Nvidia's Nemotron 3.5 Transcribe tops open ASR leaderboards with only 4B active parameters
Nvidia Nemotron 3.5 Transcribe: open ASR for 40 languages
Nvidia's Nemotron 3.5 Transcribe is an open-weight, promptable speech model that handles transcription in 40 languages, translation in 31, and speaker attribution in 13, all from a single checkpoint. It leads the Hugging Face Open ASR Leaderboard for English at 5.33% WER and ties for first on Artificial Analysis at 2.8% AA-WER while activating roughly 4B of its 30B parameters. The release ships with Granary v2, an open 463,062-hour multilingual data pipeline, and fine-tuning recipes that cut medical-domain WER by 45.8%.
For the VoxPopuli dataset, it reduced 354k hours to 111k hours across 22 European languages.
- xpct
I played around with Nemotron 3 recently. It's OK, but makes an error every few sentences, at least on the data I worked with (standard vocab). I'm sure we'll get there, but I was pleasantly surprised by how I can recognize words better than it can.
- solarkraft
> Transcribe is built for recorded audio, not live streaming.
Bit unfortunate! Nemotron Streaming in Handy has basically solved dictation for me. It recognizes German mixed with English terms and brand names really well.
- AkashKaStudio
From the model card, this supports near-field only speaker transcription, experimental but still something not available in any other OSS Speech LLM.