#2419 Application started at 2026-01-02 11:51:26 load subtitles area...Set ui text...Set style...Set controls style...Set menu

183.213* Posted at: 5 hours ago 👁9


Application started at 2026-01-02 11:51:26

load subtitles area...
Set ui text...
Set style...
Set controls style...
Set menu...
Load function window...
Set default params
Bind signal...
2026-01-02 11:51:34,415 - INFO - 平台: Windows。正在按优先级检测最佳的 'h264' 编码器: ['nvenc', 'qsv', 'amf']
Import torch...
2026-01-02 11:51:38,584 - INFO - 正在测试编码器是否可用: h264_nvenc...
end
2026-01-02 11:51:39,232 - INFO - 硬件编码器 'h264_nvenc' 可用。
2026-01-02 11:51:39,233 - INFO - 已选择硬件编码器: h264_nvenc
2026-01-02 11:51:39,233 - INFO - 最终确定使用的编码器: h264_nvenc
2026-01-02 11:51:39,233 - INFO - 平台: Windows。正在按优先级检测最佳的 'hevc' 编码器: ['nvenc', 'qsv', 'amf']
2026-01-02 11:51:39,234 - INFO - 正在测试编码器是否可用: hevc_nvenc...
2026-01-02 11:51:39,492 - INFO - 硬件编码器 'hevc_nvenc' 可用。
2026-01-02 11:51:39,493 - INFO - 已选择硬件编码器: hevc_nvenc
2026-01-02 11:51:39,493 - INFO - 最终确定使用的编码器: hevc_nvenc
self.proxy_str=None,self.uuid=None
2026-01-02 11:51:47,845 - INFO -
[deepseek]发送请求数据:message=[{'role': 'system', 'content'
......
11:58:09,990 - DEBUG - Compression ratio threshold is not met with temperature 0.2 (18.666667 > 2.400000)
2026-01-02 11:58:14,101 - DEBUG - Compression ratio threshold is not met with temperature 0.4 (18.666667 > 2.400000)
2026-01-02 11:58:18,217 - DEBUG - Compression ratio threshold is not met with temperature 0.6 (18.666667 > 2.400000)
2026-01-02 11:58:22,333 - DEBUG - Compression ratio threshold is not met with temperature 0.8 (18.666667 > 2.400000)
2026-01-02 11:58:26,439 - DEBUG - Compression ratio threshold is not met with temperature 1.0 (18.666667 > 2.400000)
2026-01-02 11:58:26,755 - DEBUG - Processing segment at 02:00.000
2026-01-02 11:58:34,598 - DEBUG - Compression ratio threshold is not met with temperature 0.0 (18.666667 > 2.400000)
2026-01-02 11:58:34,598 - DEBUG - Log probability threshold is not met with temperature 0.0 (-291.128876 < -1.000000)
2026-01-02 11:58:38,705 - DEBUG - Compression ratio threshold is not met with temperature 0.2 (18.666667 > 2.400000)
2026-01-02 11:58:42,838 - DEBUG - Compression ratio threshold is not met with temperature 0.4 (18.666667 > 2.400000)
2026-01-02 11:58:46,942 - DEBUG - Compression ratio threshold is not met with temperature 0.6 (18.666667 > 2.400000)
2026-01-02 11:58:51,061 - DEBUG - Compression ratio threshold is not met with temperature 0.8 (18.666667 > 2.400000)
2026-01-02 11:58:55,196 - DEBUG - Compression ratio threshold is not met with temperature 1.0 (18.666667 > 2.400000)
Traceback (most recent call last):
File "videotrans\task\_only_one.py", line 43, in run
File "videotrans\task\trans_create.py", line 379, in recogn
RuntimeError: 1.mp4没有识别出字幕,请检查是否包含人类说话声音,以及说话语言是否和你选择的原始语言匹配
self.proxy_str=None,self.uuid=None
2026-01-02 12:02:39,145 - INFO - 最终配置信息:self.cfg=TaskCfg(cache_folder='D:/py/win-pyvideotrans-v3.92/tmp/5012/2fa6946c81', target_dir='D:/py/video/_video_out/1-mp4', remove_noise=False, is_separate=False, detect_language='en', subtitle_language=None, source_language='英语', target_language='简体中文', source_language_code='en', target_language_code='zh-cn', source_sub='D:/py/video/_video_out/1-mp4/en.srt', target_sub='D:/py/video/_video_out/1-mp4/zh-cn.srt', source_wav='D:/py/win-pyvideotrans-v3.92/tmp/5012/2fa6946c81/en.wav', source_wav_output='D:/py/video/_video_out/1-mp4/en.m4a', target_wav='D:/py/win-pyvideotrans-v3.92/tmp/5012/2fa6946c81/target.wav', target_wav_output='D:/py/video/_video_out/1-mp4/zh-cn.m4a', subtitles='', novoice_mp4='D:/py/win-pyvideotrans-v3.92/tmp/5012/2fa6946c81/novoice.mp4', noextname='1', shibie_audio=None, targetdir_mp4='D:/py/video/_video_out/1-mp4/1.mp4', instrument=None, vocal=None, back_audio='', background_music=None, app_mode='biaozhun', subtitle_type=1, volume='+0%', pitch='+0Hz', voice_rate='+0%', voice_role='Xiaoyi(Female/CN)', copysrt_rawvideo=False, clear_cache=True, translate_type=4, tts_type=0, recogn_type=0, model_name='large-v3-turbo', split_type=1, voice_autorate=False, video_autorate=False, cuda=True, name='D:/py/video/1.mp4', basename='1.mp4', ext='mp4', dirname='D:/py/video', shound_del_name=None, uuid='2fa6946c81', remove_silent_mid=False, align_sub_audio=True, enable_diariz=False, nums_diariz=0, rephrase=1, only_out_mp4=False)
2026-01-02 12:02:39,365 - INFO - cmd=['ffmpeg', '-hide_banner', '-ignore_unknown', '-threads', '0', '-y', '-i', 'D:/py/video/1.mp4', '-vn', '-ar', '16000', '-c:a', 'pcm_s16le', 'D:/py/win-pyvideotrans-v3.92/tmp/5012/2fa6946c81/en.wav']
2026-01-02 12:02:39,376 - INFO - cmd=['ffmpeg', '-hide_banner', '-ignore_unknown', '-threads', '0', '-y', '-fflags', '+genpts', '-i', 'D:/py/video/1.mp4', '-an', '-c:v', 'copy', 'D:/py/win-pyvideotrans-v3.92/tmp/5012/2fa6946c81/novoice.mp4']
self.proxy_str=None,self.uuid='2fa6946c81'
2026-01-02 12:02:43,319 - INFO - 开始创建 pid:self.pidfile='D:/py/win-pyvideotrans-v3.92/tmp/5012/16224.lock'
2026-01-02 12:02:48,878 - INFO - Processing audio with duration 02:27.075
2026-01-02 12:02:49,375 - INFO - VAD filter removed 00:18.496 of audio

1 Replies

根据日志,您的问题是视频“1.mp4”没有识别出字幕。错误信息明确指出:“请检查是否包含人类说话声音,以及说话语言是否和你选择的原始语言匹配”。日志显示VAD(语音活动检测)过滤掉了18.496秒的音频,但剩余的语音片段未能被faster-whisper模型成功转录。

核心原因与解决方案:

  1. 视频内容问题:请确认您的视频确实包含清晰的人类说话声,而不是纯音乐、环境音或噪音。软件无法处理无说话声的视频。
  2. 语言设置错误:您在配置中设置了 source_language='英语' (en)。请检查视频中的实际说话语言是否为英语。如果不是,请将“发音语言”更改为正确的语言或选择 auto(自动检测)。
  3. 语音识别模型或参数问题:您使用的是 faster-whisperlarge-v3-turbo 模型,并启用了CUDA加速。日志中反复出现 Compression ratio threshold is not met 的调试信息,这通常意味着模型对音频的“置信度”不足,无法生成可靠的转录结果。

请按以下步骤排查和尝试:

第一步:基础检查

  • 用播放器打开视频,确认有人类说话声。
  • 在软件主界面,确保“发音语言”选择正确。如果不确定,请选择 auto

第二步:调整语音识别设置

  1. 在软件主界面,将“识别模式”从“整体识别”切换为“批量推理”。这能改变处理方式,有时能解决特定音频的识别问题。
  2. 如果问题依旧,尝试更换更小或更稳定的模型。在“语音识别”下拉框中,将 large-v3-turbo 改为 large-v2medium

第三步:微调高级参数(如果上述无效)

  1. 点击主界面右下角的“设置更多参数”。
  2. 勾选“降噪”选项,这有助于在识别前清理背景噪音。
  3. 在“VAD参数”区域,尝试将“语音阈值”从默认的 0.45 调低(例如改为 0.3),使语音检测更敏感。
  4. 在“Faster/OpenAI/Whisper 语音识别微调”区域,将“识别准确度 (beam_size)”和“识别准确度 (best_of)”都设置为 1。这能显著降低显存消耗并避免某些复杂情况下的错误。

第四步:尝试其他识别渠道
如果本地模型始终无法识别,可以尝试使用在线API(如OpenAI Whisper API),这通常识别率更高。但需在“设置”中配置相应的API密钥。

完成调整后,务必勾选主界面第一行的“清理已生成”,然后重新处理视频。

请查看更多参考文档:

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.

Donate: https://ko-fi.com/jianchang512

Trending Questions