**Know Who Spoke When: Build Real-Time, Multi-Speaker AI with NVIDIA Nemotron 3 Diarization**
NVIDIA open-sources Nemotron 3 diarization — speaker identification at real-time speed now available to builders.
Why it matters
Speaker diarization (who said what, when) is table-stakes for voice AI products. NVIDIA's open-weight release lowers the barrier to multi-speaker audio understanding for practitioners building voice agents, meeting transcription, and conversational AI.
The key facts
9 to knowNVIDIA Nemotron 3 diarization model open-sourced
Real-time multi-speaker identification capability
Open-weight release on Hugging Face
Addresses speaker tracking in audio streams
Published September 23, 2026
NVIDIA Nemotron 3 diarization — open-weight release
Real-time, multi-speaker capability
Published Sept 23, 2026 on Hugging Face
Reduces dependency on proprietary speaker-diarization APIs for voice AI builders
Go to the source
Hugging Face Bloghuggingface.co