What was announced?
Nvidia released a free speech-to-text model capable of transcribing up to 40 languages. The model converts recorded audio into text.
Limitations
The model works only on recorded audio files and cannot process live audio streams in real time. As a result, use cases like instant translation or live captioning are currently out of scope.
Who benefits?
It can be used for post-recording transcription, meeting notes, and medical documentation. Real-time audio processing remains the domain of rivals such as ElevenLabs.
