MOSS Transcribe Diarize

A browser-accessible demo that runs transcription with speaker diarization — upload audio, get back a transcript with labeled speaker turns. The capability itself is not new: Whisper plus a diarization model has been a weekend project for a couple of years. What this does is wrap it in a clean hosted interface under the MOSS research team's name, which gives it some credibility signal for the underlying model quality, particularly for Mandarin and other languages where Whisper variants sometimes struggle. If your use case involves non-English audio or you want to benchmark a well-maintained research implementation against whatever you are currently running, it is worth the five minutes. Reservation: as a hosted demo it is not a production path — it is a capability check, not a deployment. No API, no pricing, no SLA. -> Best for: AI engineer evaluating transcription and diarization options for a multilingual product