Nvidia Releases Free Speech-to-Text Model Covering 40 Languages
One-line summary: Nvidia has released a free speech-to-text model that can transcribe up to 40 languages.
Key points
- The open transcription model is free to use and supports up to 40 languages.
- It works only on recorded audio, not live speech.
- It is seen as a significant step for meeting and medical transcription tools.
- Real-time transcription remains the territory of rivals such as ElevenLabs.
Why it matters
A free open model can lower build costs for companies making meeting and medical transcription products. But without real-time support, the competitive picture for live voice services stays largely unchanged.