File size: 6,091 Bytes
071812d
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
a97e72d
071812d
 
 
 
 
 
 
a97e72d
071812d
a97e72d
071812d
 
 
 
 
a97e72d
071812d
 
 
 
a97e72d
 
 
 
 
 
 
 
 
 
 
 
071812d
 
 
 
 
a97e72d
071812d
 
 
 
 
a97e72d
071812d
 
 
 
 
 
 
 
 
 
 
a97e72d
 
 
 
 
071812d
 
 
 
 
 
 
a97e72d
 
 
 
071812d
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
a97e72d
 
 
 
 
 
 
 
 
 
 
071812d
 
a97e72d
 
 
071812d
 
a97e72d
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
071812d
 
 
 
a97e72d
071812d
 
 
 
 
 
a97e72d
071812d
 
a97e72d
071812d
 
 
 
 
 
588da9c
a97e72d
071812d
 
 
a97e72d
 
 
 
 
 
 
071812d
 
 
 
a97e72d
 
 
 
 
 
 
071812d
 
a97e72d
071812d
 
a97e72d
 
 
 
071812d
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
#!/usr/bin/env python3
#
# Copyright      2022  Xiaomi Corp.        (authors: Fangjun Kuang)
#
# See LICENSE for clarification regarding multiple authors
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

# References:
# https://gradio.app/docs/#dropdown

import logging
import os
import time
from datetime import datetime

import gradio as gr
import torchaudio

from model import get_pretrained_model, language_to_models, sample_rate

languages = sorted(language_to_models.keys())


def convert_to_wav(in_filename: str) -> str:
    """Convert the input audio file to a wave file"""
    out_filename = in_filename + ".wav"
    logging.info(f"Converting '{in_filename}' to '{out_filename}'")
    _ = os.system(f"ffmpeg -hide_banner -i '{in_filename}' '{out_filename}'")
    return out_filename


def process(
    in_filename: str,
    language: str,
    repo_id: str,
    decoding_method: str,
    num_active_paths: int,
) -> str:
    logging.info(f"in_filename: {in_filename}")
    logging.info(f"language: {language}")
    logging.info(f"repo_id: {repo_id}")
    logging.info(f"decoding_method: {decoding_method}")
    logging.info(f"num_active_paths: {num_active_paths}")

    filename = convert_to_wav(in_filename)

    now = datetime.now()
    date_time = now.strftime("%Y-%m-%d %H:%M:%S.%f")
    logging.info(f"Started at {date_time}")

    start = time.time()
    wave, wave_sample_rate = torchaudio.load(filename)

    if wave_sample_rate != sample_rate:
        logging.info(
            f"Expected sample rate: {sample_rate}. Given: {wave_sample_rate}. "
            f"Resampling to {sample_rate}."
        )

        wave = torchaudio.functional.resample(
            wave,
            orig_freq=wave_sample_rate,
            new_freq=sample_rate,
        )
    wave = wave[0]  # use only the first channel.

    hyp = get_pretrained_model(repo_id).decode_waves(
        [wave],
        decoding_method=decoding_method,
        num_active_paths=num_active_paths,
    )[0]

    date_time = now.strftime("%Y-%m-%d %H:%M:%S.%f")
    end = time.time()

    duration = wave.shape[0] / sample_rate
    rtf = (end - start) / duration

    logging.info(f"Finished at {date_time} s. Elapsed: {end - start: .3f} s")
    logging.info(f"Duration {duration: .3f} s")
    logging.info(f"RTF {rtf: .3f}")
    logging.info(f"hyp:\n{hyp}")

    return hyp


title = "# Automatic Speech Recognition with Next-gen Kaldi"
description = """
This space shows how to do automatic speech recognition with Next-gen Kaldi.

See more information by visiting the following links:

- <https://github.com/k2-fsa/icefall>
- <https://github.com/k2-fsa/sherpa>
- <https://github.com/k2-fsa/k2>
- <https://github.com/lhotse-speech/lhotse>
"""


def update_model_dropdown(language: str):
    if language in language_to_models:
        choices = language_to_models[language]
        return gr.Dropdown.update(choices=choices, value=choices[0])

    raise ValueError(f"Unsupported language: {language}")


demo = gr.Blocks()

with demo:
    gr.Markdown(title)
    language_choices = list(language_to_models.keys())

    language_radio = gr.Radio(
        label="Language",
        choices=language_choices,
        value=language_choices[0],
    )
    model_dropdown = gr.Dropdown(
        choices=language_to_models[language_choices[0]],
        label="Select a model",
        value=language_to_models[language_choices[0]][0],
    )

    language_radio.change(
        update_model_dropdown,
        inputs=language_radio,
        outputs=model_dropdown,
    )

    decoding_method_radio = gr.Radio(
        label="Decoding method",
        choices=["greedy_search", "modified_beam_search"],
        value="greedy_search",
    )

    num_active_paths_slider = gr.Slider(
        minimum=1,
        value=4,
        step=1,
        label="Number of active paths for modified_beam_search",
    )

    with gr.Tabs():
        with gr.TabItem("Upload from disk"):
            uploaded_file = gr.Audio(
                source="upload",  # Choose between "microphone", "upload"
                type="filepath",
                optional=False,
                label="Upload from disk",
            )
            upload_button = gr.Button("Submit for recognition")
            uploaded_output = gr.Textbox(label="Recognized speech from uploaded file")

        with gr.TabItem("Record from microphone"):
            microphone = gr.Audio(
                source="microphone",  # Choose between "microphone", "upload"
                type="filepath",
                optional=False,
                label="Record from microphone",
            )

            record_button = gr.Button("Submit for recognition")
            recorded_output = gr.Textbox(label="Recognized speech from recordings")

        upload_button.click(
            process,
            inputs=[
                uploaded_file,
                language_radio,
                model_dropdown,
                decoding_method_radio,
                num_active_paths_slider,
            ],
            outputs=uploaded_output,
        )
        record_button.click(
            process,
            inputs=[
                microphone,
                language_radio,
                model_dropdown,
                decoding_method_radio,
                num_active_paths_slider,
            ],
            outputs=recorded_output,
        )
    gr.Markdown(description)

if __name__ == "__main__":
    formatter = "%(asctime)s %(levelname)s [%(filename)s:%(lineno)d] %(message)s"

    logging.basicConfig(format=formatter, level=logging.INFO)

    demo.launch()