#3631 卡到降噪这里用3.92版本覆盖3.95补丁包就没这个问题

58.16* Posted at: 1 hour ago 👁17


Application started at 2026-03-16 11:37:11

[DEBUG] 返回缓存的编解码器 Windows-264: h264_nvenc
[DEBUG] 返回缓存的编解码器 Windows-265: hevc_nvenc
First searching GPU...
Found 1 GPUs, cost=3s
start workers...
NVIDIA_GPU_NUMS=1
process_max_gpu=0
multi_gpus=False
Concurrent task_nums=1
process_max=0
start 8 jobs
ended workers
[DEBUG] 最终配置信息:self.cfg=TaskCfg(cache_folder='F:/pyvideotrans/tmp/8016/3700875821', target_dir='D:/Downloads/_video_out/天启-restored-mp4', remove_noise=True, is_separate=False, detect_language='ja', subtitle_language=None, source_language='日语', target_language='日语', source_language_code='ja', target_language_code='ja', source_sub='D:/Downloads/_video_out/天启-restored-mp4/ja.srt', target_sub='D:/Downloads/_video_out/天启-restored-mp4/ja.srt', source_wav='F:/pyvideotrans/tmp/8016/3700875821/ja.wav', source_wav_output='D:/Downloads/_video_out/天启-restored-mp4/ja.m4a', target_wav='F:/pyvideotrans/tmp/8016/3700875821/target.wav', target_wav_output='D:/Downloads/_video_out/天启-restored-mp4/ja.m4a', subtitles='', novoice_mp4='F:/pyvideotrans/tmp/8016/3700875821/novoice.mp4', noextname='天启-restored', shibie_audio=None, targetdir_mp4='D:/Downloads/_video_out/天启-restored-mp4/天启-restored.mp4', instrument=None, vocal=None, back_audio='', background_music=None, app_mode='tiqu', subtitle_type=0, volume='+0%', pitch='+0Hz', voice_rate='+0%', voice_role='No', copysrt_rawvideo=False, clear_cache=False, translate_type=0, tts_type=0, recogn_type=0, model_name='large-v3', voice_autorate=False, video_autorate=False, cuda=True, name='D:/Downloads/天启-restored.mp4', basename='天启-restored.mp4', ext='mp4', dirname='D:/Downloads', shound_del_name=None, uuid='3700875821', remove_silent_mid=False, align_sub_audio=True, enable_diariz=False, nums_diariz=0, rephrase=0, only_out_mp4=False, fix_punc=False, recogn2pass=True)
[DEBUG] cmd=['F:/pyvideotrans/ffmpeg/ffmpeg.exe', '-hide_banner', '-ignore_unknown', '-threads', '0', '-y', '-i', 'D:/Downloads/天启-restored.mp4', '-vn', '-ac', '1', '-ar', '16000', '-c:a', 'pcm_s16le', 'F:/pyvideotrans/tmp/8016/3700875821/ja.wav']
[DEBUG] [job] 进入执行语音识别阶段 trk.cfg=TaskCfg(cache_folder='F:/pyvideotrans/tmp/8016/3700875821', target_dir='D:/Downloads/_video_out/天启-restored-mp4', remove_noise=True, is_separate=False, detect_language='ja', subtitle_language=None, source_language='日语', target_language='日语', source_language_code='ja', target_language_code='ja', source_sub='D:/Downloads/_video_out/天启-restored-mp4/ja.srt', target_sub='D:/Downloads/_video_out/天启-restored-mp4/ja.srt', source_wav='F:/pyvideotrans/tmp/8016/3700875821/ja.wav', source_wav_output='D:/Downloads/_video_out/天启-restored-mp4/ja.m4a', target_wav='F:/pyvideotrans/tmp/8016/3700875821/target.wav', target_wav_output='D:/Downloads/_video_out/天启-restored-mp4/ja.m4a', subtitles='', novoice_mp4='F:/pyvideotrans/tmp/8016/3700875821/novoice.mp4', noextname='天启-restored', shibie_audio=None, targetdir_mp4='D:/Downloads/_video_out/天启-restored-mp4/天启-restored.mp4', instrument=None, vocal=None, back_audio='', background_music=None, app_mode='tiqu', subtitle_type=0, volume='+0%', pitch='+0Hz', voice_rate='+0%', voice_role='No', copysrt_rawvideo=False, clear_cache=False, translate_type=0, tts_type=0, recogn_type=0, model_name='large-v3', voice_autorate=False, video_autorate=False, cuda=True, name='D:/Downloads/天启-restored.mp4', basename='天启-restored.mp4', ext='mp4', dirname='D:/Downloads', shound_del_name=None, uuid='3700875821', remove_silent_mid=False, align_sub_audio=True, enable_diariz=False, nums_diariz=0, rephrase=0, only_out_mp4=False, fix_punc=False, recogn2pass=True)
2026-03-16 11:37:43,805 - modelscope - WARNING - We can not confirm the cached file is for revision: master

[DEBUG] GPU进程池:max_workers=1
[INFO] 开始语音降噪

1 Replies

这是由于降噪功能在 v3.95 版本中可能存在问题。根据你的描述,使用 v3.92 版本覆盖 v3.95 补丁包后问题消失,这表明 v3.95 版本在降噪环节可能存在兼容性或执行错误。

解决方案:

  1. 降级或使用完整包:如果你当前使用的是 v3.95 的补丁包,建议下载 v3.98 的完整包重新安装。如果问题依旧,可以暂时回退到 v3.92 的完整版本。
  2. 关闭降噪功能:在“翻译视频或音频”界面,取消勾选“降噪”选项。降噪功能并非必需,对于大多数清晰音频,关闭后不影响识别准确度。
  3. 检查高级设置:确保“菜单-工具-高级选项-语音识别参数”中,没有启用可能导致冲突的选项(如特定的 VAD 设置)。
  4. 清理缓存:在处理新任务时,勾选“清理已生成”选项,避免旧缓存文件干扰。

如果以上方法无效,请提供 logs 文件夹中最新日志文件的底部约30行报错信息,以便进一步分析。

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