Tencent Cloud & Inworld AI partner for real-time voice AI with sub-130ms TTS

Tencent Cloud and Inworld AI announced a strategic partnership to deliver a one-stop, lifelike, real-time voice AI solution. Inworld's text-to-speech (TTS) models, ranked #1 on the Artificial Analysis Speech Arena, are deeply integrated into Tencent Real-Time Communication (Tencent RTC). This gives developers access to ultra-realistic, context-aware speech with sub-130ms first-chunk latency, over 100 languages, and instant cross-lingual conversion while preserving a consistent speaker voice. Developers can clone voices, design custom voices, control vocal style, and stream natural responses directly via the Tencent RTC console and SDK.

On the infrastructure side, Tencent RTC provides an enterprise-grade backbone with more than 3,200 global nodes and sub-300ms worldwide latency. Advanced features like AI noise suppression and weak-network resilience ensure seamless performance even in challenging connectivity environments. The partnership also includes a joint Conversational AI Demo with curated voices for different languages and scenarios. Tencent Cloud's Wison Xie emphasized that production-grade voice AI requires both a capable model and a capable network, and this partnership delivers a complete path from prototype to production for developers building emotionally intelligent voice applications.