Upload 4 files

config.py

@@ -0,0 +1,79 @@
+import argparse
+import sys
+import torch
+from multiprocessing import cpu_count
+
+class Config:
+    def __init__(self):
+        self.device = "cuda:0"
+        self.is_half = True
+        self.n_cpu = 0
+        self.gpu_name = None
+        self.gpu_mem = None
+        self.python_cmd, self.listen_port, self.iscolab, self.noparallel, self.noautoopen = self.arg_parse()
+        self.x_pad, self.x_query, self.x_center, self.x_max = self.device_config()
+
+    @staticmethod
+    def arg_parse() -> tuple:
+        exe = sys.executable or "python"
+        parser = argparse.ArgumentParser()
+        parser.add_argument("--port", type=int, default=7865, help="Listen port")
+        parser.add_argument("--pycmd", type=str, default=exe, help="Python command")
+        parser.add_argument("--colab", action="store_true", help="Launch in colab")
+        parser.add_argument("--noparallel", action="store_true", help="Disable parallel processing")
+        parser.add_argument("--noautoopen", action="store_true", help="Do not open in browser automatically")
+        
+        if len(sys.argv) > 1 and sys.argv[0].endswith("colab_kernel_launcher.py"):
+            args = parser.parse_known_args(sys.argv[1:])[0]
+        else:
+            args = parser.parse_args()
+
+        args.port = args.port if 0 <= args.port <= 65535 else 7865
+
+        return args.pycmd, args.port, args.colab, args.noparallel, args.noautoopen
+
+    @staticmethod
+    def has_mps() -> bool:
+        if not torch.backends.mps.is_available():
+            return False
+        try:
+            torch.zeros(1).to(torch.device("mps"))
+            return True
+        except Exception:
+            return False
+
+    def device_config(self) -> tuple:
+        if torch.cuda.is_available():
+            i_device = int(self.device.split(":")[-1])
+            self.gpu_name = torch.cuda.get_device_name(i_device)
+            if ("16" in self.gpu_name and "V100" not in self.gpu_name.upper()) or "P40" in self.gpu_name.upper() or "1060" in self.gpu_name or "1070" in self.gpu_name or "1080" in self.gpu_name:
+                print("Found GPU", self.gpu_name, ", force to fp32")
+                self.is_half = False
+            else:
+                print("Found GPU", self.gpu_name)
+            self.gpu_mem = int(torch.cuda.get_device_properties(i_device).total_memory / 1024 / 1024 / 1024 + 0.4)
+        elif self.has_mps():
+            print("No supported Nvidia GPU found, use MPS instead")
+            self.device = "mps"
+            self.is_half = False
+        else:
+            print("No supported Nvidia GPU found, use CPU instead")
+            self.device = "cpu"
+            self.is_half = False
+
+        if self.n_cpu == 0:
+            self.n_cpu = cpu_count()
+
+        if self.is_half:
+            x_pad, x_query, x_center, x_max = 3, 10, 60, 65
+        else:
+            x_pad, x_query, x_center, x_max = 1, 6, 38, 41
+
+        if self.gpu_mem is not None and self.gpu_mem <= 4:
+            x_pad, x_query, x_center, x_max = 1, 5, 30, 32
+
+        return x_pad, x_query, x_center, x_max
+
+if __name__ == "__main__":
+    config = Config()
+    print(config.__dict__)

handler.py

@@ -1,118 +1,75 @@
-from pydantic import BaseModel
-from environs import Env
-from typing import List, Dict, Any
-import os
-import base64
-import numpy as np
-import librosa
-from scipy.io import wavfile
-import asyncio
-import shutil
-import zipfile
-import requests
-
-
-def download_and_extract_files():
-    files_to_download = [
-        ("config.py", "https://www.dropbox.com/scl/fi/ls7vmjk75uou8ayfn6kj4/config.py?rlkey=4qluzl5l07zq1j9mkl9n6j66u&st=0yit9dzx&dl=1"),
-        ("hubert_base.pt", "https://www.dropbox.com/scl/fi/g7oohuwfzlnrbd8zic6gj/hubert_base.pt?rlkey=ddeyqex1morsm54azyakmd62e&st=rsrvf964&dl=1"),
-        ("lib.zip", "https://www.dropbox.com/scl/fi/ia6p6cf5xvcbi78dmkbbz/lib.zip?rlkey=k3chc1nlaswsqdo7slqco56wi&st=19n9syfd&dl=1"),
-        ("rmvpe.pt", "https://www.dropbox.com/scl/fi/7pl7u6fvydwgtz19n8nzx/rmvpe.pt?rlkey=tnbxmarogivbw3qy34hgy7r7q&st=um8d4230&dl=1"),
-        ("rmvpe.py", "https://www.dropbox.com/scl/fi/i2shk4otwyg4ns8yod5h1/rmvpe.py?rlkey=l7313htdh1ihylb6bx91el0lv&st=xhkfog8j&dl=1"),
-        ("vc_infer_pipeline.py", "https://www.dropbox.com/scl/fi/bvz7s2wf2y67twpg583lg/vc_infer_pipeline.py?rlkey=q4w7oww5e7e2qdfh3herofk4o&st=4sck87ny&dl=1"),
-        ("voice_processing.py", "https://www.dropbox.com/scl/fi/emrmjsuz0mod4r2x9e43f/voice_processing.py?rlkey=6baomwowns9y3yq1pl6syer0t&st=d9u51gba&dl=1"),
-        ("weights.zip", "https://www.dropbox.com/scl/fi/tr5a04wlow5go8cv3d6qp/weights.zip?rlkey=qvpwax97nn5a4iv79g76lcbz2&st=5ueb2gva&dl=1")
-    ]
-
-    for file_name, url in files_to_download:
-        if not os.path.exists(file_name):
-            response = requests.get(url)
-            with open(file_name, "wb") as file:
-                file.write(response.content)
-
-            if file_name.endswith(".zip"):
-                with zipfile.ZipFile(file_name, "r") as zip_ref:
-                    extract_to = os.path.splitext(file_name)[0]
-                    for member in zip_ref.namelist():
-                        # Extract files into the target directory without the first part of the path
-                        member_path = os.path.join(extract_to, *member.split('/')[1:])
-                        if member.endswith('/'):
-                            os.makedirs(member_path, exist_ok=True)
-                        else:
-                            os.makedirs(os.path.dirname(member_path), exist_ok=True)
-                            with open(member_path, 'wb') as f:
-                                f.write(zip_ref.read(member))
-
-                # Optionally, remove the zip file after extraction
-                os.remove(file_name)
-
-# Run the function
-download_and_extract_files()
-
-from voice_processing import tts, get_model_names, voice_mapping, get_unique_filename
-
-
-class EndpointHandler:
-    def __init__(self, model_dir=None):
-        self.model_dir = model_dir
-
-    def __call__(self, data: Dict[str, Any]) -> Dict[str, Any]:
-        try:
-            if "inputs" in data:  # Check if data is in Hugging Face JSON format
-                return self.process_hf_input(data)
-            else:
-                return self.process_json_input(data)
-        except ValueError as e:
-            return {"error": str(e)}
-        except Exception as e:
-            return {"error": str(e)}
-
-    def process_json_input(self, json_data):
-        if all(key in json_data for key in ["model_name", "tts_text", "selected_voice", "slang_rate", "use_uploaded_voice"]):
-            model_name = json_data["model_name"]
-            tts_text = json_data["tts_text"]
-            selected_voice = json_data["selected_voice"]
-            slang_rate = json_data["slang_rate"]
-            use_uploaded_voice = json_data["use_uploaded_voice"]
-            voice_upload_file = json_data.get("voice_upload_file", None)
-
-            edge_tts_voice = voice_mapping.get(selected_voice)
-            if not edge_tts_voice:
-                raise ValueError(f"Invalid voice '{selected_voice}'.")
-
-            info, edge_tts_output_path, tts_output_data, edge_output_file = asyncio.run(tts(
-                model_name, tts_text, edge_tts_voice, slang_rate, use_uploaded_voice, voice_upload_file
-            ))
-
-            if edge_output_file and os.path.exists(edge_output_file):
-                os.remove(edge_output_file)
-
-            _, audio_output = tts_output_data
-
-            audio_file_path = self.save_audio_data_to_file(audio_output) if isinstance(audio_output, np.ndarray) else audio_output
-
-            try:
-                with open(audio_file_path, 'rb') as file:
-                    audio_bytes = file.read()
-                audio_data_uri = f"data:audio/wav;base64,{base64.b64encode(audio_bytes).decode('utf-8')}"
-            except Exception as e:
-                raise Exception(f"Failed to read audio file: {e}")
-            finally:
-                if os.path.exists(audio_file_path):
-                    os.remove(audio_file_path)
-
-            return {"info": info, "audio_data_uri": audio_data_uri}
-        else:
-            raise ValueError("Invalid JSON structure.")
-
-    def process_hf_input(self, hf_data):
-        if "inputs" in hf_data:
-            actual_data = hf_data["inputs"]
-            return self.process_json_input(actual_data)
-        else:
-            return {"error": "Invalid Hugging Face JSON structure."}
-
-    def save_audio_data_to_file(self, audio_data, sample_rate=40000):
-        file_path = get_unique_filename('wav')
-        wavfile.write(file_path, sample_rate, audio_data)
+from pydantic import BaseModel
+from environs import Env
+from typing import List, Dict, Any
+import os
+import base64
+import numpy as np
+import librosa
+from scipy.io import wavfile
+import asyncio
+from voice_processing import tts, get_model_names, voice_mapping, get_unique_filename
+
+class EndpointHandler:
+    def __init__(self, model_dir=None):
+        self.model_dir = model_dir
+
+    def __call__(self, data: Dict[str, Any]) -> Dict[str, Any]:
+        try:
+            if "inputs" in data:  # Check if data is in Hugging Face JSON format
+                return self.process_hf_input(data)
+            else:
+                return self.process_json_input(data)
+        except ValueError as e:
+            return {"error": str(e)}
+        except Exception as e:
+            return {"error": str(e)}
+
+    def process_json_input(self, json_data):
+        if all(key in json_data for key in ["model_name", "tts_text", "selected_voice", "slang_rate", "use_uploaded_voice"]):
+            model_name = json_data["model_name"]
+            tts_text = json_data["tts_text"]
+            selected_voice = json_data["selected_voice"]
+            slang_rate = json_data["slang_rate"]
+            use_uploaded_voice = json_data["use_uploaded_voice"]
+            voice_upload_file = json_data.get("voice_upload_file", None)
+
+            edge_tts_voice = voice_mapping.get(selected_voice)
+            if not edge_tts_voice:
+                raise ValueError(f"Invalid voice '{selected_voice}'.")
+
+            info, edge_tts_output_path, tts_output_data, edge_output_file = asyncio.run(tts(
+                model_name, tts_text, edge_tts_voice, slang_rate, use_uploaded_voice, voice_upload_file
+            ))
+
+            if edge_output_file and os.path.exists(edge_output_file):
+                os.remove(edge_output_file)
+
+            _, audio_output = tts_output_data
+
+            audio_file_path = self.save_audio_data_to_file(audio_output) if isinstance(audio_output, np.ndarray) else audio_output
+
+            try:
+                with open(audio_file_path, 'rb') as file:
+                    audio_bytes = file.read()
+                audio_data_uri = f"data:audio/wav;base64,{base64.b64encode(audio_bytes).decode('utf-8')}"
+            except Exception as e:
+                raise Exception(f"Failed to read audio file: {e}")
+            finally:
+                if os.path.exists(audio_file_path):
+                    os.remove(audio_file_path)
+
+            return {"info": info, "audio_data_uri": audio_data_uri}
+        else:
+            raise ValueError("Invalid JSON structure.")
+
+    def process_hf_input(self, hf_data):
+        if "inputs" in hf_data:
+            actual_data = hf_data["inputs"]
+            return self.process_json_input(actual_data)
+        else:
+            return {"error": "Invalid Hugging Face JSON structure."}
+
+    def save_audio_data_to_file(self, audio_data, sample_rate=40000):
+        file_path = get_unique_filename('wav')
+        wavfile.write(file_path, sample_rate, audio_data)
         return file_path

vc_infer_pipeline.py

@@ -0,0 +1,451 @@
+import os
+import sys
+import traceback
+from functools import lru_cache
+from time import time as ttime
+
+import faiss
+import librosa
+import numpy as np
+import parselmouth
+import pyworld
+import torch
+import torch.nn.functional as F
+import torchcrepe
+from scipy import signal
+
+now_dir = os.getcwd()
+sys.path.append(now_dir)
+
+bh, ah = signal.butter(N=5, Wn=48, btype="high", fs=16000)
+
+input_audio_path2wav = {}
+
+
+@lru_cache
+def cache_harvest_f0(input_audio_path, fs, f0max, f0min, frame_period):
+    audio = input_audio_path2wav[input_audio_path]
+    f0, t = pyworld.harvest(
+        audio,
+        fs=fs,
+        f0_ceil=f0max,
+        f0_floor=f0min,
+        frame_period=frame_period,
+    )
+    f0 = pyworld.stonemask(audio, f0, t, fs)
+    return f0
+
+
+def change_rms(data1, sr1, data2, sr2, rate):  # 1是输入音频，2是输出音频,rate是2的占比
+    # print(data1.max(),data2.max())
+    rms1 = librosa.feature.rms(
+        y=data1, frame_length=sr1 // 2 * 2, hop_length=sr1 // 2
+    )  # 每半秒一个点
+    rms2 = librosa.feature.rms(y=data2, frame_length=sr2 // 2 * 2, hop_length=sr2 // 2)
+    rms1 = torch.from_numpy(rms1)
+    rms1 = F.interpolate(
+        rms1.unsqueeze(0), size=data2.shape[0], mode="linear"
+    ).squeeze()
+    rms2 = torch.from_numpy(rms2)
+    rms2 = F.interpolate(
+        rms2.unsqueeze(0), size=data2.shape[0], mode="linear"
+    ).squeeze()
+    rms2 = torch.max(rms2, torch.zeros_like(rms2) + 1e-6)
+    data2 *= (
+        torch.pow(rms1, torch.tensor(1 - rate))
+        * torch.pow(rms2, torch.tensor(rate - 1))
+    ).numpy()
+    return data2
+
+
+class VC(object):
+    def __init__(self, tgt_sr, config):
+        self.x_pad, self.x_query, self.x_center, self.x_max, self.is_half = (
+            config.x_pad,
+            config.x_query,
+            config.x_center,
+            config.x_max,
+            config.is_half,
+        )
+        self.sr = 16000  # hubert输入采样率
+        self.window = 160  # 每帧点数
+        self.t_pad = self.sr * self.x_pad  # 每条前后pad时间
+        self.t_pad_tgt = tgt_sr * self.x_pad
+        self.t_pad2 = self.t_pad * 2
+        self.t_query = self.sr * self.x_query  # 查询切点前后查询时间
+        self.t_center = self.sr * self.x_center  # 查询切点位置
+        self.t_max = self.sr * self.x_max  # 免查询时长阈值
+        self.device = config.device
+
+    def get_f0(
+        self,
+        input_audio_path,
+        x,
+        p_len,
+        f0_up_key,
+        f0_method,
+        filter_radius,
+        inp_f0=None,
+    ):
+        global input_audio_path2wav
+        time_step = self.window / self.sr * 1000
+        f0_min = 50
+        f0_max = 1100
+        f0_mel_min = 1127 * np.log(1 + f0_min / 700)
+        f0_mel_max = 1127 * np.log(1 + f0_max / 700)
+        if f0_method == "pm":
+            f0 = (
+                parselmouth.Sound(x, self.sr)
+                .to_pitch_ac(
+                    time_step=time_step / 1000,
+                    voicing_threshold=0.6,
+                    pitch_floor=f0_min,
+                    pitch_ceiling=f0_max,
+                )
+                .selected_array["frequency"]
+            )
+            pad_size = (p_len - len(f0) + 1) // 2
+            if pad_size > 0 or p_len - len(f0) - pad_size > 0:
+                f0 = np.pad(
+                    f0, [[pad_size, p_len - len(f0) - pad_size]], mode="constant"
+                )
+        elif f0_method == "harvest":
+            input_audio_path2wav[input_audio_path] = x.astype(np.double)
+            f0 = cache_harvest_f0(input_audio_path, self.sr, f0_max, f0_min, 10)
+            if filter_radius > 2:
+                f0 = signal.medfilt(f0, 3)
+        elif f0_method == "crepe":
+            model = "full"
+            # Pick a batch size that doesn't cause memory errors on your gpu
+            batch_size = 512
+            # Compute pitch using first gpu
+            audio = torch.tensor(np.copy(x))[None].float()
+            f0, pd = torchcrepe.predict(
+                audio,
+                self.sr,
+                self.window,
+                f0_min,
+                f0_max,
+                model,
+                batch_size=batch_size,
+                device=self.device,
+                return_periodicity=True,
+            )
+            pd = torchcrepe.filter.median(pd, 3)
+            f0 = torchcrepe.filter.mean(f0, 3)
+            f0[pd < 0.1] = 0
+            f0 = f0[0].cpu().numpy()
+        elif f0_method == "rmvpe":
+            if hasattr(self, "model_rmvpe") == False:
+                from rmvpe import RMVPE
+
+                print("loading rmvpe model")
+                self.model_rmvpe = RMVPE(
+                    "rmvpe.pt", is_half=self.is_half, device=self.device
+                )
+            f0 = self.model_rmvpe.infer_from_audio(x, thred=0.03)
+        f0 *= pow(2, f0_up_key / 12)
+        # with open("test.txt","w")as f:f.write("\n".join([str(i)for i in f0.tolist()]))
+        tf0 = self.sr // self.window  # 每秒f0点数
+        if inp_f0 is not None:
+            delta_t = np.round(
+                (inp_f0[:, 0].max() - inp_f0[:, 0].min()) * tf0 + 1
+            ).astype("int16")
+            replace_f0 = np.interp(
+                list(range(delta_t)), inp_f0[:, 0] * 100, inp_f0[:, 1]
+            )
+            shape = f0[self.x_pad * tf0 : self.x_pad * tf0 + len(replace_f0)].shape[0]
+            f0[self.x_pad * tf0 : self.x_pad * tf0 + len(replace_f0)] = replace_f0[
+                :shape
+            ]
+        # with open("test_opt.txt","w")as f:f.write("\n".join([str(i)for i in f0.tolist()]))
+        f0bak = f0.copy()
+        f0_mel = 1127 * np.log(1 + f0 / 700)
+        f0_mel[f0_mel > 0] = (f0_mel[f0_mel > 0] - f0_mel_min) * 254 / (
+            f0_mel_max - f0_mel_min
+        ) + 1
+        f0_mel[f0_mel <= 1] = 1
+        f0_mel[f0_mel > 255] = 255
+        f0_coarse = np.rint(f0_mel).astype(int)
+        return f0_coarse, f0bak  # 1-0
+
+    def vc(
+        self,
+        model,
+        net_g,
+        sid,
+        audio0,
+        pitch,
+        pitchf,
+        times,
+        index,
+        big_npy,
+        index_rate,
+        version,
+        protect,
+    ):  # ,file_index,file_big_npy
+        feats = torch.from_numpy(audio0)
+        if self.is_half:
+            feats = feats.half()
+        else:
+            feats = feats.float()
+        if feats.dim() == 2:  # double channels
+            feats = feats.mean(-1)
+        assert feats.dim() == 1, feats.dim()
+        feats = feats.view(1, -1)
+        padding_mask = torch.BoolTensor(feats.shape).to(self.device).fill_(False)
+
+        inputs = {
+            "source": feats.to(self.device),
+            "padding_mask": padding_mask,
+            "output_layer": 9 if version == "v1" else 12,
+        }
+        t0 = ttime()
+        with torch.no_grad():
+            logits = model.extract_features(**inputs)
+            feats = model.final_proj(logits[0]) if version == "v1" else logits[0]
+        if protect < 0.5 and pitch != None and pitchf != None:
+            feats0 = feats.clone()
+        if (
+            isinstance(index, type(None)) == False
+            and isinstance(big_npy, type(None)) == False
+            and index_rate != 0
+        ):
+            npy = feats[0].cpu().numpy()
+            if self.is_half:
+                npy = npy.astype("float32")
+
+            # _, I = index.search(npy, 1)
+            # npy = big_npy[I.squeeze()]
+
+            score, ix = index.search(npy, k=8)
+            weight = np.square(1 / score)
+            weight /= weight.sum(axis=1, keepdims=True)
+            npy = np.sum(big_npy[ix] * np.expand_dims(weight, axis=2), axis=1)
+
+            if self.is_half:
+                npy = npy.astype("float16")
+            feats = (
+                torch.from_numpy(npy).unsqueeze(0).to(self.device) * index_rate
+                + (1 - index_rate) * feats
+            )
+
+        feats = F.interpolate(feats.permute(0, 2, 1), scale_factor=2).permute(0, 2, 1)
+        if protect < 0.5 and pitch != None and pitchf != None:
+            feats0 = F.interpolate(feats0.permute(0, 2, 1), scale_factor=2).permute(
+                0, 2, 1
+            )
+        t1 = ttime()
+        p_len = audio0.shape[0] // self.window
+        if feats.shape[1] < p_len:
+            p_len = feats.shape[1]
+            if pitch != None and pitchf != None:
+                pitch = pitch[:, :p_len]
+                pitchf = pitchf[:, :p_len]
+
+        if protect < 0.5 and pitch != None and pitchf != None:
+            pitchff = pitchf.clone()
+            pitchff[pitchf > 0] = 1
+            pitchff[pitchf < 1] = protect
+            pitchff = pitchff.unsqueeze(-1)
+            feats = feats * pitchff + feats0 * (1 - pitchff)
+            feats = feats.to(feats0.dtype)
+        p_len = torch.tensor([p_len], device=self.device).long()
+        with torch.no_grad():
+            if pitch != None and pitchf != None:
+                audio1 = (
+                    (net_g.infer(feats, p_len, pitch, pitchf, sid)[0][0, 0])
+                    .data.cpu()
+                    .float()
+                    .numpy()
+                )
+            else:
+                audio1 = (
+                    (net_g.infer(feats, p_len, sid)[0][0, 0]).data.cpu().float().numpy()
+                )
+        del feats, p_len, padding_mask
+        if torch.cuda.is_available():
+            torch.cuda.empty_cache()
+        t2 = ttime()
+        times[0] += t1 - t0
+        times[2] += t2 - t1
+        return audio1
+
+    def pipeline(
+        self,
+        model,
+        net_g,
+        sid,
+        audio,
+        input_audio_path,
+        times,
+        f0_up_key,
+        f0_method,
+        file_index,
+        # file_big_npy,
+        index_rate,
+        if_f0,
+        filter_radius,
+        tgt_sr,
+        resample_sr,
+        rms_mix_rate,
+        version,
+        protect,
+        f0_file=None,
+    ):
+        if (
+            file_index != ""
+            # and file_big_npy != ""
+            # and os.path.exists(file_big_npy) == True
+            and os.path.exists(file_index) == True
+            and index_rate != 0
+        ):
+            try:
+                index = faiss.read_index(file_index)
+                # big_npy = np.load(file_big_npy)
+                big_npy = index.reconstruct_n(0, index.ntotal)
+            except:
+                traceback.print_exc()
+                index = big_npy = None
+        else:
+            index = big_npy = None
+        audio = signal.filtfilt(bh, ah, audio)
+        audio_pad = np.pad(audio, (self.window // 2, self.window // 2), mode="reflect")
+        opt_ts = []
+        if audio_pad.shape[0] > self.t_max:
+            audio_sum = np.zeros_like(audio)
+            for i in range(self.window):
+                audio_sum += audio_pad[i : i - self.window]
+            for t in range(self.t_center, audio.shape[0], self.t_center):
+                opt_ts.append(
+                    t
+                    - self.t_query
+                    + np.where(
+                        np.abs(audio_sum[t - self.t_query : t + self.t_query])
+                        == np.abs(audio_sum[t - self.t_query : t + self.t_query]).min()
+                    )[0][0]
+                )
+        s = 0
+        audio_opt = []
+        t = None
+        t1 = ttime()
+        audio_pad = np.pad(audio, (self.t_pad, self.t_pad), mode="reflect")
+        p_len = audio_pad.shape[0] // self.window
+        inp_f0 = None
+        if hasattr(f0_file, "name") == True:
+            try:
+                with open(f0_file.name, "r") as f:
+                    lines = f.read().strip("\n").split("\n")
+                inp_f0 = []
+                for line in lines:
+                    inp_f0.append([float(i) for i in line.split(",")])
+                inp_f0 = np.array(inp_f0, dtype="float32")
+            except:
+                traceback.print_exc()
+        sid = torch.tensor(sid, device=self.device).unsqueeze(0).long()
+        pitch, pitchf = None, None
+        if if_f0 == 1:
+            pitch, pitchf = self.get_f0(
+                input_audio_path,
+                audio_pad,
+                p_len,
+                f0_up_key,
+                f0_method,
+                filter_radius,
+                inp_f0,
+            )
+            pitch = pitch[:p_len]
+            pitchf = pitchf[:p_len]
+            if self.device == "mps":
+                pitchf = pitchf.astype(np.float32)
+            pitch = torch.tensor(pitch, device=self.device).unsqueeze(0).long()
+            pitchf = torch.tensor(pitchf, device=self.device).unsqueeze(0).float()
+        t2 = ttime()
+        times[1] += t2 - t1
+        for t in opt_ts:
+            t = t // self.window * self.window
+            if if_f0 == 1:
+                audio_opt.append(
+                    self.vc(
+                        model,
+                        net_g,
+                        sid,
+                        audio_pad[s : t + self.t_pad2 + self.window],
+                        pitch[:, s // self.window : (t + self.t_pad2) // self.window],
+                        pitchf[:, s // self.window : (t + self.t_pad2) // self.window],
+                        times,
+                        index,
+                        big_npy,
+                        index_rate,
+                        version,
+                        protect,
+                    )[self.t_pad_tgt : -self.t_pad_tgt]
+                )
+            else:
+                audio_opt.append(
+                    self.vc(
+                        model,
+                        net_g,
+                        sid,
+                        audio_pad[s : t + self.t_pad2 + self.window],
+                        None,
+                        None,
+                        times,
+                        index,
+                        big_npy,
+                        index_rate,
+                        version,
+                        protect,
+                    )[self.t_pad_tgt : -self.t_pad_tgt]
+                )
+            s = t
+        if if_f0 == 1:
+            audio_opt.append(
+                self.vc(
+                    model,
+                    net_g,
+                    sid,
+                    audio_pad[t:],
+                    pitch[:, t // self.window :] if t is not None else pitch,
+                    pitchf[:, t // self.window :] if t is not None else pitchf,
+                    times,
+                    index,
+                    big_npy,
+                    index_rate,
+                    version,
+                    protect,
+                )[self.t_pad_tgt : -self.t_pad_tgt]
+            )
+        else:
+            audio_opt.append(
+                self.vc(
+                    model,
+                    net_g,
+                    sid,
+                    audio_pad[t:],
+                    None,
+                    None,
+                    times,
+                    index,
+                    big_npy,
+                    index_rate,
+                    version,
+                    protect,
+                )[self.t_pad_tgt : -self.t_pad_tgt]
+            )
+        audio_opt = np.concatenate(audio_opt)
+        if rms_mix_rate != 1:
+            audio_opt = change_rms(audio, 16000, audio_opt, tgt_sr, rms_mix_rate)
+        if resample_sr >= 16000 and tgt_sr != resample_sr:
+            audio_opt = librosa.resample(
+                audio_opt, orig_sr=tgt_sr, target_sr=resample_sr
+            )
+        audio_max = np.abs(audio_opt).max() / 0.99
+        max_int16 = 32768
+        if audio_max > 1:
+            max_int16 /= audio_max
+        audio_opt = (audio_opt * max_int16).astype(np.int16)
+        del pitch, pitchf, sid
+        if torch.cuda.is_available():
+            torch.cuda.empty_cache()
+        return audio_opt

voice_processing.py

@@ -0,0 +1,230 @@
+import nest_asyncio
+nest_asyncio.apply()
+
+import asyncio
+import datetime
+import logging
+import os
+import time
+import traceback
+import tempfile
+
+import edge_tts
+import librosa
+import torch
+from fairseq import checkpoint_utils
+import uuid
+
+from config import Config
+from lib.infer_pack.models import (
+    SynthesizerTrnMs256NSFsid,
+    SynthesizerTrnMs256NSFsid_nono,
+    SynthesizerTrnMs768NSFsid,
+    SynthesizerTrnMs768NSFsid_nono,
+)
+from rmvpe import RMVPE
+from vc_infer_pipeline import VC
+
+# Set logging levels
+logging.getLogger("fairseq").setLevel(logging.WARNING)
+logging.getLogger("numba").setLevel(logging.WARNING)
+logging.getLogger("markdown_it").setLevel(logging.WARNING)
+logging.getLogger("urllib3").setLevel(logging.WARNING)
+logging.getLogger("matplotlib").setLevel(logging.WARNING)
+
+limitation = os.getenv("SYSTEM") == "spaces"
+
+config = Config()
+
+# Edge TTS
+tts_voice_list = asyncio.run(edge_tts.list_voices())
+tts_voices = ["mn-MN-BataaNeural", "mn-MN-YesuiNeural"]  # Specific voices
+
+# RVC models
+model_root = "weights"
+models = [d for d in os.listdir(model_root) if os.path.isdir(f"{model_root}/{d}")]
+models.sort()
+
+def get_unique_filename(extension):
+    return f"{uuid.uuid4()}.{extension}"
+
+def model_data(model_name):
+    pth_path = [
+        f"{model_root}/{model_name}/{f}"
+        for f in os.listdir(f"{model_root}/{model_name}")
+        if f.endswith(".pth")
+    ][0]
+    print(f"Loading {pth_path}")
+    cpt = torch.load(pth_path, map_location="cpu")
+    tgt_sr = cpt["config"][-1]
+    cpt["config"][-3] = cpt["weight"]["emb_g.weight"].shape[0]  # n_spk
+    if_f0 = cpt.get("f0", 1)
+    version = cpt.get("version", "v1")
+    if version == "v1":
+        if if_f0 == 1:
+            net_g = SynthesizerTrnMs256NSFsid(*cpt["config"], is_half=config.is_half)
+        else:
+            net_g = SynthesizerTrnMs256NSFsid_nono(*cpt["config"])
+    elif version == "v2":
+        if if_f0 == 1:
+            net_g = SynthesizerTrnMs768NSFsid(*cpt["config"], is_half=config.is_half)
+        else:
+            net_g = SynthesizerTrnMs768NSFsid_nono(*cpt["config"])
+    else:
+        raise ValueError("Unknown version")
+    del net_g.enc_q
+    net_g.load_state_dict(cpt["weight"], strict=False)
+    print("Model loaded")
+    net_g.eval().to(config.device)
+    if config.is_half:
+        net_g = net_g.half()
+    else:
+        net_g = net_g.float()
+    vc = VC(tgt_sr, config)
+    index_files = [
+        f"{model_root}/{model_name}/{f}"
+        for f in os.listdir(f"{model_root}/{model_name}")
+        if f.endswith(".index")
+    ]
+    if len(index_files) == 0:
+        print("No index file found")
+        index_file = ""
+    else:
+        index_file = index_files[0]
+        print(f"Index file found: {index_file}")
+
+    return tgt_sr, net_g, vc, version, index_file, if_f0
+
+def load_hubert():
+    models, _, _ = checkpoint_utils.load_model_ensemble_and_task(
+        ["hubert_base.pt"],
+        suffix="",
+    )
+    hubert_model = models[0]
+    hubert_model = hubert_model.to(config.device)
+    if config.is_half:
+        hubert_model = hubert_model.half()
+    else:
+        hubert_model = hubert_model.float()
+    return hubert_model.eval()
+
+def get_model_names():
+    model_root = "weights"  # Assuming this is where your models are stored
+    return [d for d in os.listdir(model_root) if os.path.isdir(f"{model_root}/{d}")]
+
+async def tts(
+    model_name,
+    tts_text,
+    tts_voice,
+    index_rate,
+    use_uploaded_voice,
+    uploaded_voice,
+):
+    speed = 0  # Default speech speed
+    f0_up_key = 0  # Default pitch adjustment
+    f0_method = "rmvpe"  # Default pitch extraction method
+    protect = 0.33  # Default protect value
+    filter_radius = 3
+    resample_sr = 0
+    rms_mix_rate = 0.25
+    edge_time = 0  # Initialize edge_time
+
+    edge_output_filename = get_unique_filename("mp3")
+
+    try:
+        if use_uploaded_voice:
+            if uploaded_voice is None:
+                return "No voice file uploaded.", None, None
+            
+            with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as tmp_file:
+                tmp_file.write(uploaded_voice)
+                uploaded_file_path = tmp_file.name
+
+            audio, sr = librosa.load(uploaded_file_path, sr=16000, mono=True)
+        else:
+            if limitation and len(tts_text) > 4000:
+                return (
+                    f"Text characters should be at most 280 in this huggingface space, but got {len(tts_text)} characters.",
+                    None,
+                    None,
+                )
+            
+            t0 = time.time()
+            speed_str = f"+{speed}%" if speed >= 0 else f"{speed}%"
+            await edge_tts.Communicate(
+                tts_text, tts_voice, rate=speed_str
+            ).save(edge_output_filename)
+            t1 = time.time()
+            edge_time = t1 - t0
+
+            audio, sr = librosa.load(edge_output_filename, sr=16000, mono=True)
+
+        duration = len(audio) / sr
+        print(f"Audio duration: {duration}s")
+        if limitation and duration >= 20:
+            return (
+                f"Audio should be less than 20 seconds in this huggingface space, but got {duration}s.",
+                None,
+                None,
+            )
+
+        f0_up_key = int(f0_up_key)
+        tgt_sr, net_g, vc, version, index_file, if_f0 = model_data(model_name)
+
+        if f0_method == "rmvpe":
+            vc.model_rmvpe = rmvpe_model
+
+        times = [0, 0, 0]
+        audio_opt = vc.pipeline(
+            hubert_model,
+            net_g,
+            0,
+            audio,
+            edge_output_filename if not use_uploaded_voice else uploaded_file_path,
+            times,
+            f0_up_key,
+            f0_method,
+            index_file,
+            index_rate,
+            if_f0,
+            filter_radius,
+            tgt_sr,
+            resample_sr,
+            rms_mix_rate,
+            version,
+            protect,
+            None,
+        )
+
+        if tgt_sr != resample_sr and resample_sr >= 16000:
+            tgt_sr = resample_sr
+        
+        info = f"Success. Time: tts: {edge_time}s, npy: {times[0]}s, f0: {times[1]}s, infer: {times[2]}s"
+        print(info)
+        return (
+            info,
+            edge_output_filename if not use_uploaded_voice else None,   
+            (tgt_sr, audio_opt),
+            edge_output_filename  
+        )
+
+    except EOFError:
+        info = (
+            "output not valid. This may occur when input text and speaker do not match."
+        )
+        print(info)
+        return info, None, None
+    except Exception as e:
+        traceback_info = traceback.format_exc()
+        print(traceback_info)
+        return str(e), None, None
+
+voice_mapping = {
+    "Mongolian Male": "mn-MN-BataaNeural",
+    "Mongolian Female": "mn-MN-YesuiNeural"
+}
+
+hubert_model = load_hubert()
+rmvpe_model = RMVPE("rmvpe.pt", config.is_half, config.device)
+
+