Support for audio transcription

[joaojhgs]

Audio transcription seems like a really good way to improve note-taking capability enhanced with the composer, allowing for drop-in of audios, or even direct recording in the chat;

Support for a local whisper model and cloud providers would be awesome!

[kevin-on]

Thanks for the suggestion! Audio transcription is on the roadmap, but there are a few more urgent priorities at the moment—including PDF support, adding MCPS, and other core features. Appreciate the feedback!

[BradKML]

ASR is definitely good, but imagine ASR + TTS together for an agent ("if there are >2s pause start inference")? Sorry that sounds too OP
P.S. Might need to look at the leaderboard since Whisper v3 Large is the standard but Nemo is strong https://huggingface.co/spaces/hf-audio/open_asr_leaderboard

[kevin-on]

Thanks for the reference. I'll take a look at the leaderboard.