WhisperX
FreeWhisper for subtitles — word-level timestamps and speaker diarization
Adds word-level timestamps and speaker diarization on top of Whisper, so subtitles align per word and multi-speaker audio is labelled by who said what. For interview transcripts, meeting notes and video subtitles it is the only one here with both out of the box.
Why we picked it
If your work is subtitles or multi-speaker transcripts, WhisperX has essentially no substitute. 23,575 stars, BSD-2-Clause, latest v3.8.6 (25 May 2026).
It handles what the other tools do not: word-level timestamps so subtitles align precisely, and speaker diarization so an interview transcript knows who said which line. Assembling those yourself is a chore; here they come built in.
The cost is a longer dependency chain and a fussier install than whisper.cpp. If you only need a block of text, skip it and take a desktop app instead.