Spaces:
Running
Running
File size: 11,532 Bytes
785fa6f |
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 |
import os
import torch
from typing import List
import whisper
import gradio as gr
from abc import ABC, abstractmethod
from typing import BinaryIO, Union, Tuple, List
import numpy as np
from datetime import datetime
from modules.subtitle_manager import get_srt, get_vtt, get_txt, write_file, safe_filename
from modules.youtube_manager import get_ytdata, get_ytaudio
from modules.whisper_parameter import *
class WhisperBase(ABC):
def __init__(self,
model_dir: str):
self.model = None
self.current_model_size = None
self.model_dir = model_dir
os.makedirs(self.model_dir, exist_ok=True)
self.available_models = whisper.available_models()
self.available_langs = sorted(list(whisper.tokenizer.LANGUAGES.values()))
self.translatable_models = ["large", "large-v1", "large-v2", "large-v3"]
self.device = self.get_device()
self.available_compute_types = ["float16", "float32"]
self.current_compute_type = "float16" if self.device == "cuda" else "float32"
@abstractmethod
def transcribe(self,
audio: Union[str, BinaryIO, np.ndarray],
progress: gr.Progress,
*whisper_params,
):
pass
@abstractmethod
def update_model(self,
model_size: str,
compute_type: str,
progress: gr.Progress
):
pass
def transcribe_file(self,
files: list,
file_format: str,
add_timestamp: bool,
progress=gr.Progress(),
*whisper_params,
) -> list:
"""
Write subtitle file from Files
Parameters
----------
files: list
List of files to transcribe from gr.Files()
file_format: str
Subtitle File format to write from gr.Dropdown(). Supported format: [SRT, WebVTT, txt]
add_timestamp: bool
Boolean value from gr.Checkbox() that determines whether to add a timestamp at the end of the subtitle filename.
progress: gr.Progress
Indicator to show progress directly in gradio.
*whisper_params: tuple
Gradio components related to Whisper. see whisper_data_class.py for details.
Returns
----------
result_str:
Result of transcription to return to gr.Textbox()
result_file_path:
Output file path to return to gr.Files()
"""
try:
files_info = {}
for file in files:
transcribed_segments, time_for_task = self.transcribe(
file.name,
progress,
*whisper_params,
)
file_name, file_ext = os.path.splitext(os.path.basename(file.name))
file_name = safe_filename(file_name)
subtitle, file_path = self.generate_and_write_file(
file_name=file_name,
transcribed_segments=transcribed_segments,
add_timestamp=add_timestamp,
file_format=file_format
)
files_info[file_name] = {"subtitle": subtitle, "time_for_task": time_for_task, "path": file_path}
total_result = ''
total_time = 0
for file_name, info in files_info.items():
total_result += '------------------------------------\n'
total_result += f'{file_name}\n\n'
total_result += f'{info["subtitle"]}'
total_time += info["time_for_task"]
result_str = f"Done in {self.format_time(total_time)}! Subtitle is in the outputs folder.\n\n{total_result}"
result_file_path = [info['path'] for info in files_info.values()]
return [result_str, result_file_path]
except Exception as e:
print(f"Error transcribing file: {e}")
finally:
self.release_cuda_memory()
if not files:
self.remove_input_files([file.name for file in files])
def transcribe_mic(self,
mic_audio: str,
file_format: str,
progress=gr.Progress(),
*whisper_params,
) -> list:
"""
Write subtitle file from microphone
Parameters
----------
mic_audio: str
Audio file path from gr.Microphone()
file_format: str
Subtitle File format to write from gr.Dropdown(). Supported format: [SRT, WebVTT, txt]
progress: gr.Progress
Indicator to show progress directly in gradio.
*whisper_params: tuple
Gradio components related to Whisper. see whisper_data_class.py for details.
Returns
----------
result_str:
Result of transcription to return to gr.Textbox()
result_file_path:
Output file path to return to gr.Files()
"""
try:
progress(0, desc="Loading Audio..")
transcribed_segments, time_for_task = self.transcribe(
mic_audio,
progress,
*whisper_params,
)
progress(1, desc="Completed!")
subtitle, result_file_path = self.generate_and_write_file(
file_name="Mic",
transcribed_segments=transcribed_segments,
add_timestamp=True,
file_format=file_format
)
result_str = f"Done in {self.format_time(time_for_task)}! Subtitle file is in the outputs folder.\n\n{subtitle}"
return [result_str, result_file_path]
except Exception as e:
print(f"Error transcribing file: {e}")
finally:
self.release_cuda_memory()
self.remove_input_files([mic_audio])
def transcribe_youtube(self,
youtube_link: str,
file_format: str,
add_timestamp: bool,
progress=gr.Progress(),
*whisper_params,
) -> list:
"""
Write subtitle file from Youtube
Parameters
----------
youtube_link: str
URL of the Youtube video to transcribe from gr.Textbox()
file_format: str
Subtitle File format to write from gr.Dropdown(). Supported format: [SRT, WebVTT, txt]
add_timestamp: bool
Boolean value from gr.Checkbox() that determines whether to add a timestamp at the end of the filename.
progress: gr.Progress
Indicator to show progress directly in gradio.
*whisper_params: tuple
Gradio components related to Whisper. see whisper_data_class.py for details.
Returns
----------
result_str:
Result of transcription to return to gr.Textbox()
result_file_path:
Output file path to return to gr.Files()
"""
try:
progress(0, desc="Loading Audio from Youtube..")
yt = get_ytdata(youtube_link)
audio = get_ytaudio(yt)
transcribed_segments, time_for_task = self.transcribe(
audio,
progress,
*whisper_params,
)
progress(1, desc="Completed!")
file_name = safe_filename(yt.title)
subtitle, result_file_path = self.generate_and_write_file(
file_name=file_name,
transcribed_segments=transcribed_segments,
add_timestamp=add_timestamp,
file_format=file_format
)
result_str = f"Done in {self.format_time(time_for_task)}! Subtitle file is in the outputs folder.\n\n{subtitle}"
return [result_str, result_file_path]
except Exception as e:
print(f"Error transcribing file: {e}")
finally:
try:
if 'yt' not in locals():
yt = get_ytdata(youtube_link)
file_path = get_ytaudio(yt)
else:
file_path = get_ytaudio(yt)
self.release_cuda_memory()
self.remove_input_files([file_path])
except Exception as cleanup_error:
pass
@staticmethod
def generate_and_write_file(file_name: str,
transcribed_segments: list,
add_timestamp: bool,
file_format: str,
) -> str:
"""
Writes subtitle file
Parameters
----------
file_name: str
Output file name
transcribed_segments: list
Text segments transcribed from audio
add_timestamp: bool
Determines whether to add a timestamp to the end of the filename.
file_format: str
File format to write. Supported formats: [SRT, WebVTT, txt]
Returns
----------
content: str
Result of the transcription
output_path: str
output file path
"""
timestamp = datetime.now().strftime("%m%d%H%M%S")
if add_timestamp:
output_path = os.path.join("outputs", f"{file_name}-{timestamp}")
else:
output_path = os.path.join("outputs", f"{file_name}")
if file_format == "SRT":
content = get_srt(transcribed_segments)
output_path += '.srt'
write_file(content, output_path)
elif file_format == "WebVTT":
content = get_vtt(transcribed_segments)
output_path += '.vtt'
write_file(content, output_path)
elif file_format == "txt":
content = get_txt(transcribed_segments)
output_path += '.txt'
write_file(content, output_path)
return content, output_path
@staticmethod
def format_time(elapsed_time: float) -> str:
"""
Get {hours} {minutes} {seconds} time format string
Parameters
----------
elapsed_time: str
Elapsed time for transcription
Returns
----------
Time format string
"""
hours, rem = divmod(elapsed_time, 3600)
minutes, seconds = divmod(rem, 60)
time_str = ""
if hours:
time_str += f"{hours} hours "
if minutes:
time_str += f"{minutes} minutes "
seconds = round(seconds)
time_str += f"{seconds} seconds"
return time_str.strip()
@staticmethod
def get_device():
if torch.cuda.is_available():
return "cuda"
elif torch.backends.mps.is_available():
return "mps"
else:
return "cpu"
@staticmethod
def release_cuda_memory():
if torch.cuda.is_available():
torch.cuda.empty_cache()
torch.cuda.reset_max_memory_allocated()
@staticmethod
def remove_input_files(file_paths: List[str]):
if not file_paths:
return
for file_path in file_paths:
if file_path and os.path.exists(file_path):
os.remove(file_path)
|