1h ago
Meta Voice Transcribe as a model option
It's SOTA on AA Speech to text benchmarks.
"The Meta Model API for real-time voice transcription supports a Zero Data Retention (ZDR) tier" so privacy concerns are mitigated. I guess the risk for the integration is the rate limits on the Meta dev platform. They haven't published any concrete limits like with their LLMs, so I assume it's high enough for this user base
They have real time transcription support and diarization
Pending