[harryf]

Have you considered supporting whisper-at - https://github.com/YuanGongND/whisper-at ? Being able to identify sounds on a timeline can be useful e.g. politicians speech and how the audience is reacting to it (e.g. clapping, applauding)