#6215 Why the step that freeze the end videos usually take about 30 minutes to do and when it done. It freeze the end so long

This is caused by the “视频慢速” (Video Slow Motion) alignment feature. When enabled, pyVideoTrans stretches the video timeline to match the dubbing duration. This is extremely CPU‑intensive (explaining the ~30‑minute final step) and has known FFmpeg setpts precision limitations, which can cause the timeline to drift and the final portion of the video to become a long, frozen frame that overwrites the original visual content. The dubbing audio is then misaligned: some voice segments are lost (truncated or placed on the wrong timeline) and reappear later inside the frozen part.

Solutions:

1. In the main interface, uncheck 视频慢速 (Video Slow Motion). Keep only 音频加速 (Audio Speedup) so the dubbing is speeded up to match the original video timeline.
2. If you must use slow motion, go to 菜单 → 工具 → 高级选项 → 字幕声音画面对齐 and reduce 视频慢放最大倍数 (maximum video slowdown ratio) — e.g., set it to 2 or lower.
3. Enable 二次识别 (second-pass recognition) so the subtitles are regenerated against the dubbing timeline for better alignment.
4. If the dubbing is longer than the original video, either enable 音频加速 or increase 配音语速 to prevent the final voice part from being truncated.
5. Avoid enabling both 视频慢速 and 分离人声背景声 at the same time — this multiplies processing time and temporary file size dramatically.

请查阅相关文档：

• 字幕语音画面对齐
• 常见问题FAQ