spock74 vumichien commited on
Commit
5a669ff
0 Parent(s):

Duplicate from vumichien/whisper-speaker-diarization

Browse files

Co-authored-by: vumichien <vumichien@users.noreply.huggingface.co>

Files changed (7) hide show
  1. .gitattributes +34 -0
  2. README.md +15 -0
  3. app.py +416 -0
  4. packages.txt +1 -0
  5. requirements.txt +20 -0
  6. sample1.wav +0 -0
  7. sample2.wav +0 -0
.gitattributes ADDED
@@ -0,0 +1,34 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ *.7z filter=lfs diff=lfs merge=lfs -text
2
+ *.arrow filter=lfs diff=lfs merge=lfs -text
3
+ *.bin filter=lfs diff=lfs merge=lfs -text
4
+ *.bz2 filter=lfs diff=lfs merge=lfs -text
5
+ *.ckpt filter=lfs diff=lfs merge=lfs -text
6
+ *.ftz filter=lfs diff=lfs merge=lfs -text
7
+ *.gz filter=lfs diff=lfs merge=lfs -text
8
+ *.h5 filter=lfs diff=lfs merge=lfs -text
9
+ *.joblib filter=lfs diff=lfs merge=lfs -text
10
+ *.lfs.* filter=lfs diff=lfs merge=lfs -text
11
+ *.mlmodel filter=lfs diff=lfs merge=lfs -text
12
+ *.model filter=lfs diff=lfs merge=lfs -text
13
+ *.msgpack filter=lfs diff=lfs merge=lfs -text
14
+ *.npy filter=lfs diff=lfs merge=lfs -text
15
+ *.npz filter=lfs diff=lfs merge=lfs -text
16
+ *.onnx filter=lfs diff=lfs merge=lfs -text
17
+ *.ot filter=lfs diff=lfs merge=lfs -text
18
+ *.parquet filter=lfs diff=lfs merge=lfs -text
19
+ *.pb filter=lfs diff=lfs merge=lfs -text
20
+ *.pickle filter=lfs diff=lfs merge=lfs -text
21
+ *.pkl filter=lfs diff=lfs merge=lfs -text
22
+ *.pt filter=lfs diff=lfs merge=lfs -text
23
+ *.pth filter=lfs diff=lfs merge=lfs -text
24
+ *.rar filter=lfs diff=lfs merge=lfs -text
25
+ *.safetensors filter=lfs diff=lfs merge=lfs -text
26
+ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
27
+ *.tar.* filter=lfs diff=lfs merge=lfs -text
28
+ *.tflite filter=lfs diff=lfs merge=lfs -text
29
+ *.tgz filter=lfs diff=lfs merge=lfs -text
30
+ *.wasm filter=lfs diff=lfs merge=lfs -text
31
+ *.xz filter=lfs diff=lfs merge=lfs -text
32
+ *.zip filter=lfs diff=lfs merge=lfs -text
33
+ *.zst filter=lfs diff=lfs merge=lfs -text
34
+ *tfevents* filter=lfs diff=lfs merge=lfs -text
README.md ADDED
@@ -0,0 +1,15 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ title: Whisper Speaker Diarization
3
+ emoji: 🎎
4
+ colorFrom: blue
5
+ colorTo: red
6
+ sdk: gradio
7
+ sdk_version: 3.9.1
8
+ app_file: app.py
9
+ pinned: false
10
+ tags:
11
+ - whisper-event
12
+ duplicated_from: vumichien/whisper-speaker-diarization
13
+ ---
14
+
15
+ Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
app.py ADDED
@@ -0,0 +1,416 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import whisper
2
+ import datetime
3
+ import subprocess
4
+ import gradio as gr
5
+ from pathlib import Path
6
+ import pandas as pd
7
+ import re
8
+ import time
9
+ import os
10
+ import numpy as np
11
+ from sklearn.cluster import AgglomerativeClustering
12
+
13
+ from pytube import YouTube
14
+ import torch
15
+ import pyannote.audio
16
+ from pyannote.audio.pipelines.speaker_verification import PretrainedSpeakerEmbedding
17
+ from pyannote.audio import Audio
18
+ from pyannote.core import Segment
19
+
20
+ from gpuinfo import GPUInfo
21
+
22
+ import wave
23
+ import contextlib
24
+ from transformers import pipeline
25
+ import psutil
26
+
27
+ whisper_models = ["base", "small", "medium", "large"]
28
+ source_languages = {
29
+ "en": "English",
30
+ "zh": "Chinese",
31
+ "de": "German",
32
+ "es": "Spanish",
33
+ "ru": "Russian",
34
+ "ko": "Korean",
35
+ "fr": "French",
36
+ "ja": "Japanese",
37
+ "pt": "Portuguese",
38
+ "tr": "Turkish",
39
+ "pl": "Polish",
40
+ "ca": "Catalan",
41
+ "nl": "Dutch",
42
+ "ar": "Arabic",
43
+ "sv": "Swedish",
44
+ "it": "Italian",
45
+ "id": "Indonesian",
46
+ "hi": "Hindi",
47
+ "fi": "Finnish",
48
+ "vi": "Vietnamese",
49
+ "he": "Hebrew",
50
+ "uk": "Ukrainian",
51
+ "el": "Greek",
52
+ "ms": "Malay",
53
+ "cs": "Czech",
54
+ "ro": "Romanian",
55
+ "da": "Danish",
56
+ "hu": "Hungarian",
57
+ "ta": "Tamil",
58
+ "no": "Norwegian",
59
+ "th": "Thai",
60
+ "ur": "Urdu",
61
+ "hr": "Croatian",
62
+ "bg": "Bulgarian",
63
+ "lt": "Lithuanian",
64
+ "la": "Latin",
65
+ "mi": "Maori",
66
+ "ml": "Malayalam",
67
+ "cy": "Welsh",
68
+ "sk": "Slovak",
69
+ "te": "Telugu",
70
+ "fa": "Persian",
71
+ "lv": "Latvian",
72
+ "bn": "Bengali",
73
+ "sr": "Serbian",
74
+ "az": "Azerbaijani",
75
+ "sl": "Slovenian",
76
+ "kn": "Kannada",
77
+ "et": "Estonian",
78
+ "mk": "Macedonian",
79
+ "br": "Breton",
80
+ "eu": "Basque",
81
+ "is": "Icelandic",
82
+ "hy": "Armenian",
83
+ "ne": "Nepali",
84
+ "mn": "Mongolian",
85
+ "bs": "Bosnian",
86
+ "kk": "Kazakh",
87
+ "sq": "Albanian",
88
+ "sw": "Swahili",
89
+ "gl": "Galician",
90
+ "mr": "Marathi",
91
+ "pa": "Punjabi",
92
+ "si": "Sinhala",
93
+ "km": "Khmer",
94
+ "sn": "Shona",
95
+ "yo": "Yoruba",
96
+ "so": "Somali",
97
+ "af": "Afrikaans",
98
+ "oc": "Occitan",
99
+ "ka": "Georgian",
100
+ "be": "Belarusian",
101
+ "tg": "Tajik",
102
+ "sd": "Sindhi",
103
+ "gu": "Gujarati",
104
+ "am": "Amharic",
105
+ "yi": "Yiddish",
106
+ "lo": "Lao",
107
+ "uz": "Uzbek",
108
+ "fo": "Faroese",
109
+ "ht": "Haitian creole",
110
+ "ps": "Pashto",
111
+ "tk": "Turkmen",
112
+ "nn": "Nynorsk",
113
+ "mt": "Maltese",
114
+ "sa": "Sanskrit",
115
+ "lb": "Luxembourgish",
116
+ "my": "Myanmar",
117
+ "bo": "Tibetan",
118
+ "tl": "Tagalog",
119
+ "mg": "Malagasy",
120
+ "as": "Assamese",
121
+ "tt": "Tatar",
122
+ "haw": "Hawaiian",
123
+ "ln": "Lingala",
124
+ "ha": "Hausa",
125
+ "ba": "Bashkir",
126
+ "jw": "Javanese",
127
+ "su": "Sundanese",
128
+ }
129
+
130
+ source_language_list = [key[0] for key in source_languages.items()]
131
+
132
+ MODEL_NAME = "vumichien/whisper-medium-jp"
133
+ lang = "ja"
134
+
135
+ device = 0 if torch.cuda.is_available() else "cpu"
136
+ pipe = pipeline(
137
+ task="automatic-speech-recognition",
138
+ model=MODEL_NAME,
139
+ chunk_length_s=30,
140
+ device=device,
141
+ )
142
+
143
+ pipe.model.config.forced_decoder_ids = pipe.tokenizer.get_decoder_prompt_ids(language=lang, task="transcribe")
144
+
145
+ embedding_model = PretrainedSpeakerEmbedding(
146
+ "speechbrain/spkrec-ecapa-voxceleb",
147
+ device=torch.device("cuda" if torch.cuda.is_available() else "cpu"))
148
+
149
+ def transcribe(microphone, file_upload):
150
+ warn_output = ""
151
+ if (microphone is not None) and (file_upload is not None):
152
+ warn_output = (
153
+ "WARNING: You've uploaded an audio file and used the microphone. "
154
+ "The recorded file from the microphone will be used and the uploaded audio will be discarded.\n"
155
+ )
156
+
157
+ elif (microphone is None) and (file_upload is None):
158
+ return "ERROR: You have to either use the microphone or upload an audio file"
159
+
160
+ file = microphone if microphone is not None else file_upload
161
+
162
+ text = pipe(file)["text"]
163
+
164
+ return warn_output + text
165
+
166
+ def _return_yt_html_embed(yt_url):
167
+ video_id = yt_url.split("?v=")[-1]
168
+ HTML_str = (
169
+ f'<center> <iframe width="500" height="320" src="https://www.youtube.com/embed/{video_id}"> </iframe>'
170
+ " </center>"
171
+ )
172
+ return HTML_str
173
+
174
+ def yt_transcribe(yt_url):
175
+ yt = YouTube(yt_url)
176
+ html_embed_str = _return_yt_html_embed(yt_url)
177
+ stream = yt.streams.filter(only_audio=True)[0]
178
+ stream.download(filename="audio.mp3")
179
+
180
+ text = pipe("audio.mp3")["text"]
181
+
182
+ return html_embed_str, text
183
+
184
+ def convert_time(secs):
185
+ return datetime.timedelta(seconds=round(secs))
186
+
187
+ def get_youtube(video_url):
188
+ yt = YouTube(video_url)
189
+ abs_video_path = yt.streams.filter(progressive=True, file_extension='mp4').order_by('resolution').desc().first().download()
190
+ print("Success download video")
191
+ print(abs_video_path)
192
+ return abs_video_path
193
+
194
+ def speech_to_text(video_file_path, selected_source_lang, whisper_model, num_speakers):
195
+ """
196
+ # Transcribe youtube link using OpenAI Whisper
197
+ 1. Using Open AI's Whisper model to seperate audio into segments and generate transcripts.
198
+ 2. Generating speaker embeddings for each segments.
199
+ 3. Applying agglomerative clustering on the embeddings to identify the speaker for each segment.
200
+
201
+ Speech Recognition is based on models from OpenAI Whisper https://github.com/openai/whisper
202
+ Speaker diarization model and pipeline from by https://github.com/pyannote/pyannote-audio
203
+ """
204
+
205
+ model = whisper.load_model(whisper_model)
206
+ time_start = time.time()
207
+ if(video_file_path == None):
208
+ raise ValueError("Error no video input")
209
+ print(video_file_path)
210
+
211
+ try:
212
+ # Read and convert youtube video
213
+ _,file_ending = os.path.splitext(f'{video_file_path}')
214
+ print(f'file enging is {file_ending}')
215
+ audio_file = video_file_path.replace(file_ending, ".wav")
216
+ print("starting conversion to wav")
217
+ os.system(f'ffmpeg -i "{video_file_path}" -ar 16000 -ac 1 -c:a pcm_s16le "{audio_file}"')
218
+
219
+ # Get duration
220
+ with contextlib.closing(wave.open(audio_file,'r')) as f:
221
+ frames = f.getnframes()
222
+ rate = f.getframerate()
223
+ duration = frames / float(rate)
224
+ print(f"conversion to wav ready, duration of audio file: {duration}")
225
+
226
+ # Transcribe audio
227
+ options = dict(language=selected_source_lang, beam_size=5, best_of=5)
228
+ transcribe_options = dict(task="transcribe", **options)
229
+ result = model.transcribe(audio_file, **transcribe_options)
230
+ segments = result["segments"]
231
+ print("starting whisper done with whisper")
232
+ except Exception as e:
233
+ raise RuntimeError("Error converting video to audio")
234
+
235
+ try:
236
+ # Create embedding
237
+ def segment_embedding(segment):
238
+ audio = Audio()
239
+ start = segment["start"]
240
+ # Whisper overshoots the end timestamp in the last segment
241
+ end = min(duration, segment["end"])
242
+ clip = Segment(start, end)
243
+ waveform, sample_rate = audio.crop(audio_file, clip)
244
+ return embedding_model(waveform[None])
245
+
246
+ embeddings = np.zeros(shape=(len(segments), 192))
247
+ for i, segment in enumerate(segments):
248
+ embeddings[i] = segment_embedding(segment)
249
+ embeddings = np.nan_to_num(embeddings)
250
+ print(f'Embedding shape: {embeddings.shape}')
251
+
252
+ # Assign speaker label
253
+ clustering = AgglomerativeClustering(num_speakers).fit(embeddings)
254
+ labels = clustering.labels_
255
+ for i in range(len(segments)):
256
+ segments[i]["speaker"] = 'SPEAKER ' + str(labels[i] + 1)
257
+
258
+ # Make output
259
+ objects = {
260
+ 'Start' : [],
261
+ 'End': [],
262
+ 'Speaker': [],
263
+ 'Text': []
264
+ }
265
+ text = ''
266
+ for (i, segment) in enumerate(segments):
267
+ if i == 0 or segments[i - 1]["speaker"] != segment["speaker"]:
268
+ objects['Start'].append(str(convert_time(segment["start"])))
269
+ objects['Speaker'].append(segment["speaker"])
270
+ if i != 0:
271
+ objects['End'].append(str(convert_time(segments[i - 1]["end"])))
272
+ objects['Text'].append(text)
273
+ text = ''
274
+ text += segment["text"] + ' '
275
+ objects['End'].append(str(convert_time(segments[i - 1]["end"])))
276
+ objects['Text'].append(text)
277
+
278
+ time_end = time.time()
279
+ time_diff = time_end - time_start
280
+ memory = psutil.virtual_memory()
281
+ gpu_utilization, gpu_memory = GPUInfo.gpu_usage()
282
+ gpu_utilization = gpu_utilization[0] if len(gpu_utilization) > 0 else 0
283
+ gpu_memory = gpu_memory[0] if len(gpu_memory) > 0 else 0
284
+ system_info = f"""
285
+ *Memory: {memory.total / (1024 * 1024 * 1024):.2f}GB, used: {memory.percent}%, available: {memory.available / (1024 * 1024 * 1024):.2f}GB.*
286
+ *Processing time: {time_diff:.5} seconds.*
287
+ *GPU Utilization: {gpu_utilization}%, GPU Memory: {gpu_memory}MiB.*
288
+ """
289
+
290
+ return pd.DataFrame(objects), system_info
291
+
292
+ except Exception as e:
293
+ raise RuntimeError("Error Running inference with local model", e)
294
+
295
+
296
+ # ---- Gradio Layout -----
297
+ # Inspiration from https://huggingface.co/spaces/RASMUS/Whisper-youtube-crosslingual-subtitles
298
+ video_in = gr.Video(label="Video file", mirror_webcam=False)
299
+ youtube_url_in = gr.Textbox(label="Youtube url", lines=1, interactive=True)
300
+ df_init = pd.DataFrame(columns=['Start', 'End', 'Speaker', 'Text'])
301
+ memory = psutil.virtual_memory()
302
+ selected_source_lang = gr.Dropdown(choices=source_language_list, type="value", value="en", label="Spoken language in video", interactive=True)
303
+ selected_whisper_model = gr.Dropdown(choices=whisper_models, type="value", value="base", label="Selected Whisper model", interactive=True)
304
+ number_speakers = gr.Number(precision=0, value=2, label="Selected number of speakers", interactive=True)
305
+ system_info = gr.Markdown(f"*Memory: {memory.total / (1024 * 1024 * 1024):.2f}GB, used: {memory.percent}%, available: {memory.available / (1024 * 1024 * 1024):.2f}GB*")
306
+ transcription_df = gr.DataFrame(value=df_init,label="Transcription dataframe", row_count=(0, "dynamic"), max_rows = 10, wrap=True, overflow_row_behaviour='paginate')
307
+ title = "Whisper speaker diarization"
308
+ demo = gr.Blocks(title=title)
309
+ demo.encrypt = False
310
+
311
+
312
+ with demo:
313
+ with gr.Tab("Whisper speaker diarization"):
314
+ gr.Markdown('''
315
+ <div>
316
+ <h1 style='text-align: center'>Whisper speaker diarization</h1>
317
+ This space uses Whisper models from <a href='https://github.com/openai/whisper' target='_blank'><b>OpenAI</b></a> to recoginze the speech and ECAPA-TDNN model from <a href='https://github.com/speechbrain/speechbrain' target='_blank'><b>SpeechBrain</b></a> to encode and clasify speakers</h2>
318
+ </div>
319
+ ''')
320
+
321
+ with gr.Row():
322
+ gr.Markdown('''
323
+ ### Transcribe youtube link using OpenAI Whisper
324
+ ##### 1. Using Open AI's Whisper model to seperate audio into segments and generate transcripts.
325
+ ##### 2. Generating speaker embeddings for each segments.
326
+ ##### 3. Applying agglomerative clustering on the embeddings to identify the speaker for each segment.
327
+ ''')
328
+
329
+ with gr.Row():
330
+ gr.Markdown('''
331
+ ### You can test by following examples:
332
+ ''')
333
+ examples = gr.Examples(examples=
334
+ [ "https://www.youtube.com/watch?v=j7BfEzAFuYc&t=32s",
335
+ "https://www.youtube.com/watch?v=-UX0X45sYe4",
336
+ "https://www.youtube.com/watch?v=7minSgqi-Gw"],
337
+ label="Examples", inputs=[youtube_url_in])
338
+
339
+
340
+ with gr.Row():
341
+ with gr.Column():
342
+ youtube_url_in.render()
343
+ download_youtube_btn = gr.Button("Download Youtube video")
344
+ download_youtube_btn.click(get_youtube, [youtube_url_in], [
345
+ video_in])
346
+ print(video_in)
347
+
348
+
349
+ with gr.Row():
350
+ with gr.Column():
351
+ video_in.render()
352
+ with gr.Column():
353
+ gr.Markdown('''
354
+ ##### Here you can start the transcription process.
355
+ ##### Please select the source language for transcription.
356
+ ##### You should select a number of speakers for getting better results.
357
+ ''')
358
+ selected_source_lang.render()
359
+ selected_whisper_model.render()
360
+ number_speakers.render()
361
+ transcribe_btn = gr.Button("Transcribe audio and diarization")
362
+ transcribe_btn.click(speech_to_text, [video_in, selected_source_lang, selected_whisper_model, number_speakers], [transcription_df, system_info])
363
+
364
+
365
+ with gr.Row():
366
+ gr.Markdown('''
367
+ ##### Here you will get transcription output
368
+ ##### ''')
369
+
370
+
371
+ with gr.Row():
372
+ with gr.Column():
373
+ transcription_df.render()
374
+ system_info.render()
375
+ gr.Markdown('''<center><img src='https://visitor-badge.glitch.me/badge?page_id=WhisperDiarizationSpeakers' alt='visitor badge'><a href="https://opensource.org/licenses/Apache-2.0"><img src='https://img.shields.io/badge/License-Apache_2.0-blue.svg' alt='License: Apache 2.0'></center>''')
376
+
377
+
378
+
379
+ with gr.Tab("Whisper Transcribe Japanese Audio"):
380
+ gr.Markdown(f'''
381
+ <div>
382
+ <h1 style='text-align: center'>Whisper Transcribe Japanese Audio</h1>
383
+ </div>
384
+ Transcribe long-form microphone or audio inputs with the click of a button! The fine-tuned
385
+ checkpoint <a href='https://huggingface.co/{MODEL_NAME}' target='_blank'><b>{MODEL_NAME}</b></a> to transcribe audio files of arbitrary length.
386
+ ''')
387
+ microphone = gr.inputs.Audio(source="microphone", type="filepath", optional=True)
388
+ upload = gr.inputs.Audio(source="upload", type="filepath", optional=True)
389
+ transcribe_btn = gr.Button("Transcribe Audio")
390
+ text_output = gr.Textbox()
391
+ with gr.Row():
392
+ gr.Markdown('''
393
+ ### You can test by following examples:
394
+ ''')
395
+ examples = gr.Examples(examples=
396
+ [ "sample1.wav",
397
+ "sample2.wav",
398
+ ],
399
+ label="Examples", inputs=[upload])
400
+ transcribe_btn.click(transcribe, [microphone, upload], outputs=text_output)
401
+
402
+ with gr.Tab("Whisper Transcribe Japanese YouTube"):
403
+ gr.Markdown(f'''
404
+ <div>
405
+ <h1 style='text-align: center'>Whisper Transcribe Japanese YouTube</h1>
406
+ </div>
407
+ Transcribe long-form YouTube videos with the click of a button! The fine-tuned checkpoint:
408
+ <a href='https://huggingface.co/{MODEL_NAME}' target='_blank'><b>{MODEL_NAME}</b></a> to transcribe audio files of arbitrary length.
409
+ ''')
410
+ youtube_link = gr.Textbox(label="Youtube url", lines=1, interactive=True)
411
+ yt_transcribe_btn = gr.Button("Transcribe YouTube")
412
+ text_output2 = gr.Textbox()
413
+ html_output = gr.Markdown()
414
+ yt_transcribe_btn.click(yt_transcribe, [youtube_link], outputs=[html_output, text_output2])
415
+
416
+ demo.launch(debug=True)
packages.txt ADDED
@@ -0,0 +1 @@
 
 
1
+ ffmpeg
requirements.txt ADDED
@@ -0,0 +1,20 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ git+https://github.com/huggingface/transformers
2
+ git+https://github.com/pyannote/pyannote-audio
3
+ git+https://github.com/openai/whisper.git
4
+ gradio==3.12
5
+ ffmpeg-python
6
+ pandas==1.5.0
7
+ pytube==12.1.0
8
+ sacremoses
9
+ sentencepiece
10
+ tokenizers
11
+ torch
12
+ torchaudio
13
+ tqdm==4.64.1
14
+ EasyNMT==2.0.2
15
+ nltk
16
+ transformers
17
+ pysrt
18
+ psutil==5.9.2
19
+ requests
20
+ gpuinfo
sample1.wav ADDED
Binary file (306 kB). View file
 
sample2.wav ADDED
Binary file (470 kB). View file