video-subtitles-and-audio-insert-workflow

Use moviepy 2.x to parse the SRT file into timed text clips, composite them over the video with CompositeVideoClip, and write the result with write_videofile using libx264. Set a bitrate around 1.2-1.5 times the source to avoid oversized outputs.

Install edge-tts and call edge_tts.Communicate with a voice such as zh-CN-YunjianNeural or zh-CN-XiaoxiaoNeural, optionally setting rate, volume, and pitch. Save the result as an MP3, then adjust its duration with FFmpeg atempo or apad filters.

Box characters mean the chosen font lacks the required glyphs, such as using Arial for Chinese text. Switch to a CJK-capable font like STHeiti or Songti on macOS, WenQuanYi Micro Hei on Linux, or Microsoft YaHei on Windows.

Yes, the moviepy 2.x workflow is cross-platform on macOS, Linux, and Windows with Python 3.x. Only the font path differs per operating system, and FFmpeg/ffprobe must be available for encoding and inspection.

The output bitrate is likely set far above the source, for example 5000k when the source is 444 kbps. Check the source bitrate with ffprobe and set the output to roughly 1.2-1.5 times that value, or use CRF encoding around 23 instead.

Measure the audio duration with ffprobe, then speed it up with the FFmpeg atempo filter if too long or pad it with apad if too short. Targeting scene duration minus about 0.2 seconds adds breathing room between scenes.