RC RANDOM CHAOS

Voxtral 3: Expressive Real-Time Voice Agent for Enterprise

· via Hacker News

Original source

Voxtral 3

Hacker News →

Mistral released Voxtral 3, a voice agent platform combining speech recognition, reasoning, and synthesis in a single real-time pipeline. The system transcribes across 50+ languages with speaker identification and context-aware accuracy, while its text-to-speech engine supports 35+ languages and 50+ dialects with dynamic emotion, pacing, and emphasis. The integrated reasoning layer handles complex multi-step tasks during live conversations.

Organizations deploy via EU-hosted API or self-hosted infrastructure, controlling where audio is processed and whether conversations are retained. The voice agent SDK completes conversation turns in under one second, handles interruptions gracefully, and executes business workflows through function calling—including identity verification, account checks, and appointment booking. Pricing is $0.0035 per minute for transcription and $13 per million characters for speech synthesis. Primary use cases span contact centers, customer service, and hands-free copilot applications.

Read the full article

Continue reading at Hacker News →

This is an AI-generated summary. Read the original for the full story.