wav
+p360/p360_131.wav
+p360/p360_226.wav
+p360/p360_069.wav
+p360/p360_408.wav
+p360/p360_302.wav
+p360/p360_015.wav
+p360/p360_305.wav
+p360/p360_012.wav
+p360/p360_136.wav
+p360/p360_221.wav
+p360/p360_379.wav
+p360/p360_144.wav
+p360/p360_282.wav
+p360/p360_253.wav
+p360/p360_195.wav
+p360/p360_138.wav
+p360/p360_377.wav
+p360/p360_060.wav
+p360/p360_401.wav
+p360/p360_228.wav
+p360/p360_370.wav
+p360/p360_067.wav
+p360/p360_406.wav
+p360/p360_143.wav
+p360/p360_285.wav
+p360/p360_254.wav
+p360/p360_192.wav
+p360/p360_394.wav
+p360/p360_052.wav
+p360/p360_083.wav
+p360/p360_345.wav
+p360/p360_261.wav
+p360/p360_176.wav
+p360/p360_339.wav
+p360/p360_266.wav
+p360/p360_171.wav
+p360/p360_029.wav
+p360/p360_393.wav
+p360/p360_055.wav
+p360/p360_084.wav
+p360/p360_342.wav
+p360/p360_268.wav
+p360/p360_027.wav
+p360/p360_330.wav
+p360/p360_214.wav
+p360/p360_103.wav
+p360/p360_213.wav
+p360/p360_104.wav
+p360/p360_178.wav
+p360/p360_020.wav
+p360/p360_337.wav
+p374/p374_417.wav
+p374/p374_076.wav
+p374/p374_361.wav
+p374/p374_239.wav
+p374/p374_183.wav
+p374/p374_245.wav
+p374/p374_294.wav
+p374/p374_152.wav
+p374/p374_184.wav
+p374/p374_242.wav
+p374/p374_293.wav
+p374/p374_155.wav
+p374/p374_410.wav
+p374/p374_071.wav
+p374/p374_366.wav
+p374/p374_129.wav
+p374/p374_003.wav
+p374/p374_314.wav
+p374/p374_368.wav
+p374/p374_230.wav
+p374/p374_127.wav
+p374/p374_419.wav
+p374/p374_078.wav
+p374/p374_237.wav
+p374/p374_120.wav
+p374/p374_004.wav
+p374/p374_313.wav
+p374/p374_115.wav
+p374/p374_202.wav
+p374/p374_326.wav
+p374/p374_031.wav
+p374/p374_169.wav
+p374/p374_321.wav
+p374/p374_036.wav
+p374/p374_279.wav
+p374/p374_112.wav
+p374/p374_205.wav
+p374/p374_038.wav
+p374/p374_160.wav
+p374/p374_277.wav
+p374/p374_353.wav
+p374/p374_095.wav
+p374/p374_044.wav
+p374/p374_382.wav
+p374/p374_354.wav
+p374/p374_092.wav
+p374/p374_043.wav
+p374/p374_422.wav
+p374/p374_385.wav
+p374/p374_328.wav
+p374/p374_167.wav
+p374/p374_270.wav
+p374/p374_052.wav
+p374/p374_394.wav
+p374/p374_345.wav
+p374/p374_083.wav
+p374/p374_261.wav
+p374/p374_176.wav
+p374/p374_339.wav
+p374/p374_266.wav
+p374/p374_171.wav
+p374/p374_029.wav
+p374/p374_055.wav
+p374/p374_393.wav
+p374/p374_342.wav
+p374/p374_084.wav
+p374/p374_268.wav
+p374/p374_027.wav
+p374/p374_330.wav
+p374/p374_214.wav
+p374/p374_103.wav
+p374/p374_213.wav
+p374/p374_104.wav
+p374/p374_178.wav
+p374/p374_020.wav
+p374/p374_337.wav
+p374/p374_131.wav
+p374/p374_226.wav
+p374/p374_408.wav
+p374/p374_069.wav
+p374/p374_302.wav
+p374/p374_015.wav
+p374/p374_305.wav
+p374/p374_012.wav
+p374/p374_136.wav
+p374/p374_221.wav
+p374/p374_379.wav
+p374/p374_282.wav
+p374/p374_144.wav
+p374/p374_195.wav
+p374/p374_253.wav
+p374/p374_138.wav
+p374/p374_377.wav
+p374/p374_401.wav
+p374/p374_060.wav
+p374/p374_228.wav
+p374/p374_370.wav
+p374/p374_406.wav
+p374/p374_067.wav
+p374/p374_285.wav
+p374/p374_143.wav
+p374/p374_192.wav
+p374/p374_254.wav
+p374/p374_355.wav
+p374/p374_093.wav
+p374/p374_042.wav
+p374/p374_384.wav
+p374/p374_423.wav
+p374/p374_329.wav
+p374/p374_166.wav
+p374/p374_271.wav
+p374/p374_039.wav
+p374/p374_161.wav
+p374/p374_276.wav
+p374/p374_352.wav
+p374/p374_094.wav
+p374/p374_045.wav
+p374/p374_424.wav
+p374/p374_383.wav
+p374/p374_320.wav
+p374/p374_037.wav
+p374/p374_278.wav
+p374/p374_113.wav
+p374/p374_204.wav
+p374/p374_114.wav
+p374/p374_203.wav
+p374/p374_327.wav
+p374/p374_030.wav
+p374/p374_168.wav
+p374/p374_418.wav
+p374/p374_079.wav
+p374/p374_236.wav
+p374/p374_121.wav
+p374/p374_005.wav
+p374/p374_312.wav
+p374/p374_002.wav
+p374/p374_315.wav
+p374/p374_369.wav
+p374/p374_231.wav
+p374/p374_126.wav
+p374/p374_185.wav
+p374/p374_243.wav
+p374/p374_292.wav
+p374/p374_154.wav
+p374/p374_411.wav
+p374/p374_070.wav
+p374/p374_367.wav
+p374/p374_128.wav
+p374/p374_416.wav
+p374/p374_077.wav
+p374/p374_360.wav
+p374/p374_238.wav
+p374/p374_182.wav
+p374/p374_244.wav
+p374/p374_295.wav
+p374/p374_153.wav
+p374/p374_229.wav
+p374/p374_371.wav
+p374/p374_407.wav
+p374/p374_066.wav
+p374/p374_284.wav
+p374/p374_142.wav
+p374/p374_193.wav
+p374/p374_255.wav
+p374/p374_283.wav
+p374/p374_145.wav
+p374/p374_194.wav
+p374/p374_252.wav
+p374/p374_139.wav
+p374/p374_376.wav
+p374/p374_400.wav
+p374/p374_061.wav
+p374/p374_304.wav
+p374/p374_013.wav
+p374/p374_137.wav
+p374/p374_220.wav
+p374/p374_378.wav
+p374/p374_130.wav
+p374/p374_227.wav
+p374/p374_409.wav
+p374/p374_068.wav
+p374/p374_303.wav
+p374/p374_014.wav
+p374/p374_212.wav
+p374/p374_105.wav
+p374/p374_179.wav
+p374/p374_021.wav
+p374/p374_336.wav
+p374/p374_269.wav
+p374/p374_026.wav
+p374/p374_331.wav
+p374/p374_215.wav
+p374/p374_102.wav
+p374/p374_267.wav
+p374/p374_170.wav
+p374/p374_028.wav
+p374/p374_054.wav
+p374/p374_392.wav
+p374/p374_343.wav
+p374/p374_085.wav
+p374/p374_053.wav
+p374/p374_395.wav
+p374/p374_344.wav
+p374/p374_082.wav
+p374/p374_260.wav
+p374/p374_177.wav
+p374/p374_338.wav
+p374/p374_146.wav
+p374/p374_280.wav
+p374/p374_251.wav
+p374/p374_197.wav
+p374/p374_309.wav
+p374/p374_375.wav
+p374/p374_062.wav
+p374/p374_403.wav
+p374/p374_372.wav
+p374/p374_065.wav
+p374/p374_404.wav
+p374/p374_141.wav
+p374/p374_287.wav
+p374/p374_256.wav
+p374/p374_190.wav
+p374/p374_019.wav
+p374/p374_133.wav
+p374/p374_224.wav
+p374/p374_289.wav
+p374/p374_258.wav
+p374/p374_300.wav
+p374/p374_017.wav
+p374/p374_148.wav
+p374/p374_199.wav
+p374/p374_307.wav
+p374/p374_010.wav
+p374/p374_134.wav
+p374/p374_223.wav
+p374/p374_025.wav
+p374/p374_332.wav
+p374/p374_216.wav
+p374/p374_101.wav
+p374/p374_059.wav
+p374/p374_088.wav
+p374/p374_211.wav
+p374/p374_106.wav
+p374/p374_398.wav
+p374/p374_349.wav
+p374/p374_022.wav
+p374/p374_335.wav
+p374/p374_108.wav
+p374/p374_396.wav
+p374/p374_050.wav
+p374/p374_081.wav
+p374/p374_347.wav
+p374/p374_263.wav
+p374/p374_174.wav
+p374/p374_264.wav
+p374/p374_173.wav
+p374/p374_218.wav
+p374/p374_391.wav
+p374/p374_057.wav
+p374/p374_086.wav
+p374/p374_340.wav
+p374/p374_162.wav
+p374/p374_275.wav
+p374/p374_097.wav
+p374/p374_351.wav
+p374/p374_380.wav
+p374/p374_046.wav
+p374/p374_209.wav
+p374/p374_090.wav
+p374/p374_356.wav
+p374/p374_387.wav
+p374/p374_420.wav
+p374/p374_041.wav
+p374/p374_119.wav
+p374/p374_165.wav
+p374/p374_272.wav
+p374/p374_358.wav
+p374/p374_389.wav
+p374/p374_117.wav
+p374/p374_200.wav
+p374/p374_324.wav
+p374/p374_033.wav
+p374/p374_323.wav
+p374/p374_034.wav
+p374/p374_099.wav
+p374/p374_048.wav
+p374/p374_110.wav
+p374/p374_207.wav
+p374/p374_001.wav
+p374/p374_316.wav
+p374/p374_188.wav
+p374/p374_159.wav
+p374/p374_232.wav
+p374/p374_125.wav
+p374/p374_235.wav
+p374/p374_122.wav
+p374/p374_006.wav
+p374/p374_311.wav
+p374/p374_249.wav
+p374/p374_298.wav
+p374/p374_074.wav
+p374/p374_415.wav
+p374/p374_363.wav
+p374/p374_008.wav
+p374/p374_247.wav
+p374/p374_181.wav
+p374/p374_150.wav
+p374/p374_296.wav
+p374/p374_318.wav
+p374/p374_240.wav
+p374/p374_186.wav
+p374/p374_157.wav
+p374/p374_291.wav
+p374/p374_073.wav
+p374/p374_412.wav
+p374/p374_364.wav
+p374/p374_265.wav
+p374/p374_172.wav
+p374/p374_219.wav
+p374/p374_390.wav
+p374/p374_056.wav
+p374/p374_087.wav
+p374/p374_341.wav
+p374/p374_109.wav
+p374/p374_397.wav
+p374/p374_051.wav
+p374/p374_080.wav
+p374/p374_346.wav
+p374/p374_262.wav
+p374/p374_175.wav
+p374/p374_210.wav
+p374/p374_107.wav
+p374/p374_399.wav
+p374/p374_348.wav
+p374/p374_023.wav
+p374/p374_334.wav
+p374/p374_024.wav
+p374/p374_333.wav
+p374/p374_217.wav
+p374/p374_100.wav
+p374/p374_058.wav
+p374/p374_089.wav
+p374/p374_149.wav
+p374/p374_198.wav
+p374/p374_306.wav
+p374/p374_011.wav
+p374/p374_135.wav
+p374/p374_222.wav
+p374/p374_132.wav
+p374/p374_225.wav
+p374/p374_288.wav
+p374/p374_259.wav
+p374/p374_301.wav
+p374/p374_016.wav
+p374/p374_373.wav
+p374/p374_064.wav
+p374/p374_405.wav
+p374/p374_140.wav
+p374/p374_286.wav
+p374/p374_257.wav
+p374/p374_191.wav
+p374/p374_018.wav
+p374/p374_147.wav
+p374/p374_281.wav
+p374/p374_250.wav
+p374/p374_196.wav
+p374/p374_308.wav
+p374/p374_374.wav
+p374/p374_063.wav
+p374/p374_402.wav
+p374/p374_319.wav
+p374/p374_241.wav
+p374/p374_187.wav
+p374/p374_156.wav
+p374/p374_290.wav
+p374/p374_072.wav
+p374/p374_413.wav
+p374/p374_365.wav
+p374/p374_075.wav
+p374/p374_414.wav
+p374/p374_362.wav
+p374/p374_009.wav
+p374/p374_246.wav
+p374/p374_180.wav
+p374/p374_151.wav
+p374/p374_297.wav
+p374/p374_234.wav
+p374/p374_123.wav
+p374/p374_007.wav
+p374/p374_310.wav
+p374/p374_248.wav
+p374/p374_299.wav
+p374/p374_317.wav
+p374/p374_189.wav
+p374/p374_158.wav
+p374/p374_233.wav
+p374/p374_124.wav
+p374/p374_322.wav
+p374/p374_035.wav
+p374/p374_098.wav
+p374/p374_049.wav
+p374/p374_111.wav
+p374/p374_206.wav
+p374/p374_359.wav
+p374/p374_388.wav
+p374/p374_116.wav
+p374/p374_201.wav
+p374/p374_325.wav
+p374/p374_032.wav
+p374/p374_091.wav
+p374/p374_357.wav
+p374/p374_421.wav
+p374/p374_386.wav
+p374/p374_040.wav
+p374/p374_118.wav
+p374/p374_164.wav
+p374/p374_273.wav
+p374/p374_163.wav
+p374/p374_274.wav
+p374/p374_096.wav
+p374/p374_350.wav
+p374/p374_381.wav
+p374/p374_047.wav
+p374/p374_208.wav

dataset.py

@@ -0,0 +1,227 @@
+import glob
+import os
+import random
+
+import librosa
+import numpy as np
+import soundfile as sf
+import torch
+from numpy.random import default_rng
+from pydtmc import MarkovChain
+from sklearn.model_selection import train_test_split
+from torch.utils.data import Dataset
+
+from config import CONFIG
+
+np.random.seed(0)
+rng = default_rng()
+
+
+def load_audio(
+        path,
+        sample_rate: int = 16000,
+        chunk_len=None,
+):
+    with sf.SoundFile(path) as f:
+        sr = f.samplerate
+        audio_len = f.frames
+
+        if chunk_len is not None and chunk_len < audio_len:
+            start_index = torch.randint(0, audio_len - chunk_len, (1,))[0]
+
+            frames = f._prepare_read(start_index, start_index + chunk_len, -1)
+            audio = f.read(frames, always_2d=True, dtype="float32")
+
+        else:
+            audio = f.read(always_2d=True, dtype="float32")
+
+    if sr != sample_rate:
+        audio = librosa.resample(np.squeeze(audio), sr, sample_rate)[:, np.newaxis]
+
+    return audio.T
+
+
+def pad(sig, length):
+    if sig.shape[1] < length:
+        pad_len = length - sig.shape[1]
+        sig = torch.hstack((sig, torch.zeros((sig.shape[0], pad_len))))
+
+    else:
+        start = random.randint(0, sig.shape[1] - length)
+        sig = sig[:, start:start + length]
+    return sig
+
+
+class MaskGenerator:
+    def __init__(self, is_train=True, probs=((0.9, 0.1), (0.5, 0.1), (0.5, 0.5))):
+        '''
+            is_train: if True, mask generator for training otherwise for evaluation
+            probs: a list of transition probability (p_N, p_L) for Markov Chain. Only allow 1 tuple if 'is_train=False'
+        '''
+        self.is_train = is_train
+        self.probs = probs
+        self.mcs = []
+        if self.is_train:
+            for prob in probs:
+                self.mcs.append(MarkovChain([[prob[0], 1 - prob[0]], [1 - prob[1], prob[1]]], ['1', '0']))
+        else:
+            assert len(probs) == 1
+            prob = self.probs[0]
+            self.mcs.append(MarkovChain([[probs[0], 1 - prob[0]], [1 - prob[1], prob[1]]], ['1', '0']))
+
+    def gen_mask(self, length, seed=0):
+        if self.is_train:
+            mc = random.choice(self.mcs)
+        else:
+            mc = self.mcs[0]
+        mask = mc.walk(length - 1, seed=seed)
+        mask = np.array(list(map(int, mask)))
+        return mask
+
+
+class TestLoader(Dataset):
+    def __init__(self):
+        dataset_name = CONFIG.DATA.dataset
+        self.mask = CONFIG.DATA.EVAL.masking
+
+        self.target_root = CONFIG.DATA.data_dir[dataset_name]['root']
+        txt_list = CONFIG.DATA.data_dir[dataset_name]['test']
+        self.data_list = self.load_txt(txt_list)
+        if self.mask == 'real':
+            trace_txt = glob.glob(os.path.join(CONFIG.DATA.EVAL.trace_path, '*.txt'))
+            trace_txt.sort()
+            self.trace_list = [1 - np.array(list(map(int, open(txt, 'r').read().strip('\n').split('\n')))) for txt in
+                               trace_txt]
+        else:
+            self.mask_generator = MaskGenerator(is_train=False, probs=CONFIG.DATA.EVAL.transition_probs)
+
+        self.sr = CONFIG.DATA.sr
+        self.stride = CONFIG.DATA.stride
+        self.window_size = CONFIG.DATA.window_size
+        self.audio_chunk_len = CONFIG.DATA.audio_chunk_len
+        self.p_size = CONFIG.DATA.EVAL.packet_size  # 20ms
+        self.hann = torch.sqrt(torch.hann_window(self.window_size))
+
+    def __len__(self):
+        return len(self.data_list)
+
+    def load_txt(self, txt_list):
+        target = []
+        with open(txt_list) as f:
+            for line in f:
+                target.append(os.path.join(self.target_root, line.strip('\n')))
+        target = list(set(target))
+        target.sort()
+        return target
+
+    def __getitem__(self, index):
+        target = load_audio(self.data_list[index], sample_rate=self.sr)
+        target = target[:, :(target.shape[1] // self.p_size) * self.p_size]
+
+        sig = np.reshape(target, (-1, self.p_size)).copy()
+        if self.mask == 'real':
+            mask = self.trace_list[index % len(self.trace_list)]
+            mask = np.repeat(mask, np.ceil(len(sig) / len(mask)), 0)[:len(sig)][:, np.newaxis]
+        else:
+            mask = self.mask_generator.gen_mask(len(sig), seed=index)[:, np.newaxis]
+        sig *= mask
+        sig = torch.tensor(sig).reshape(-1)
+
+        target = torch.tensor(target).squeeze(0)
+
+        sig_wav = sig.clone()
+        target_wav = target.clone()
+
+        target = torch.stft(target, self.window_size, self.stride, window=self.hann,
+                            return_complex=False).permute(2, 0, 1)
+        sig = torch.stft(sig, self.window_size, self.stride, window=self.hann, return_complex=False).permute(2, 0, 1)
+        return sig.float(), target.float(), sig_wav, target_wav
+
+
+class BlindTestLoader(Dataset):
+    def __init__(self, test_dir):
+        self.data_list = glob.glob(os.path.join(test_dir, '*.wav'))[:10]
+        self.sr = CONFIG.DATA.sr
+        self.stride = CONFIG.DATA.stride
+        self.chunk_len = CONFIG.DATA.window_size
+        self.hann = torch.sqrt(torch.hann_window(self.chunk_len))
+
+    def __len__(self):
+        return len(self.data_list)
+
+    def __getitem__(self, index):
+        sig = load_audio(self.data_list[index], sample_rate=self.sr)
+        sig = torch.from_numpy(sig).squeeze(0)
+        sig = torch.stft(sig, self.chunk_len, self.stride, window=self.hann, return_complex=False).permute(2, 0, 1)
+        return sig.float()
+
+
+class TrainDataset(Dataset):
+
+    def __init__(self, mode='train'):
+        dataset_name = CONFIG.DATA.dataset
+        self.target_root = CONFIG.DATA.data_dir[dataset_name]['root']
+
+        txt_list = CONFIG.DATA.data_dir[dataset_name]['train']
+        self.data_list = self.load_txt(txt_list)
+
+        if mode == 'train':
+            self.data_list, _ = train_test_split(self.data_list, test_size=CONFIG.TRAIN.val_split, random_state=0)
+
+        elif mode == 'val':
+            _, self.data_list = train_test_split(self.data_list, test_size=CONFIG.TRAIN.val_split, random_state=0)
+
+        self.p_sizes = CONFIG.DATA.TRAIN.packet_sizes
+        self.mode = mode
+        self.sr = CONFIG.DATA.sr
+        self.window = CONFIG.DATA.audio_chunk_len
+        self.stride = CONFIG.DATA.stride
+        self.chunk_len = CONFIG.DATA.window_size
+        self.hann = torch.sqrt(torch.hann_window(self.chunk_len))
+        self.mask_generator = MaskGenerator(is_train=True, probs=CONFIG.DATA.TRAIN.transition_probs)
+
+    def __len__(self):
+        return len(self.data_list)
+
+    def load_txt(self, txt_list):
+        target = []
+        with open(txt_list) as f:
+            for line in f:
+                target.append(os.path.join(self.target_root, line.strip('\n')))
+        target = list(set(target))
+        target.sort()
+        return target
+
+    def fetch_audio(self, index):
+        sig = load_audio(self.data_list[index], sample_rate=self.sr, chunk_len=self.window)
+        while sig.shape[1] < self.window:
+            idx = torch.randint(0, len(self.data_list), (1,))[0]
+            pad_len = self.window - sig.shape[1]
+            if pad_len < 0.02 * self.sr:
+                padding = np.zeros((1, pad_len), dtype=np.float)
+            else:
+                padding = load_audio(self.data_list[idx], sample_rate=self.sr, chunk_len=pad_len)
+            sig = np.hstack((sig, padding))
+        return sig
+
+    def __getitem__(self, index):
+        sig = self.fetch_audio(index)
+
+        sig = sig.reshape(-1).astype(np.float32)
+
+        sig = sig.reshape((1, -1))
+        target = torch.tensor(sig.copy())
+        p_size = random.choice(self.p_sizes)
+
+        sig = np.reshape(sig, (-1, p_size))
+        mask = self.mask_generator.gen_mask(len(sig), seed=index)[:, np.newaxis]
+        sig *= mask
+        sig = torch.tensor(sig.copy())
+
+        sig = sig.reshape(1, -1)
+
+        target = torch.stft(target.squeeze(0), self.chunk_len, self.stride, window=self.hann,
+                            return_complex=False).permute(2, 0, 1).float()
+        sig = torch.stft(sig.squeeze(0), self.chunk_len, self.stride, window=self.hann, return_complex=False)
+        sig = sig.permute(2, 0, 1).float()
+        return sig, target

index.html

@@ -0,0 +1,139 @@
+<!DOCTYPE html>
+<html>
+<head>
+    <link href="css/styles.css" rel="stylesheet">
+
+    <title>Full-band Recurrent Network</title>
+</head>
+<body>
+<nav>
+    <ul>
+        <!--        <li><a href="/">Home</a></li> -->
+        <li><a href="https://github.com/Crystalsound/FRN/">Github</a></li>
+        <li><a href="https://arxiv.org/abs/2211.04071">Arxiv</a></li>
+        <li><a href="https://www.namitech.io/">Website</a></li>
+    </ul>
+</nav>
+<div class=”container”>
+    <div class=”blurb”>
+        <h1>Audio samples</h1>
+        <p><b>Improving performance of real-time full-band blind packet-loss concealment with predictive network</b></a>
+        </p>
+        <p><i>Viet-Anh Nguyen<sup>1</sup>, Anh H. T. Nguyen<sup>1</sup>, and Andy W. H. Khong<sup>2</sup></i>
+            <br><sup>1</sup>Crystalsound Team, NamiTech JSC, Ho Chi Minh City, Vietnam
+            <br><sup>2</sup>Nanyang Technological University, Singapore
+            <br><TT>{vietanh.nguyen, anh.nguyen}@namitech.io, andykhong@ntu.edu.sg
+    </div>
+</div>
+<h3> Audio samples of our full-band recurrent network (FRN) versus TFGAN and tPLCNet for blind packet loss concealment
+    (PLC)</h3>
+Audio files are at 48 kHz sampling rate with packet size of 20 ms. Our FRN is a causal and blind PLC model while TFGAN
+is non-causal and tPLC is an informed PLC model.
+<br> </br>
+<table>
+    <thead>
+    <tr>
+        <th align="middle">Clean target</th>
+        <th align="middle">Lossy input</th>
+        <th align="middle">TFGAN</th>
+        <th align="middle">tPLCNet</th>
+        <th align="middle">FRN (Ours)</th>
+    </tr>
+    </thead>
+
+    <tbody>
+    <tr>
+        <td>
+            <audio controls style="width: 250px; height: 50px">
+                <source src="audio_samples/sample_1/clean.wav" type="audio/wav">
+            </audio>
+        </td>
+        <td>
+            <audio controls style="width: 250px; height: 50px">
+                <source src="audio_samples/sample_1/lossy.wav" type="audio/wav">
+            </audio>
+        </td>
+        <td>
+            <audio controls style="width: 250px; height: 50px">
+                <source src="audio_samples/sample_1/TFGAN_enhanced.wav" type="audio/wav">
+            </audio>
+        </td>
+        <td>
+            <audio controls style="width: 250px; height: 50px">
+                <source src="audio_samples/sample_1/tPLC_enhanced.wav" type="audio/wav">
+            </audio>
+        </td>
+        <td>
+            <audio controls style="width: 250px; height: 50px">
+                <source src="audio_samples/sample_1/FRN_enhanced.wav" type="audio/wav">
+            </audio>
+        </td>
+    </tr>
+
+    <tr>
+        <td>
+            <audio controls style="width: 250px; height: 50px">
+                <source src="audio_samples/sample_2/clean.wav" type="audio/wav">
+            </audio>
+        </td>
+        <td>
+            <audio controls style="width: 250px; height: 50px">
+                <source src="audio_samples/sample_2/lossy.wav" type="audio/wav">
+            </audio>
+        </td>
+        <td>
+            <audio controls style="width: 250px; height: 50px">
+                <source src="audio_samples/sample_2/TFGAN_enhanced.wav" type="audio/wav">
+            </audio>
+        </td>
+        <td>
+            <audio controls style="width: 250px; height: 50px">
+                <source src="audio_samples/sample_2/tPLC_enhanced.wav" type="audio/wav">
+            </audio>
+        </td>
+        <td>
+            <audio controls style="width: 250px; height: 50px">
+                <source src="audio_samples/sample_2/FRN_enhanced.wav" type="audio/wav">
+            </audio>
+        </td>
+    </tr>
+
+    <tr>
+        <td>
+            <audio controls style="width: 250px; height: 50px">
+                <source src="audio_samples/sample_3/clean.wav" type="audio/wav">
+            </audio>
+        </td>
+        <td>
+            <audio controls style="width: 250px; height: 50px">
+                <source src="audio_samples/sample_3/lossy.wav" type="audio/wav">
+            </audio>
+        </td>
+        <td>
+            <audio controls style="width: 250px; height: 50px">
+                <source src="audio_samples/sample_3/TFGAN_enhanced.wav" type="audio/wav">
+            </audio>
+        </td>
+        <td>
+            <audio controls style="width: 250px; height: 50px">
+                <source src="audio_samples/sample_3/tPLC_enhanced.wav" type="audio/wav">
+            </audio>
+        </td>
+        <td>
+            <audio controls style="width: 250px; height: 50px">
+                <source src="audio_samples/sample_3/FRN_enhanced.wav" type="audio/wav">
+            </audio>
+        </td>
+    </tr>
+
+
+    </tbody>
+</table>
+<!--    <footer>
+   <ul>
+     <li><a href=”mailto:YOUREMAIL”>YOUREMAIL</a></li>
+   </ul>
+   </footer> -->
+
+</body>
+</html>

inference_onnx.py

@@ -0,0 +1,63 @@
+import argparse
+import glob
+import os
+
+import librosa
+import numpy as np
+import onnx
+import onnxruntime
+import soundfile as sf
+import torch
+import tqdm
+
+from config import CONFIG
+
+parser = argparse.ArgumentParser()
+
+parser.add_argument('--onnx_path', default=None,
+                    help='path to onnx')
+args = parser.parse_args()
+
+if __name__ == '__main__':
+    path = args.onnx_path
+    window = CONFIG.DATA.window_size
+    stride = CONFIG.DATA.stride
+    onnx_model = onnx.load(path)
+    options = onnxruntime.SessionOptions()
+    options.intra_op_num_threads = 8
+    options.graph_optimization_level = onnxruntime.GraphOptimizationLevel.ORT_ENABLE_ALL
+    session = onnxruntime.InferenceSession(path, options)
+    input_names = [x.name for x in session.get_inputs()]
+    output_names = [x.name for x in session.get_outputs()]
+    print(input_names)
+    print(output_names)
+
+    audio_files = glob.glob(os.path.join(CONFIG.TEST.in_dir, '*.wav'))
+    hann = torch.sqrt(torch.hann_window(window))
+    os.makedirs(CONFIG.TEST.out_dir, exist_ok=True)
+    for file in tqdm.tqdm(audio_files, total=len(audio_files)):
+        sig, _ = librosa.load(file, sr=48000)
+        sig = torch.tensor(sig)
+        re_im = torch.stft(sig, window, stride, window=hann, return_complex=False).permute(2, 0, 1).unsqueeze(
+            0).numpy().astype(np.float32)
+
+        inputs = {input_names[i]: np.zeros([d.dim_value for d in _input.type.tensor_type.shape.dim],
+                                           dtype=np.float32)
+                  for i, _input in enumerate(onnx_model.graph.input)
+                  }
+
+        output_audio = []
+        for t in range(re_im.shape[-1]):
+            ri_t = re_im[:, :, :, t:t + 1]
+            out, prev_mag, predictor_state, mlp_state = session.run(output_names, inputs)
+            inputs[input_names[1]] = prev_mag
+            inputs[input_names[2]] = predictor_state
+            inputs[input_names[3]] = mlp_state
+            output_audio.append(out)
+
+        output_audio = torch.tensor(np.concatenate(output_audio, 0))
+        output_audio = output_audio.permute(1, 0, 2).contiguous()
+        output_audio = torch.view_as_complex(output_audio)
+        output_audio = torch.istft(output_audio, window, stride, window=hann)
+        sf.write(os.path.join(CONFIG.TEST.out_dir, os.path.basename(file)), output_audio, samplerate=48000,
+                 subtype='PCM_16')

lightning_logs/predictor/checkpoints/predictor.ckpt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1f3679c9431666575eb7899e556d040073aa74956c48f122b16b30b9efa2e93b
+size 14985163

lightning_logs/predictor/hparams.yaml

@@ -0,0 +1,6 @@
+batch_size: 90
+input: mag
+lstm_dim: 512
+lstm_layers: 1
+output: mag
+window_size: 960

lightning_logs/version_0/checkpoints/frn-epoch=65-val_loss=0.2290.ckpt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4061bb0f6e669315e00878009440dab749f60f823d5bf863bfa4b8172d96d073
+size 109184745

lightning_logs/version_0/checkpoints/frn.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e373d149496278e2368fb371802878f22ed650ce65f4bf13f73446c9df6ac56e
+size 36527867

lightning_logs/version_0/hparams.yaml

@@ -0,0 +1,6 @@
+batch_size: 90
+cnn_dim: 64
+cnn_layers: 5
+lstm_dim: 512
+lstm_layers: 1
+window_size: 960

loss.py

@@ -0,0 +1,145 @@
+import librosa
+import pytorch_lightning as pl
+import torch
+from auraloss.freq import STFTLoss, MultiResolutionSTFTLoss, apply_reduction, SpectralConvergenceLoss, STFTMagnitudeLoss
+
+from config import CONFIG
+
+
+class STFTLossDDP(STFTLoss):
+    def __init__(self,
+                 fft_size=1024,
+                 hop_size=256,
+                 win_length=1024,
+                 window="hann_window",
+                 w_sc=1.0,
+                 w_log_mag=1.0,
+                 w_lin_mag=0.0,
+                 w_phs=0.0,
+                 sample_rate=None,
+                 scale=None,
+                 n_bins=None,
+                 scale_invariance=False,
+                 eps=1e-8,
+                 output="loss",
+                 reduction="mean",
+                 device=None):
+        super(STFTLoss, self).__init__()
+        self.fft_size = fft_size
+        self.hop_size = hop_size
+        self.win_length = win_length
+        self.window = getattr(torch, window)(win_length)
+        self.w_sc = w_sc
+        self.w_log_mag = w_log_mag
+        self.w_lin_mag = w_lin_mag
+        self.w_phs = w_phs
+        self.sample_rate = sample_rate
+        self.scale = scale
+        self.n_bins = n_bins
+        self.scale_invariance = scale_invariance
+        self.eps = eps
+        self.output = output
+        self.reduction = reduction
+        self.device = device
+
+        self.spectralconv = SpectralConvergenceLoss()
+        self.logstft = STFTMagnitudeLoss(log=True, reduction=reduction)
+        self.linstft = STFTMagnitudeLoss(log=False, reduction=reduction)
+
+        # setup mel filterbank
+        if self.scale == "mel":
+            assert (sample_rate is not None)  # Must set sample rate to use mel scale
+            assert (n_bins <= fft_size)  # Must be more FFT bins than Mel bins
+            fb = librosa.filters.mel(sample_rate, fft_size, n_mels=n_bins)
+            self.fb = torch.tensor(fb).unsqueeze(0)
+        elif self.scale == "chroma":
+            assert (sample_rate is not None)  # Must set sample rate to use chroma scale
+            assert (n_bins <= fft_size)  # Must be more FFT bins than chroma bins
+            fb = librosa.filters.chroma(sample_rate, fft_size, n_chroma=n_bins)
+            self.fb = torch.tensor(fb).unsqueeze(0)
+
+        if scale is not None and device is not None:
+            self.fb = self.fb.to(self.device)  # move filterbank to device
+
+    def compressed_loss(self, x, y, alpha=None):
+        self.window = self.window.to(x.device)
+        x_mag, x_phs = self.stft(x.view(-1, x.size(-1)))
+        y_mag, y_phs = self.stft(y.view(-1, y.size(-1)))
+
+        if alpha is not None:
+            x_mag = x_mag ** alpha
+            y_mag = y_mag ** alpha
+
+        # apply relevant transforms
+        if self.scale is not None:
+            x_mag = torch.matmul(self.fb.to(x_mag.device), x_mag)
+            y_mag = torch.matmul(self.fb.to(y_mag.device), y_mag)
+
+        # normalize scales
+        if self.scale_invariance:
+            alpha = (x_mag * y_mag).sum([-2, -1]) / ((y_mag ** 2).sum([-2, -1]))
+            y_mag = y_mag * alpha.unsqueeze(-1)
+
+        # compute loss terms
+        sc_loss = self.spectralconv(x_mag, y_mag) if self.w_sc else 0.0
+        mag_loss = self.logstft(x_mag, y_mag) if self.w_log_mag else 0.0
+        lin_loss = self.linstft(x_mag, y_mag) if self.w_lin_mag else 0.0
+
+        # combine loss terms
+        loss = (self.w_sc * sc_loss) + (self.w_log_mag * mag_loss) + (self.w_lin_mag * lin_loss)
+        loss = apply_reduction(loss, reduction=self.reduction)
+        return loss
+
+    def forward(self, x, y):
+        return self.compressed_loss(x, y, 0.3)
+
+
+class MRSTFTLossDDP(MultiResolutionSTFTLoss):
+    def __init__(self,
+                 fft_sizes=(1024, 2048, 512),
+                 hop_sizes=(120, 240, 50),
+                 win_lengths=(600, 1200, 240),
+                 window="hann_window",
+                 w_sc=1.0,
+                 w_log_mag=1.0,
+                 w_lin_mag=0.0,
+                 w_phs=0.0,
+                 sample_rate=None,
+                 scale=None,
+                 n_bins=None,
+                 scale_invariance=False,
+                 **kwargs):
+        super(MultiResolutionSTFTLoss, self).__init__()
+        assert len(fft_sizes) == len(hop_sizes) == len(win_lengths)  # must define all
+        self.stft_losses = torch.nn.ModuleList()
+        for fs, ss, wl in zip(fft_sizes, hop_sizes, win_lengths):
+            self.stft_losses += [STFTLossDDP(fs,
+                                             ss,
+                                             wl,
+                                             window,
+                                             w_sc,
+                                             w_log_mag,
+                                             w_lin_mag,
+                                             w_phs,
+                                             sample_rate,
+                                             scale,
+                                             n_bins,
+                                             scale_invariance,
+                                             **kwargs)]
+
+
+class Loss(pl.LightningModule):
+    def __init__(self):
+        super(Loss, self).__init__()
+        self.stft_loss = MRSTFTLossDDP(sample_rate=CONFIG.DATA.sr, device="cpu", w_log_mag=0.0, w_lin_mag=1.0)
+        self.window = torch.sqrt(torch.hann_window(CONFIG.DATA.window_size))
+
+    def forward(self, x, y):
+        x = x.permute(0, 2, 3, 1)
+        y = y.permute(0, 2, 3, 1)
+        wave_x = torch.istft(torch.view_as_complex(x.contiguous()), CONFIG.DATA.window_size, CONFIG.DATA.stride,
+                             window=self.window.to(x.device))
+        wave_y = torch.istft(torch.view_as_complex(y.contiguous()), CONFIG.DATA.window_size, CONFIG.DATA.stride,
+                             window=self.window.to(y.device))
+        loss = self.stft_loss(wave_x, wave_y)
+        return loss

main.py

@@ -0,0 +1,132 @@
+import argparse
+import os
+
+import pytorch_lightning as pl
+import soundfile as sf
+import torch
+from pytorch_lightning.callbacks import ModelCheckpoint
+from pytorch_lightning.utilities.model_summary import summarize
+from torch.utils.data import DataLoader
+
+from config import CONFIG
+from dataset import TrainDataset, TestLoader, BlindTestLoader
+from models.frn import PLCModel, OnnxWrapper
+from utils.tblogger import TensorBoardLoggerExpanded
+from utils.utils import mkdir_p
+
+parser = argparse.ArgumentParser()
+
+parser.add_argument('--version', default=None,
+                    help='version to resume')
+parser.add_argument('--mode', default='train',
+                    help='training or testing mode')
+
+args = parser.parse_args()
+os.environ["CUDA_VISIBLE_DEVICES"] = str(CONFIG.gpus)
+assert args.mode in ['train', 'eval', 'test', 'onnx'], "--mode should be 'train', 'eval', 'test' or 'onnx'"
+
+
+def resume(train_dataset, val_dataset, version):
+    print("Version", version)
+    model_path = os.path.join(CONFIG.LOG.log_dir, 'version_{}/checkpoints/'.format(str(version)))
+    config_path = os.path.join(CONFIG.LOG.log_dir, 'version_{}/'.format(str(version)) + 'hparams.yaml')
+    model_name = [x for x in os.listdir(model_path) if x.endswith(".ckpt")][0]
+    ckpt_path = model_path + model_name
+    checkpoint = PLCModel.load_from_checkpoint(ckpt_path,
+                                               strict=True,
+                                               hparams_file=config_path,
+                                               train_dataset=train_dataset,
+                                               val_dataset=val_dataset,
+                                               window_size=CONFIG.DATA.window_size)
+
+    return checkpoint
+
+
+def train():
+    train_dataset = TrainDataset('train')
+    val_dataset = TrainDataset('val')
+    checkpoint_callback = ModelCheckpoint(monitor='val_loss', mode='min', verbose=True,
+                                          filename='frn-{epoch:02d}-{val_loss:.4f}', save_weights_only=False)
+    gpus = CONFIG.gpus.split(',')
+    logger = TensorBoardLoggerExpanded(CONFIG.DATA.sr)
+    if args.version is not None:
+        model = resume(train_dataset, val_dataset, args.version)
+    else:
+        model = PLCModel(train_dataset,
+                         val_dataset,
+                         window_size=CONFIG.DATA.window_size,
+                         enc_layers=CONFIG.MODEL.enc_layers,
+                         enc_in_dim=CONFIG.MODEL.enc_in_dim,
+                         enc_dim=CONFIG.MODEL.enc_dim,
+                         pred_dim=CONFIG.MODEL.pred_dim,
+                         pred_layers=CONFIG.MODEL.pred_layers)
+
+    trainer = pl.Trainer(logger=logger,
+                         gradient_clip_val=CONFIG.TRAIN.clipping_val,
+                         gpus=len(gpus),
+                         max_epochs=CONFIG.TRAIN.epochs,
+                         accelerator="ddp" if len(gpus) > 1 else None,
+                         stochastic_weight_avg=True,
+                         callbacks=[checkpoint_callback]
+                         )
+
+    print(model.hparams)
+    print(
+        'Dataset: {}, Train files: {}, Val files {}'.format(CONFIG.DATA.dataset, len(train_dataset), len(val_dataset)))
+    trainer.fit(model)
+
+
+def to_onnx(model, onnx_path):
+    model.eval()
+
+    model = OnnxWrapper(model)
+
+    torch.onnx.export(model,
+                      model.sample,
+                      onnx_path,
+                      export_params=True,
+                      opset_version=12,
+                      input_names=model.input_names,
+                      output_names=model.output_names,
+                      do_constant_folding=True,
+                      verbose=False)
+
+
+if __name__ == '__main__':
+
+    if args.mode == 'train':
+        train()
+    else:
+        model = resume(None, None, args.version)
+        print(model.hparams)
+        print(summarize(model))
+
+        model.eval()
+        model.freeze()
+        if args.mode == 'eval':
+            model.cuda(device=0)
+            trainer = pl.Trainer(gpus=1)
+            testset = TestLoader()
+            test_loader = DataLoader(testset, batch_size=1, num_workers=4)
+            trainer.test(model, test_loader)
+            print('Version', args.version)
+            masking = CONFIG.DATA.EVAL.masking
+            prob = CONFIG.DATA.EVAL.transition_probs[0]
+            loss_percent = (1 - prob[0]) / (2 - prob[0][prob[1]]) * 100
+            print('Evaluate with real trace' if masking == 'real' else
+                  'Evaluate with generated trace {}% packet loss'.format(str(prob)))
+        elif args.mode == 'test':
+            model.cuda(device=0)
+            testset = BlindTestLoader(test_dir=CONFIG.TEST.in_dir)
+            test_loader = DataLoader(testset, batch_size=1, num_workers=4)
+            trainer = pl.Trainer(gpus=1)
+            preds = trainer.predict(model, test_loader, return_predictions=True)
+            mkdir_p(CONFIG.TEST.out_dir)
+            for idx, path in enumerate(test_loader.dataset.data_list):
+                out_path = os.path.join(CONFIG.TEST.out_dir, os.path.basename(path))
+                sf.write(out_path, preds[idx], samplerate=CONFIG.DATA.sr, subtype='PCM_16')
+
+        else:
+            onnx_path = 'lightning_logs/version_{}/checkpoints/frn.onnx'.format(str(args.version))
+            to_onnx(model, onnx_path)
+            print('ONNX model saved to', onnx_path)

models/blocks.py

@@ -0,0 +1,142 @@
+import librosa
+import pytorch_lightning as pl
+import torch
+from einops.layers.torch import Rearrange
+from torch import nn
+
+
+class Aff(nn.Module):
+    def __init__(self, dim):
+        super().__init__()
+
+        self.alpha = nn.Parameter(torch.ones([1, 1, dim]))
+        self.beta = nn.Parameter(torch.zeros([1, 1, dim]))
+
+    def forward(self, x):
+        x = x * self.alpha + self.beta
+        return x
+
+
+class FeedForward(nn.Module):
+    def __init__(self, dim, hidden_dim, dropout=0.):
+        super().__init__()
+        self.net = nn.Sequential(
+            nn.Linear(dim, hidden_dim),
+            nn.GELU(),
+            nn.Dropout(dropout),
+            nn.Linear(hidden_dim, dim),
+            nn.Dropout(dropout)
+        )
+
+    def forward(self, x):
+        return self.net(x)
+
+
+class MLPBlock(nn.Module):
+
+    def __init__(self, dim, mlp_dim, dropout=0., init_values=1e-4):
+        super().__init__()
+
+        self.pre_affine = Aff(dim)
+        self.inter = nn.LSTM(input_size=dim, hidden_size=dim, num_layers=1,
+                             bidirectional=False, batch_first=True)
+        self.ff = nn.Sequential(
+            FeedForward(dim, mlp_dim, dropout),
+        )
+        self.post_affine = Aff(dim)
+        self.gamma_1 = nn.Parameter(init_values * torch.ones(dim), requires_grad=True)
+        self.gamma_2 = nn.Parameter(init_values * torch.ones(dim), requires_grad=True)
+
+    def forward(self, x, state=None):
+        x = self.pre_affine(x)
+        if state is None:
+            inter, _ = self.inter(x)
+        else:
+            inter, state = self.inter(x, (state[0], state[1]))
+        x = x + self.gamma_1 * inter
+        x = self.post_affine(x)
+        x = x + self.gamma_2 * self.ff(x)
+        if state is None:
+            return x
+        state = torch.stack(state, 0)
+        return x, state
+
+
+class Encoder(nn.Module):
+
+    def __init__(self, in_dim, dim, depth, mlp_dim):
+        super().__init__()
+        self.in_dim = in_dim
+        self.dim = dim
+        self.depth = depth
+        self.mlp_dim = mlp_dim
+        self.to_patch_embedding = nn.Sequential(
+            Rearrange('b c f t -> b t (c f)'),
+            nn.Linear(in_dim, dim),
+            nn.GELU()
+        )
+
+        self.mlp_blocks = nn.ModuleList([])
+
+        for _ in range(depth):
+            self.mlp_blocks.append(MLPBlock(self.dim, mlp_dim, dropout=0.15))
+
+        self.affine = nn.Sequential(
+            Aff(self.dim),
+            nn.Linear(dim, in_dim),
+            Rearrange('b t (c f) -> b c f t', c=2),
+        )
+
+    def forward(self, x_in, states=None):
+        x = self.to_patch_embedding(x_in)
+        if states is not None:
+            out_states = []
+        for i, mlp_block in enumerate(self.mlp_blocks):
+            if states is None:
+                x = mlp_block(x)
+            else:
+                x, state = mlp_block(x, states[i])
+                out_states.append(state)
+        x = self.affine(x)
+        x = x + x_in
+        if states is None:
+            return x
+        else:
+            return x, torch.stack(out_states, 0)
+
+
+class Predictor(pl.LightningModule):  # mel
+    def __init__(self, window_size=1536, sr=48000, lstm_dim=256, lstm_layers=3, n_mels=64):
+        super(Predictor, self).__init__()
+        self.window_size = window_size
+        self.hop_size = window_size // 2
+        self.lstm_dim = lstm_dim
+        self.n_mels = n_mels
+        self.lstm_layers = lstm_layers
+
+        fb = librosa.filters.mel(sr=sr, n_fft=self.window_size, n_mels=self.n_mels)[:, 1:]
+        self.fb = torch.from_numpy(fb).unsqueeze(0).unsqueeze(0)
+        self.lstm = nn.LSTM(input_size=self.n_mels, hidden_size=self.lstm_dim, bidirectional=False,
+                            num_layers=self.lstm_layers)
+        self.expand_dim = nn.Linear(self.lstm_dim, self.n_mels)
+        self.inv_mel = nn.Linear(self.n_mels, self.hop_size)
+
+    def forward(self, x, state=None):  # B, 2, F, T
+
+        self.fb = self.fb.to(x.device)
+        x = torch.log(torch.matmul(self.fb, x) + 1e-8)
+        B, C, F, T = x.shape
+        x = x.reshape(B, F * C, T)
+        x = x.permute(0, 2, 1)
+        if state is None:
+            x, _ = self.lstm(x)
+        else:
+            x, state = self.lstm(x, (state[0], state[1]))
+        x = self.expand_dim(x)
+        x = torch.abs(self.inv_mel(torch.exp(x)))
+        x = x.permute(0, 2, 1)
+        x = x.reshape(B, C, -1, T)
+        if state is None:
+            return x
+        else:
+            return x, torch.stack(state, 0)

models/frn.py

@@ -0,0 +1,220 @@
+import os
+
+import librosa
+import pytorch_lightning as pl
+import soundfile as sf
+import torch
+from torch import nn
+from torch.utils.data import DataLoader
+from torchmetrics.audio.pesq import PerceptualEvaluationSpeechQuality as PESQ
+from torchmetrics.audio.stoi import ShortTimeObjectiveIntelligibility as STOI
+
+from PLCMOS.plc_mos import PLCMOSEstimator
+from config import CONFIG
+from loss import Loss
+from models.blocks import Encoder, Predictor
+from utils.utils import visualize, LSD
+
+plcmos = PLCMOSEstimator()
+
+
+class PLCModel(pl.LightningModule):
+    def __init__(self, train_dataset=None, val_dataset=None, window_size=960, enc_layers=4, enc_in_dim=384, enc_dim=768,
+                 pred_dim=512, pred_layers=1, pred_ckpt_path='lightning_logs/predictor/checkpoints/predictor.ckpt'):
+        super(PLCModel, self).__init__()
+        self.window_size = window_size
+        self.hop_size = window_size // 2
+        self.learning_rate = CONFIG.TRAIN.lr
+        self.hparams.batch_size = CONFIG.TRAIN.batch_size
+
+        self.enc_layers = enc_layers
+        self.enc_in_dim = enc_in_dim
+        self.enc_dim = enc_dim
+        self.pred_dim = pred_dim
+        self.pred_layers = pred_layers
+        self.train_dataset = train_dataset
+        self.val_dataset = val_dataset
+        self.stoi = STOI(48000)
+        self.pesq = PESQ(16000, 'wb')
+
+        if pred_ckpt_path is not None:
+            self.predictor = Predictor.load_from_checkpoint(pred_ckpt_path)
+        else:
+            self.predictor = Predictor(window_size=self.window_size, lstm_dim=self.pred_dim,
+                                       lstm_layers=self.pred_layers)
+        self.joiner = nn.Sequential(
+            nn.Conv2d(3, 48, kernel_size=(9, 1), stride=1, padding=(4, 0), padding_mode='reflect',
+                      groups=3),
+            nn.LeakyReLU(0.2),
+            nn.Conv2d(48, 2, kernel_size=1, stride=1, padding=0, groups=2),
+        )
+
+        self.encoder = Encoder(in_dim=self.window_size, dim=self.enc_in_dim, depth=self.enc_layers,
+                               mlp_dim=self.enc_dim)
+
+        self.loss = Loss()
+        self.window = torch.sqrt(torch.hann_window(self.window_size))
+        self.save_hyperparameters('window_size', 'enc_layers', 'enc_in_dim', 'enc_dim', 'pred_dim', 'pred_layers')
+
+    def forward(self, x):
+        """
+        Input: real-imaginary; shape (B, F, T, 2); F = hop_size + 1
+        Output: real-imaginary
+        """
+
+        B, C, F, T = x.shape
+
+        x = x.permute(3, 0, 1, 2).unsqueeze(-1)
+        prev_mag = torch.zeros((B, 1, F, 1), device=x.device)
+        predictor_state = torch.zeros((2, self.predictor.lstm_layers, 1, self.predictor.lstm_dim), device=x.device)
+        mlp_state = torch.zeros((self.encoder.depth, 2, 1, B, self.encoder.dim), device=x.device)
+        result = []
+        for step in x:
+            feat, mlp_state = self.encoder(step, mlp_state)
+            prev_mag, predictor_state = self.predictor(prev_mag, predictor_state)
+            feat = torch.cat((feat, prev_mag), 1)
+            feat = self.joiner(feat)
+            feat = feat + step
+            result.append(feat)
+            prev_mag = torch.linalg.norm(feat, dim=1, ord=1, keepdims=True)  # compute magnitude
+        output = torch.cat(result, -1)
+        return output
+
+    def forward_onnx(self, x, prev_mag, predictor_state=None, mlp_state=None):
+        prev_mag, predictor_state = self.predictor(prev_mag, predictor_state)
+        feat, mlp_state = self.encoder(x, mlp_state)
+
+        feat = torch.cat((feat, prev_mag), 1)
+        feat = self.joiner(feat)
+        prev_mag = torch.linalg.norm(feat, dim=1, ord=1, keepdims=True)
+        feat = feat + x
+        return feat, prev_mag, predictor_state, mlp_state
+
+    def train_dataloader(self):
+        return DataLoader(self.train_dataset, shuffle=False, batch_size=self.hparams.batch_size,
+                          num_workers=CONFIG.TRAIN.workers)
+
+    def val_dataloader(self):
+        return DataLoader(self.val_dataset, shuffle=False, batch_size=self.hparams.batch_size,
+                          num_workers=CONFIG.TRAIN.workers)
+
+    def training_step(self, batch, batch_idx):
+        x_in, y = batch
+        f_0 = x_in[:, :, 0:1, :]
+        x = x_in[:, :, 1:, :]
+
+        x = self(x)
+        x = torch.cat([f_0, x], dim=2)
+
+        loss = self.loss(x, y)
+        self.log('train_loss', loss, logger=True)
+        return loss
+
+    def validation_step(self, val_batch, batch_idx):
+        x, y = val_batch
+        f_0 = x[:, :, 0:1, :]
+        x_in = x[:, :, 1:, :]
+
+        pred = self(x_in)
+        pred = torch.cat([f_0, pred], dim=2)
+
+        loss = self.loss(pred, y)
+        self.window = self.window.to(pred.device)
+        pred = torch.view_as_complex(pred.permute(0, 2, 3, 1).contiguous())
+        pred = torch.istft(pred, self.window_size, self.hop_size, window=self.window)
+        y = torch.view_as_complex(y.permute(0, 2, 3, 1).contiguous())
+        y = torch.istft(y, self.window_size, self.hop_size, window=self.window)
+
+        self.log('val_loss', loss, on_step=False, on_epoch=True, logger=True, prog_bar=True, sync_dist=True)
+
+        if batch_idx == 0:
+            i = torch.randint(0, x.shape[0], (1,)).item()
+            x = torch.view_as_complex(x.permute(0, 2, 3, 1).contiguous())
+            x = torch.istft(x[i], self.window_size, self.hop_size, window=self.window)
+
+            self.trainer.logger.log_spectrogram(y[i], x, pred[i], self.current_epoch)
+            self.trainer.logger.log_audio(y[i], x, pred[i], self.current_epoch)
+
+    def test_step(self, test_batch, batch_idx):
+        inp, tar, inp_wav, tar_wav = test_batch
+        inp_wav = inp_wav.squeeze()
+        tar_wav = tar_wav.squeeze()
+        f_0 = inp[:, :, 0:1, :]
+        x = inp[:, :, 1:, :]
+        pred = self(x)
+        pred = torch.cat([f_0, pred], dim=2)
+        pred = torch.istft(pred.squeeze(0).permute(1, 2, 0), self.window_size, self.hop_size,
+                           window=self.window.to(pred.device))
+        stoi = self.stoi(pred, tar_wav)
+
+        tar_wav = tar_wav.cpu().numpy()
+        inp_wav = inp_wav.cpu().numpy()
+        pred = pred.detach().cpu().numpy()
+        lsd, _ = LSD(tar_wav, pred)
+
+        if batch_idx in [3, 5, 7]:
+            sample_path = os.path.join(CONFIG.LOG.sample_path)
+            path = os.path.join(sample_path, 'sample_' + str(batch_idx))
+            visualize(tar_wav, inp_wav, pred, path)
+            sf.write(os.path.join(path, 'enhanced_output.wav'), pred, samplerate=CONFIG.DATA.sr, subtype='PCM_16')
+            sf.write(os.path.join(path, 'lossy_input.wav'), inp_wav, samplerate=CONFIG.DATA.sr, subtype='PCM_16')
+            sf.write(os.path.join(path, 'target.wav'), tar_wav, samplerate=CONFIG.DATA.sr, subtype='PCM_16')
+        if CONFIG.DATA.sr != 16000:
+            pred = librosa.resample(pred, 48000, 16000)
+            tar_wav = librosa.resample(tar_wav, 48000, 16000, res_type='kaiser_fast')
+        ret = plcmos.run(pred, tar_wav)
+        pesq = self.pesq(torch.tensor(pred), torch.tensor(tar_wav))
+        metrics = {
+            "Intrusive": ret[0],
+            "Non-intrusive": ret[1],
+            'LSD': lsd,
+            'STOI': stoi,
+            'PESQ': pesq,
+        }
+        self.log_dict(metrics)
+        return metrics
+
+    def predict_step(self, batch, batch_idx: int, dataloader_idx: int = 0):
+        f_0 = batch[:, :, 0:1, :]
+        x = batch[:, :, 1:, :]
+        pred = self(x)
+        pred = torch.cat([f_0, pred], dim=2)
+        pred = torch.istft(pred.squeeze(0).permute(1, 2, 0), self.window_size, self.hop_size,
+                           window=self.window.to(pred.device))
+        return pred
+
+    def configure_optimizers(self):
+        optimizer = torch.optim.Adam(self.parameters(), lr=self.learning_rate)
+        lr_scheduler = torch.optim.lr_scheduler.ReduceLROnPlateau(optimizer, patience=CONFIG.TRAIN.patience,
+                                                                  factor=CONFIG.TRAIN.factor, verbose=True)
+
+        scheduler = {
+            'scheduler': lr_scheduler,
+            'reduce_on_plateau': True,
+            'monitor': 'val_loss'
+        }
+        return [optimizer], [scheduler]
+
+
+class OnnxWrapper(pl.LightningModule):
+    def __init__(self, model, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+        self.model = model
+        batch_size = 1
+        pred_states = torch.zeros((2, 1, 1, model.predictor.lstm_dim))
+        mlp_states = torch.zeros((model.encoder.depth, 2, 1, batch_size, model.encoder.dim))
+        mag = torch.zeros((batch_size, 1, model.hop_size, 1))
+        x = torch.randn(batch_size, model.hop_size + 1, 2)
+        self.sample = (x, mag, pred_states, mlp_states)
+        self.input_names = ['input', 'mag_in_cached_', 'pred_state_in_cached_', 'mlp_state_in_cached_']
+        self.output_names = ['output', 'mag_out_cached_', 'pred_state_out_cached_', 'mlp_state_out_cached_']
+
+    def forward(self, x, prev_mag, predictor_state=None, mlp_state=None):
+        x = x.permute(0, 2, 1).unsqueeze(-1)
+        f_0 = x[:, :, 0:1, :]
+        x = x[:, :, 1:, :]
+
+        output, prev_mag, predictor_state, mlp_state = self.model.forward_onnx(x, prev_mag, predictor_state, mlp_state)
+        output = torch.cat([f_0, output], dim=2)
+        output = output.squeeze(-1).permute(0, 2, 1)
+        return output, prev_mag, predictor_state, mlp_state

requirements.txt

@@ -0,0 +1,17 @@
+auraloss==0.3.0
+einops==0.6.0
+librosa==0.9.2
+matplotlib==3.5.3
+numpy==1.22.3
+onnxruntime==1.13.1
+pandas==1.5.3
+pydtmc==7.0.0
+pytorch_lightning==1.9.0
+scikit_learn==1.2.1
+soundfile==0.11.0
+torch==1.13.1
+torchmetrics==0.11.0
+tqdm==4.64.0
+stoi==0.3.3
+pesq==0.0.4
+onnx==1.13.0

utils/stft.py

@@ -0,0 +1,23 @@
+import torch
+import torch.nn as nn
+
+
+class STFTMag(nn.Module):
+    def __init__(self,
+                 nfft=1024,
+                 hop=256):
+        super().__init__()
+        self.nfft = nfft
+        self.hop = hop
+        self.register_buffer('window', torch.hann_window(nfft), False)
+
+    # x: [B,T] or [T]
+    @torch.no_grad()
+    def forward(self, x):
+        stft = torch.stft(x.cpu(),
+                          self.nfft,
+                          self.hop,
+                          window=self.window,
+                          )  # return_complex=False)  #[B, F, TT,2]
+        mag = torch.norm(stft, p=2, dim=-1)  # [B, F, TT]
+        return mag

utils/tblogger.py

@@ -0,0 +1,71 @@
+from os import path
+
+import librosa as rosa
+import matplotlib
+import matplotlib.pyplot as plt
+import numpy as np
+from pytorch_lightning.loggers import TensorBoardLogger
+from pytorch_lightning.utilities import rank_zero_only
+
+from utils.stft import STFTMag
+
+matplotlib.use('Agg')
+
+
+class TensorBoardLoggerExpanded(TensorBoardLogger):
+    def __init__(self, sr=16000):
+        super().__init__(save_dir='lightning_logs', default_hp_metric=False, name='')
+        self.sr = sr
+        self.stftmag = STFTMag()
+
+    def fig2np(self, fig):
+        data = np.fromstring(fig.canvas.tostring_rgb(), dtype=np.uint8, sep='')
+        data = data.reshape(fig.canvas.get_width_height()[::-1] + (3,))
+        return data
+
+    def plot_spectrogram_to_numpy(self, y, y_low, y_recon, step):
+        name_list = ['y', 'y_low', 'y_recon']
+        fig = plt.figure(figsize=(9, 15))
+        fig.suptitle(f'Epoch_{step}')
+        for i, yy in enumerate([y, y_low, y_recon]):
+            if yy.dim() == 1:
+                yy = self.stftmag(yy)
+            ax = plt.subplot(3, 1, i + 1)
+            ax.set_title(name_list[i])
+            plt.imshow(rosa.amplitude_to_db(yy.numpy(),
+                                            ref=np.max, top_db=80.),
+                       # vmin = -20,
+                       vmax=0.,
+                       aspect='auto',
+                       origin='lower',
+                       interpolation='none')
+            plt.colorbar()
+            plt.xlabel('Frames')
+            plt.ylabel('Channels')
+            plt.tight_layout()
+
+        fig.canvas.draw()
+        data = self.fig2np(fig)
+
+        plt.close()
+        return data
+
+    @rank_zero_only
+    def log_spectrogram(self, y, y_low, y_recon, epoch):
+        y, y_low, y_recon = y.detach().cpu(), y_low.detach().cpu(), y_recon.detach().cpu()
+        spec_img = self.plot_spectrogram_to_numpy(y, y_low, y_recon, epoch)
+        self.experiment.add_image(path.join(self.save_dir, 'result'),
+                                  spec_img,
+                                  epoch,
+                                  dataformats='HWC')
+        self.experiment.flush()
+        return
+
+    @rank_zero_only
+    def log_audio(self, y, y_low, y_recon, epoch):
+        y, y_low, y_recon = y.detach().cpu(), y_low.detach().cpu(), y_recon.detach().cpu(),
+        name_list = ['y', 'y_low', 'y_recon']
+        for n, yy in zip(name_list, [y, y_low, y_recon]):
+            self.experiment.add_audio(n, yy, epoch, self.sr)
+        self.experiment.flush()
+        return

utils/utils.py

@@ -0,0 +1,67 @@
+import os
+
+import librosa
+import librosa.display
+import matplotlib.pyplot as plt
+import numpy as np
+from matplotlib.backends.backend_agg import FigureCanvasAgg as FigureCanvas
+
+from config import CONFIG
+
+
+def mkdir_p(mypath):
+    """Creates a directory. equivalent to using mkdir -p on the command line"""
+
+    from errno import EEXIST
+    from os import makedirs, path
+
+    try:
+        makedirs(mypath)
+    except OSError as exc:  # Python >2.5
+        if exc.errno == EEXIST and path.isdir(mypath):
+            pass
+        else:
+            raise
+
+
+def visualize(hr, lr, recon, path):
+    sr = CONFIG.DATA.sr
+    window_size = 1024
+    window = np.hanning(window_size)
+
+    stft_hr = librosa.core.spectrum.stft(hr, n_fft=window_size, hop_length=512, window=window)
+    stft_hr = 2 * np.abs(stft_hr) / np.sum(window)
+
+    stft_lr = librosa.core.spectrum.stft(lr, n_fft=window_size, hop_length=512, window=window)
+    stft_lr = 2 * np.abs(stft_lr) / np.sum(window)
+
+    stft_recon = librosa.core.spectrum.stft(recon, n_fft=window_size, hop_length=512, window=window)
+    stft_recon = 2 * np.abs(stft_recon) / np.sum(window)
+
+    fig, (ax1, ax2, ax3) = plt.subplots(3, 1, sharey=True, sharex=True, figsize=(16, 10))
+    ax1.title.set_text('HR signal')
+    ax2.title.set_text('LR signal')
+    ax3.title.set_text('Reconstructed signal')
+
+    canvas = FigureCanvas(fig)
+    p = librosa.display.specshow(librosa.amplitude_to_db(stft_hr), ax=ax1, y_axis='linear', x_axis='time', sr=sr)
+    p = librosa.display.specshow(librosa.amplitude_to_db(stft_lr), ax=ax2, y_axis='linear', x_axis='time', sr=sr)
+    p = librosa.display.specshow(librosa.amplitude_to_db(stft_recon), ax=ax3, y_axis='linear', x_axis='time', sr=sr)
+    mkdir_p(path)
+    fig.savefig(os.path.join(path, 'spec.png'))
+
+
+def get_power(x, nfft):
+    S = librosa.stft(x, nfft)
+    S = np.log(np.abs(S) ** 2 + 1e-8)
+    return S
+
+
+def LSD(x_hr, x_pr):
+    S1 = get_power(x_hr, nfft=2048)
+    S2 = get_power(x_pr, nfft=2048)
+    lsd = np.mean(np.sqrt(np.mean((S1 - S2) ** 2 + 1e-8, axis=-1)), axis=0)
+    S1 = S1[-(len(S1) - 1) // 2:, :]
+    S2 = S2[-(len(S2) - 1) // 2:, :]
+    lsd_high = np.mean(np.sqrt(np.mean((S1 - S2) ** 2 + 1e-8, axis=-1)), axis=0)
+    return lsd, lsd_high