~/AI/tencent-hunyuan-releases-hy-realtime-a-singing-and-tool-calling-speech-model

Tencent Hunyuan Releases Hy Realtime: A Singing and Tool-Calling Speech Model

Tencent Hunyuan has launched Hy Realtime, a new real-time speech model featuring advanced voice comprehension, anthropomorphic expression, agentic tool-calling, and singing capabilities. The model is already integrated into Tencent's Yuanbao app and customer service products, with Tencent Cloud API services arriving soon. This release represents a significant advancement in conversational AI, shifting voice models from basic speech-to-text assistants to interactive agents capable of complex reasoning, web searches, and musical expression. By integrating these capabilities into widespread consumer apps, Tencent is pushing the boundaries of natural, multimodal human-computer interaction. Hy Realtime handles ambiguous speech inputs and long conversation contexts effectively, while its singing capability was co-designed with Tencent Music Entertainment's (TME) Lingyin AI team. Users can trigger requests based on complex criteria such as artist, genre, release year, or popularity.

## BACKGROUND

Real-time speech models represent a shift from traditional multi-step pipelines (speech-to-text, text processing, text-to-speech) to direct end-to-end voice interactions that minimize latency and capture emotional nuances. Agentic tool-calling allows AI models to autonomously decide when and how to utilize external tools like web search engines to fulfill user requests during a conversation.

## REFERENCES

## KEYWORDS

#AI#Speech Model#Tencent#Real-time#Conversational AI

$ subscribe --daily