Mistral releases four audio models in the Voxtral 3 series in public preview
Mistral has made four models in the Voxtral 3 series available in public preview. They offer speech generation with voice cloning, real-time transcription and transcription of recordings with speaker differentiation and timestamps.
Mistral announced the release of four audio models in the Voxtral 3 series on 20 October. All are available in public preview and cover both text-to-speech and speech-to-text conversion.
According to Mistral, Voxtral TTS 3 offers voice cloning without prior fine-tuning, streaming speech generation and context-aware delivery. Voxtral Mini Transcribe Realtime 3 provides real-time transcription. According to the company, Voxtral Mini Transcribe 3 and Voxtral Small Transcribe 3 can transcribe recordings while distinguishing between speakers and adding timestamps.
Why it matters
Depending on the task, interested users can try generating spoken content, streaming transcription or processing existing recordings. Speaker differentiation and timestamps may make interview recordings easier to navigate; availability remains limited to public preview.
Two audiences, two different impacts
What this means
For individuals
Individuals can try converting their own text into speech with delivery that, according to the company, adapts to the context.
For a business
Two models are available for processing business recordings. According to the company, they combine transcription with speaker differentiation and timestamps. Their usefulness in preparing interview records can therefore be tested.
ProcessesCheck the original
Event sources
clearly official source · 1 publisher, 0 independent. We count feeds from the same owner only once.