Boilerplate available to render video of per-word closed captions?

[ArcAster]

Curious if core devs or someone using this could post some boilerplate / example code that was used to generate the live transcription videos in the readme?

Thanks!

[jianfch]

Those are just hardsubs on a black background, encoded with FFmpeg:

ffmpeg -f lavfi -i color=size=720x120:rate=25:color=black -i audio.mp3 -vf "subtitles=audio.ass:force_style='Fontsize=90'" -shortest audio_subbed.mp4

The same be done with a video:

ffmpeg -i video.mp4 -vf "subtitles=audio.ass:force_style='Fontsize=90'" video_subbed.mp4

For subtitles={subtitle file}, the file could be any format supported by FFmpeg.
stable-ts supports .srt and .ass outputs with these functions:
results_to_sentence_srt
results_to_word_srt
results_to_token_srt
results_to_sentence_word_ass