Transcribes, translates, and answers questions about recordings in 40 languages, with support for audio up to one hour. Runs offline and can produce speaker labels and word-level timestamps on request.
