# Nvidia Releases a 100-Million-Parameter Speaker Diarization Model Under a Commercial Open License

Nemotron 3 Diarization labels up to eight speakers in a single end-to-end pass, runs streaming at a 320-millisecond input buffer, and is downloadable from Hugging Face for commercial use.

- Published: 2026-09-25T06:22:14.083Z
- Canonical: https://polylog.news/ai/2026-09-25/nvidia-releases-a-100-million-parameter-speaker-diarization
- Publisher: Polylog (AI desk)
- Section: tech
- Sources: [Polylog editors](https://polylog.news), [MarkTechPost](https://www.marktechpost.com/2026/09/23/nvidia-releases-nemotron-3-diarization/), [Hugging Face](https://huggingface.co/nvidia/Nemotron-3-Diarization)

Speaker diarization, the task of marking who spoke when in an audio stream, has long been served by a pipeline: segment the audio, embed each segment, cluster the embeddings, then tune the thresholds for every new domain. Nvidia has release…

This story is for subscribers. Read it in full at https://polylog.news/ai/2026-09-25/nvidia-releases-a-100-million-parameter-speaker-diarization (subscription information: https://polylog.news/pricing).