Model releaseModerate global significanceConfirmed confidence
Deepgram releases Nova-3
Deepgram releases Nova-3 for real-time and batch transcription, with stronger handling of noise, overlapping speech and word-level timestamps.
What happened
Event details
Nova-3 provides real-time and prerecorded-audio transcription through a hosted API and can integrate a versioned diarizer to produce word-level speaker labels.
Assessment
Why it matters
It combines low latency, word-level timestamps, overlapping-speech handling and optional speaker diarization in a commercial real-time ASR service.
68/100Global significance score. Regional effects are recorded only when the evidence supports a meaningful difference.
Availability
Access notes
The hosted API is generally available. Self-hosting is offered for enterprise deployments, and diarization is integrated through a separately versioned diarizer.