Technology
Voice AI in vogue: Nvidia releases new open transcription model
Source: TechCrunch · Published
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.
Related stories
- Xreal Aura is a new wired headset powered by Android XR
TechCrunch · 28 Oct 2026, 17:00
- As AI agents reshape the web, Google opens Gemini and Search to more apps
TechCrunch · 28 Oct 2026, 17:00
- Chrome’s AI-powered Autofill can now dig through your Gmail to fill out forms
TechCrunch · 28 Oct 2026, 17:00
- Reddit is the latest platform to embrace the ‘For You’ feed
TechCrunch · 28 Oct 2026, 16:00
- AI execs may be too out of touch to build for the general public
TechCrunch · 28 Oct 2026, 15:53
- Trump DOE loans $5B to nuclear plants that will power data centers
TechCrunch · 28 Oct 2026, 15:53
We are a news aggregator: we show the headline and a short excerpt with a link to the original source. We never republish full articles.