NVIDIA releases Nemotron 3.5 Transcribe, an open speech LLM for 40 languages
AINVIDIA has released Nemotron 3.5 Transcribe, an open-weights speech LLM that handles transcription in 40 languages for audio up to one hour. The model reports 4.71% mean WER on nine public English Open ASR sets, the best open-weights score in NVIDIA's report, while commercial APIs still lead.
Loading source posts…
This story is featured in top picks: “NVIDIA releases Nemotron 3.5 Transcribe, a 30B-A4B open-weight speech model”
Source details
MarkTechPost · Asif Razzaq · marktechpost.comPublished · added here
Original title: NVIDIA Releases Nemotron 3.5 Transcribe: An Open 30B MoE Model for 40-Language Transcription with ~4B Active Parameters