英偉達發布免費100M參數Nemotron 3模型，實現即時識別八個說話者
英偉達發布nemotron 3，一款100m參數的ai模型，可即時識別最多八個說話者，模型權重免費開放。
nvidia released nemotron 3, a 100m-parameter ai model for real-time speaker diarization. it can identify up to eight speakers and detect when multiple people talk at the same time. in the voicearena diarization benchmark v1, nemotron 3 leads with a der of 14.7% and outperforms its predecessor, streaming sortformer, by 41%. the model works with both recordings and live audio, and can produce transcripts with speaker labels when paired with a speech recognition system.