Home/Events/NVIDIA Nemotron 3 Diarization: Real-Time, Multi-Speaker AI

NVIDIA Nemotron 3 Diarization: Real-Time, Multi-Speaker AI

Confirmed

Confidence

90%

Impact: 80%

Updated 1h ago

Consensus Brief

NVIDIA has introduced the Nemotron 3 Diarization model, an open-weight, 100M-parameter system that excels in speaker diarization, achieving a 14.72% Diarization Error Rate (DER). This model supports up to eight speakers and effectively manages overlapping speech in both live and recorded conversations.

Sourced from

Hugging FaceT1

Primary: Hugging Face

What Changed Since Last Update

1h ago

Nemotron 3 Diarization expands the speaker support from four to eight speakers compared to earlier models.

Claim Ledger

4 claims tracked across sources

Confirmed Fact

Nemotron 3 Diarization ranks #1 on Voice Arena's Diarization-Bench leaderboard.

Confirmed Fact

The model has a 14.72% Diarization Error Rate (DER).

Confirmed Fact

The model supports up to eight speakers across live and recorded conversations.

Confirmed Fact

The model was trained using public and licensed speech data, including multispeaker-annotated conversations.

Role-Based Impact Analysis

Source Timeline

1 source corroborating

**Know Who Spoke When: Build Real-Time, Multi-Speaker AI with NVIDIA Nemotron 3 Diarization**

Hugging Face·1h ago