File size: 13,370 Bytes
7ed286f a6a5509 7ed286f b710448 7ed286f 07903ab 24020bb 07903ab 24020bb 7ed286f 07903ab 7ed286f 07903ab 7ed286f 07903ab 7ed286f 24020bb 7ed286f 07903ab 7ed286f 07903ab 7ed286f d486a74 24020bb 7ed286f 07903ab 7ed286f 07903ab 7ed286f 07903ab 7ed286f 07903ab 7ed286f 07903ab 7ed286f 07903ab 7ed286f 07903ab 7ed286f 07903ab 7ed286f 07903ab 7ed286f 07903ab 7ed286f 07903ab |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 |
import gradio as gr
from pathlib import Path
import time
import pandas as pd
import re
import time
import os
import requests
import json
import whisper
from pytube import YouTube
import psutil
import torch
# is cuda available?
num_cores = psutil.cpu_count()
os.environ["OMP_NUM_THREADS"] = f"{num_cores}"
headers = {'Authorization': os.environ['DeepL_API_KEY']}
device = "cpu"#torch.device("cuda" if torch.cuda.is_available() else "cpu")
print("DEVICE IS: ")
print(device)
asr_model_base = whisper.load_model("base", device=device)
asr_model_small = whisper.load_model("small", device=device)
whisper_models_dict = {
'base': asr_model_base,
'small': asr_model_small
}
whisper_models = ["base", "small"]
transcribe_options = dict(beam_size=3, best_of=3, without_timestamps=False)
source_languages = {
"Afrikaans":"af",
"Amharic":"am",
"Arabic":"ar",
"Asturian ":"st",
"Azerbaijani":"az",
"Bashkir":"ba",
"Belarusian":"be",
"Bulgarian":"bg",
"Bengali":"bn",
"Breton":"br",
"Bosnian":"bs",
"Catalan; Valencian":"ca",
"Cebuano":"eb",
"Czech":"cs",
"Welsh":"cy",
"Danish":"da",
"German":"de",
"Greeek":"el",
"English":"en",
"Spanish":"es",
"Estonian":"et",
"Persian":"fa",
"Fulah":"ff",
"Finnish":"fi",
"French":"fr",
"Western Frisian":"fy",
"Irish":"ga",
"Gaelic; Scottish Gaelic":"gd",
"Galician":"gl",
"Gujarati":"gu",
"Hausa":"ha",
"Hebrew":"he",
"Hindi":"hi",
"Croatian":"hr",
"Haitian; Haitian Creole":"ht",
"Hungarian":"hu",
"Armenian":"hy",
"Indonesian":"id",
"Igbo":"ig",
"Iloko":"lo",
"Icelandic":"is",
"Italian":"it",
"Japanese":"ja",
"Javanese":"jv",
"Georgian":"ka",
"Kazakh":"kk",
"Central Khmer":"km",
"Kannada":"kn",
"Korean":"ko",
"Luxembourgish; Letzeburgesch":"lb",
"Ganda":"lg",
"Lingala":"ln",
"Lao":"lo",
"Lithuanian":"lt",
"Latvian":"lv",
"Malagasy":"mg",
"Macedonian":"mk",
"Malayalam":"ml",
"Mongolian":"mn",
"Marathi":"mr",
"Malay":"ms",
"Burmese":"my",
"Nepali":"ne",
"Dutch; Flemish":"nl",
"Norwegian":"no",
"Northern Sotho":"ns",
"Occitan (post 1500)":"oc",
"Oriya":"or",
"Panjabi; Punjabi":"pa",
"Polish":"pl",
"Pushto; Pashto":"ps",
"Portuguese":"pt",
"Romanian; Moldavian; Moldovan":"ro",
"Russian":"ru",
"Sindhi":"sd",
"Sinhala; Sinhalese":"si",
"Slovak":"sk",
"Slovenian":"sl",
"Somali":"so",
"Albanian":"sq",
"Serbian":"sr",
"Swati":"ss",
"Sundanese":"su",
"Swedish":"sv",
"Swahili":"sw",
"Tamil":"ta",
"Thai":"th",
"Tagalog":"tl",
"Tswana":"tn",
"Turkish":"tr",
"Ukrainian":"uk",
"Urdu":"ur",
"Uzbek":"uz",
"Vietnamese":"vi",
"Wolof":"wo",
"Xhosa":"xh",
"Yiddish":"yi",
"Yoruba":"yo",
"Chinese":"zh",
"Zulu":"zu",
"Let the model analyze": "Let the model analyze"
}
DeepL_language_codes_for_translation = {
"Bulgarian": "BG",
"Czech": "CS",
"Danish": "DA",
"German": "DE",
"Greek": "EL",
"English": "EN",
"Spanish": "ES",
"Estonian": "ET",
"Finnish": "FI",
"French": "FR",
"Hungarian": "HU",
"Indonesian": "ID",
"Italian": "IT",
"Japanese": "JA",
"Lithuanian": "LT",
"Latvian": "LV",
"Dutch": "NL",
"Polish": "PL",
"Portuguese": "PT",
"Romanian": "RO",
"Russian": "RU",
"Slovak": "SK",
"Slovenian": "SL",
"Swedish": "SV",
"Turkish": "TR",
"Ukrainian": "UK",
"Chinese": "ZH"
}
source_language_list = [key[0] for key in source_languages.items()]
translation_models_list = [key[0] for key in DeepL_language_codes_for_translation.items()]
videos_out_path = Path("./videos_out")
videos_out_path.mkdir(parents=True, exist_ok=True)
def get_youtube(video_url):
yt = YouTube(video_url)
abs_video_path = yt.streams.filter(progressive=True, file_extension='mp4').order_by('resolution').desc().first().download()
print("LADATATTU POLKUUN")
print(abs_video_path)
return abs_video_path
def speech_to_text(video_file_path, selected_translation_lang, whisper_model):
"""
# Youtube with translated subtitles using OpenAI Whisper and Opus-MT models.
# Currently supports only English audio
This space allows you to:
1. Download youtube video with a given url
2. Watch it in the first video component
3. Run automatic speech recognition on the video using fast Whisper models
4. Translate the recognized transcriptions to 26 languages supported by deepL (If source language not supported this will return original transciption)
5. Burn the translations to the original video and watch the video in the 2nd video component
Speech Recognition is based on models from OpenAI Whisper https://github.com/openai/whisper
This space is using c++ implementation by https://github.com/ggerganov/whisper.cpp
"""
if(video_file_path == None):
raise ValueError("Error no video input")
print(video_file_path)
try:
audio = whisper.load_audio(video_file_path)
except Exception as e:
raise RuntimeError("Error converting video to audio")
last_time = time.time()
try:
print(f'Transcribing via local model')
transcribe_options = dict(beam_size=3, best_of=3, without_timestamps=False)
transcription = whisper_models_dict.get(whisper_model).transcribe(audio, **transcribe_options)
df = pd.DataFrame(columns=['start','end','text'])
for i,segment in enumerate(transcription['segments']):
new_row = {'start': segment['start'],
'end': segment['end'],
'text': segment['text']
}
df = df.append(new_row, ignore_index=True)
return (df)
except Exception as e:
raise RuntimeError("Error Running inference with local model", e)
def translate_transcriptions(df, selected_translation_lang_2):
if selected_translation_lang_2 is None:
selected_translation_lang_2 = 'English'
df.reset_index(inplace=True)
print("start_translation")
translations = []
text_combined = ""
for i, sentence in enumerate(df['text']):
if i == 0:
text_combined = sentence
else:
text_combined = text_combined + '\n' + sentence
data = {'text': text_combined,
'tag_spitting': 'xml',
'target_lang': DeepL_language_codes_for_translation.get(selected_translation_lang_2)
}
try:
response = requests.post('https://api-free.deepl.com/v2/translate', headers=headers, data=data)
# Print the response from the server
translated_sentences = json.loads(response.text)
translated_sentences = translated_sentences['translations'][0]['text'].split('\n')
df['translation'] = translated_sentences
except Exception as e:
print(e)
df['translation'] = df['text']
print("translations done")
return df
def create_srt_and_burn(df, video_in):
print("Starting creation of video wit srt")
with open('testi.srt','w', encoding="utf-8") as file:
for i in range(len(df)):
file.write(str(i+1))
file.write('\n')
start = df.iloc[i]['start']
milliseconds = round(start * 1000.0)
hours = milliseconds // 3_600_000
milliseconds -= hours * 3_600_000
minutes = milliseconds // 60_000
milliseconds -= minutes * 60_000
seconds = milliseconds // 1_000
milliseconds -= seconds * 1_000
file.write(f"{hours}:{minutes:02d}:{seconds:02d}.{milliseconds:03d}")
stop = df.iloc[i]['end']
milliseconds = round(stop * 1000.0)
hours = milliseconds // 3_600_000
milliseconds -= hours * 3_600_000
minutes = milliseconds // 60_000
milliseconds -= minutes * 60_000
seconds = milliseconds // 1_000
milliseconds -= seconds * 1_000
file.write(' --> ')
file.write(f"{hours}:{minutes:02d}:{seconds:02d}.{milliseconds:03d}")
file.write('\n')
file.writelines(df.iloc[i]['translation'])
if int(i) != len(df)-1:
file.write('\n\n')
print("SRT DONE")
try:
file1 = open('./testi.srt', 'r', encoding="utf-8")
Lines = file1.readlines()
count = 0
# Strips the newline character
for line in Lines:
count += 1
print("{}".format(line))
print(type(video_in))
print(video_in)
video_out = video_in.replace('.mp4', '_out.mp4')
print(video_out)
command = 'ffmpeg -i "{}" -y -vf subtitles=./testi.srt "{}"'.format(video_in, video_out)
print(command)
os.system(command)
return video_out
except Exception as e:
print(e)
return video_out
# ---- Gradio Layout -----
video_in = gr.Video(label="Video file", mirror_webcam=False)
youtube_url_in = gr.Textbox(label="Youtube url", lines=1, interactive=True)
video_out = gr.Video(label="Video Out", mirror_webcam=False)
df_init = pd.DataFrame(columns=['start','end','text','translation'])
selected_source_lang = gr.Dropdown(choices=source_language_list, type="value", value="Let the model analyze", label="Spoken language in video", interactive=True)
selected_translation_lang_2 = gr.Dropdown(choices=translation_models_list, type="value", value="English", label="In which language you want the transcriptions?", interactive=True)
selected_whisper_model = gr.Dropdown(choices=whisper_models, type="value", value="base", label="Selected Whisper model", interactive=True)
transcription_df = gr.DataFrame(value=df_init,label="Transcription dataframe", row_count=(0, "dynamic"), max_rows = 10, wrap=True, overflow_row_behaviour='paginate')
transcription_and_translation_df = gr.DataFrame(value=df_init,label="Transcription and translation dataframe", max_rows = 10, wrap=True, overflow_row_behaviour='paginate')
demo = gr.Blocks(css='''
#cut_btn, #reset_btn { align-self:stretch; }
#\\31 3 { max-width: 540px; }
.output-markdown {max-width: 65ch !important;}
''')
demo.encrypt = False
with demo:
transcription_var = gr.Variable()
with gr.Row():
with gr.Column():
gr.Markdown('''
### This space allows you to:
##### 1. Download youtube video with a given URL
##### 2. Watch it in the first video component
##### 3. Run automatic speech recognition on the video using Whisper
##### 4. Translate the recognized transcriptions to 26 languages supported by deepL
##### 5. Burn the translations to the original video and watch the video in the 2nd video component
''')
with gr.Column():
gr.Markdown('''
### 1. Insert Youtube URL below. Some test videos below:
##### 1. https://www.youtube.com/watch?v=nlMuHtV82q8&ab_channel=NothingforSale24
##### 2. https://www.youtube.com/watch?v=JzPfMbG1vrE&ab_channel=ExplainerVideosByLauren
##### 3. https://www.youtube.com/watch?v=S68vvV0kod8&ab_channel=Pearl-CohnTelevision
''')
with gr.Row():
with gr.Column():
youtube_url_in.render()
download_youtube_btn = gr.Button("Step 1. Download Youtube video")
download_youtube_btn.click(get_youtube, [youtube_url_in], [
video_in])
print(video_in)
with gr.Row():
with gr.Column():
video_in.render()
with gr.Column():
gr.Markdown('''
##### Here you can start the transcription and translation process.
##### Be aware that processing will last some time. With base model it is around 3x speed
''')
selected_source_lang.render()
selected_whisper_model.render()
transcribe_btn = gr.Button("Step 2. Transcribe audio")
transcribe_btn.click(speech_to_text, [video_in, selected_source_lang, selected_whisper_model], transcription_df)
with gr.Row():
gr.Markdown('''
##### Here you will get transcription output
##### ''')
with gr.Row():
with gr.Column():
transcription_df.render()
with gr.Row():
with gr.Column():
gr.Markdown('''
##### Here you will get translated transcriptions.
##### Please remember to select target language
##### ''')
selected_translation_lang_2.render()
translate_transcriptions_button = gr.Button("Step 3. Translate transcription")
translate_transcriptions_button.click(translate_transcriptions, [transcription_df, selected_translation_lang_2], transcription_and_translation_df)
transcription_and_translation_df.render()
with gr.Row():
with gr.Column():
gr.Markdown('''
##### Now press the Step 4. Button to create output video with translated transcriptions
##### ''')
translate_and_make_srt_btn = gr.Button("Step 4. Create and burn srt to video")
print(video_in)
translate_and_make_srt_btn.click(create_srt_and_burn, [transcription_and_translation_df,video_in], [
video_out])
video_out.render()
demo.launch() |