Inworld AI Launches Realtime TTS-2: A Closed-Loop Voice Model That Adapts to How You Actually Talk

Inworld AI launched Realtime TTS-2, a research-preview voice model for more natural conversational speech.

The closed-loop system listens to prior turns and supports inline voice direction, nonverbal cues, and better tone, pacing, and emotion.

It maintains a consistent voice across 100+ languages and can generate reusable voices from written prompts without reference audio.

The model is available through the Inworld API and Inworld Realtime API.