Hume AI launches EVI 3 empathic voice API for real-time agent conversations

Read the original article →

What happened

Hume AI launched EVI 3 on July 26, an empathic voice interface API that adjusts tone, pacing, and word choice based on detected user emotion in real time.

The API supports streaming speech-to-text and text-to-speech in a single WebSocket connection with sub-300ms round-trip latency. Hume targets therapy intake, sales calls, and customer-support agents. Pricing starts at $0.07 per minute of conversation.

Why it matters

Most voice agents sound flat regardless of user frustration or confusion. Emotion-aware responses could improve completion rates in support and intake flows where tone mismatch causes drop-off.

Combining STT and TTS in one connection also simplifies the integration stack for voice-agent builders.

MintedBrain take

Emotion detection is probabilistic and culturally variable. Test EVI 3 on your actual user population before deploying in sensitive contexts like healthcare intake.

References

This article was originally published at VentureBeat. For the full piece, read the original article.

Discussion

  • Loading…

← Back to News