Auto-generated captions
Captions generated by YouTube's speech-recognition system. Available on most videos but less accurate than human-authored tracks.
Auto-generated captions come from YouTube's ASR. Japanese quality has improved since 2023 but still lags human-authored tracks on named entities, jargon, and fast speech. SumTube's prompts are tuned to survive reasonable ASR noise by relying on context rather than exact transcription.