NVIDIA Nemotron 3.5 Transcribe and Granary v2: Intelligent Multilingual Transcription You Can Prompt and Adapt
3.78T1.5 sourceHugging Face Blog
Source record
Published by Hugging Face Blog (T1.5 source). The original is at https://huggingface.co/blog/nvidia/nemotron-transcribe-granary.
Pipeline notes
The summary and note below are generated by the signal pipeline — they are Beyond Desk’s reading, not quotations from the source.
SummaryNVIDIA released Nemotron 3.5 Transcribe, an open-weight, promptable speech model combining transcription, diarization, translation, and audio question answering. It supports 40 languages, 31 translation directions, up to eight speakers, one-hour recordings, domain adaptation, and single-GPU deployment.
Why it mattersBuilders can assess one model for multilingual call, meeting, and dictation workflows; Granary v2 also exposes the data curation pipeline behind the release.
Cited by
No citations on record.