Back to events
Models
Score 82Heat 0

NVIDIA Releases Nemotron 3.5 Transcribe Model and Granary v2 Data Pipeline

Open SourceSpeechTrainingOpen WeightsMultimodalNVIDIANemotron 3.5 TranscribeHugging FaceModelScopeGranary v2

Summary

NVIDIA has released Nemotron 3.5 Transcribe, an open-weight speech model supporting multilingual transcription, translation, and speaker attribution. Alongside the model, NVIDIA introduced Granary v2, an open data curation pipeline and dataset manifest designed to aid speech model training and domain adaptation.

Why it matters

Nemotron 3.5 Transcribe provides top-tier open-weight speech recognition capabilities with promptable control and efficient performance, lowering the barrier for developers building localized audio applications.

Editorial score

1 review
8
Significance
7
Novelty
9
Credibility
8
Reach
9
Actionability