NVIDIA Releases Nemotron 3 Diarization: A 100M-Parameter Open-Weight Model That Tracks 8 Speakers in Real Time

NVIDIA has released Nemotron 3 Diarization, a 100M-parameter open-weight model on Hugging Face. This tool identifies who spoke when in conversations, tracking up to 8 speakers including overlapping voices. A single checkpoint supports both offline recordings and real-time streaming. By open-sourcing Nemotron 3, NVIDIA aims to revolutionize real-time audio processing and improve AI integration for multi-speaker applications. The model is fully deployable, offering a significant advancement in how systems handle complex, real-time audio tracking tasks.