Voice AI in vogue: Nvidia releases new open transcription model
Nvidia just released a free speech-to-text model that can transcribe up to 40 languages, but it only works on recordings, not live audio, so it’s a big step for meeting and medical transcription tools while still leaving real-time use to rivals like ElevenLabs.
TL;DR
- Nvidia released Nemotron 3.5 Transcribe, a free, open-weight AI model for audio transcription.
- The model supports up to 40 languages and can translate 31 languages to/from English.
- Nemotron 3.5 can distinguish up to eight speakers and handle recordings up to an hour long.
- It is designed for developers building business meeting and medical transcription applications.
- The model claims to outperform rivals on benchmarks and can run on a single chip.
- Nvidia also released version 2 of its open-source speech dataset, Granary.
- A limitation is that Transcribe only works with recorded audio, not live streams.