Mistral Showcases Voxtral Speech-to-Text Demo

sophiamyang · x · 2026-07-17

Mistral's official account shared a demo using Voxtral speech-to-text: the same audio snippet can yield two "completely different transcripts," creating a contrasting showcase.

The post included a football-related joke, mentioning "football isn't coming home," hinting that this transcription capability can handle voice recognition demos across various accents and scenarios.

Original post →

More from Multimodal

Multimodal channel →