dasemu/Transcriptarr
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Update subgen.py

Browse Source
This commit is contained in:
McCloudS
2024-03-22 09:08:43 -06:00
committed by GitHub
parent 5af06045c5
commit a844fe5966
1 changed files with 9 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files

18
subgen.py
View File

@@ -59,7 +59,7 @@ append = convert_to_bool(os.getenv('APPEND', False))
reload_script_on_change = convert_to_bool(os.getenv('RELOAD_SCRIPT_ON_CHANGE', False)) reload_script_on_change = convert_to_bool(os.getenv('RELOAD_SCRIPT_ON_CHANGE', False))
model_prompt = os.getenv('USE_MODEL_PROMPT', 'False') model_prompt = os.getenv('USE_MODEL_PROMPT', 'False')
custom_model_prompt = os.getenv('CUSTOM_MODEL_PROMPT', '') custom_model_prompt = os.getenv('CUSTOM_MODEL_PROMPT', '')
custom_parameters = os.getenv('CUSTOM_PARAMETERS', '') custom_regroup = os.getenv('CUSTOM_REGROUP', '')
if transcribe_device == "gpu": if transcribe_device == "gpu":
transcribe_device = "cuda" transcribe_device = "cuda"
@@ -337,10 +337,10 @@ def asr(
audio_data = np.frombuffer(audio_file.file.read(), np.int16).flatten().astype(np.float32) / 32768.0 audio_data = np.frombuffer(audio_file.file.read(), np.int16).flatten().astype(np.float32) / 32768.0
if model_prompt: if model_prompt:
custom_model_prompt = greetings_translations.get(language, '') or custom_model_prompt custom_model_prompt = greetings_translations.get(language, '') or custom_model_prompt
if custom_parameters: if custom_regroup:
# Convert the string to a dictionary result = model.transcribe_stable(audio_data, task=task, input_sr=16000, language=language, progress_callback=progress, initial_prompt=custom_model_prompt, regroup=custom_regroup)
params_dict = dict(param.split('=') for param in custom_parameters.split(', ')) else:
result = model.transcribe_stable(audio_data, task=task, input_sr=16000, language=language, progress_callback=progress, initial_prompt=custom_model_prompt, **params_dict) result = model.transcribe_stable(audio_data, task=task, input_sr=16000, language=language, progress_callback=progress, initial_prompt=custom_model_prompt)
appendLine(result) appendLine(result)
elapsed_time = time.time() - start_time elapsed_time = time.time() - start_time
minutes, seconds = divmod(int(elapsed_time), 60) minutes, seconds = divmod(int(elapsed_time), 60)
@@ -440,10 +440,10 @@ def gen_subtitles(file_path: str, transcribe_or_translate: str, front=True, forc
if force_detected_language_to: if force_detected_language_to:
forceLanguage = force_detected_language_to forceLanguage = force_detected_language_to
logging.info(f"Forcing language to {forceLanguage}") logging.info(f"Forcing language to {forceLanguage}")
if custom_parameters: if custom_regroup:
# Convert the string to a dictionary result = model.transcribe_stable(file_path, language=forceLanguage, task=transcribe_or_translate, progress_callback=progress, initial_prompt=custom_model_prompt, regroup=custom_regroup)
params_dict = dict(param.split('=') for param in custom_parameters.split(', ')) else:
result = model.transcribe_stable(file_path, language=forceLanguage, task=transcribe_or_translate, progress_callback=progress, initial_prompt=custom_model_prompt, **params_dict) result = model.transcribe_stable(file_path, language=forceLanguage, task=transcribe_or_translate, progress_callback=progress, initial_prompt=custom_model_prompt)
appendLine(result) appendLine(result)
result.to_srt_vtt(get_file_name_without_extension(file_path) + subextension, word_level=word_level_highlight) result.to_srt_vtt(get_file_name_without_extension(file_path) + subextension, word_level=word_level_highlight)
elapsed_time = time.time() - start_time elapsed_time = time.time() - start_time