Mistral Releases Voxtral TTS: Open-Weight 4B Streaming Speech Model
Open-weight 4B TTS model with sub-200ms latency, 9 languages, and zero-shot voice cloning, making voice-enabled agents practical to self-host.
Open-weight 4B TTS model with sub-200ms latency, 9 languages, and zero-shot voice cloning, making voice-enabled agents practical to self-host.