#3873 TaskCfgVTT(is_cuda=True, uuid='4cebd95d1a', cache_folder='C:/Users/Usuario/Downloads/win-pyvideotrans-v3.98-312/tmp/1064

2800:484* Posted at: 2 days ago 👁22

Merge Error An error has occurred. There may be insufficient memory or video memory.
A child process terminated abruptly, the process pool is not usable anymore
Traceback (most recent call last):
File "videotrans\configure\_base.py", line 280, in _new_process
File "videotrans\process\signelobj.py", line 81, in submit_task_gpu
File "concurrent\futures\process.py", line 720, in submit
concurrent.futures.process.BrokenProcessPool: A child process terminated abruptly, the process pool is not usable anymore

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
File "videotrans\task\job.py", line 288, in run
File "videotrans\task\trans_create.py", line 789, in assembling
File "videotrans\task\trans_create.py", line 1251, in _join_video_audio_srt
File "videotrans\task\trans_create.py", line 530, in recogn2pass
File "videotrans\recognition\__init__.py", line 265, in run
File "videotrans\recognition\_base.py", line 143, in run
File "videotrans\recognition\_overall.py", line 33, in _exec
File "videotrans\recognition\_overall.py", line 105, in _faster
File "videotrans\configure\_base.py", line 294, in _new_process
RuntimeError: An error has occurred. There may be insufficient memory or video memory.
A child process terminated abruptly, the process pool is not usable anymore
TaskCfgVTT(is_cuda=True, uuid='4cebd95d1a', cache_folder='C:/Users/Usuario/Downloads/win-pyvideotrans-v3.98-312/tmp/10644/4cebd95d1a', target_dir='L:/_video_out/12-ts', source_language='Portuguese', source_language_code='pt', source_sub='L:/_video_out/12-ts/pt.srt', source_wav='C:/Users/Usuario/Downloads/win-pyvideotrans-v3.98-312/tmp/10644/4cebd95d1a/pt.wav', source_wav_output='L:/_video_out/12-ts/pt.m4a', target_language='Spanish', target_language_code='es', target_sub='L:/_video_out/12-ts/es.srt', target_wav='C:/Users/Usuario/Downloads/win-pyvideotrans-v3.98-312/tmp/10644/4cebd95d1a/target.wav', target_wav_output='L:/_video_out/12-ts/es.m4a', name='L:/video/12.ts', noextname='12', basename='12.ts', ext='ts', dirname='//RaiDrive-Usuario/Google Drive (14)/video', shound_del_name=None, translate_type=0, tts_type=0, volume='+0%', pitch='+0Hz', voice_rate='+0%', voice_role='Victor(Male/PR)', voice_autorate=True, video_autorate=False, remove_silent_mid=False, align_sub_audio=True, detect_language='pt', recogn_type=0, model_name='large-v3-turbo', shibie_audio=None, remove_noise=False, enable_diariz=False, nums_diariz=0, rephrase=0, fix_punc=False, subtitle_language=None, app_mode='biaozhun', subtitles='', targetdir_mp4='L:/_video_out/12-ts/12.mp4', novoice_mp4='C:/Users/Usuario/Downloads/win-pyvideotrans-v3.98-312/tmp/10644/4cebd95d1a/novoice.mp4', is_separate=False, instrument=None, vocal=None, back_audio='', clear_cache=False, background_music=None, subtitle_type=0, only_out_mp4=False, recogn2pass=True, output_srt=0, copysrt_rawvideo=False)

=
system:Windows-10-10.0.26200-SP0
version:v3.98
frozen:True
language:en
root_dir:C:/Users/Usuario/Downloads/win-pyvideotrans-v3.98-312

1 Replies

Floor #1 DeepSeek AI reply at 2 days ago

This error indicates insufficient GPU memory (VRAM) or system memory (RAM) during the "二次识别" (second-pass recognition) phase, which is triggered because you have enabled the "二次识别" option. The process pool crashed because a child process terminated abruptly due to memory constraints.

The configuration shows you are using:

  • faster-whisper(本地) with the large-v3-turbo model.
  • is_cuda=True (GPU acceleration enabled).
  • recogn2pass=True (二次识别 enabled).

The large-v3-turbo model requires significant VRAM. When "二次识别" is enabled, it runs a second instance of the speech recognition model on the generated配音 audio, potentially doubling the VRAM usage if multiple videos are being processed concurrently.

Solution Steps:

  1. Disable "二次识别" (Second-pass Recognition): In the main translation interface, uncheck the "二次识别" option. This is the most direct way to prevent the memory spike during the final stage. You can still achieve good sync by enabling "配音加速" and/or "视频慢速".
  2. Use a Smaller Model: If you need to keep "二次识别" enabled, switch to a smaller, less memory-intensive model. In the "语音识别" dropdown, change from large-v3-turbo to medium, small, or base.

  3. Adjust Advanced Settings (if keeping the large model): Go to Menu -> Tools/Options -> Advanced Options. In the "语音识别参数" section, make the following changes to reduce VRAM usage at the cost of some precision:

    • CUDA数据类型: Change from float32 to float16 or int8.
    • 识别准确度beam_size: Change from 5 to 1.
    • 识别准确度best_of: Change from 5 to 1.
    • 启用上下文感知: Ensure this is set to false.
  4. Ensure Serial Processing for Batch Jobs: If you are translating multiple videos, go to Menu -> Tools/Options -> Advanced Options. In the "通用设置" section, check the option "批量翻译时强制串行". This ensures videos are processed one after another, preventing multiple models from loading into VRAM simultaneously.
  5. Check GPU Task Concurrency: In the same "通用设置" section, ensure "GPU同时任务数[重启生效]" is set to 1 (the default and safest value unless you have a GPU with >20GB VRAM).

请查阅相关文档：

Post Your Reply
Open source and free maintenance is not easy. If this project is helpful to you, please consider making a small donation to help the project continue to maintain and update.

Related resource