Newshot Tencent Hunyuan Releases Real-Time Speech Model Hy Realtime: It Can Talk, Sing, and Call Tools, Now Live in Yuanbao Voice Calls, Tencent Customer Service, and More
IT之家 November 13 news, Tencent Hunyuan today released its new real-time speech model Hy Realtime. The model performs excellently in foundational conversational capabilities such as speech understanding, instruction following, and anthropomorphic expression, while adding Agentic tool calling and singing capabilities.
Hy Realtime is already live in products including Yuanbao voice calls and Tencent Customer Service, and will soon launch on Tencent Cloud to provide API services externally.
According to the official statement, Hy Realtime is still being continuously iterated, and the next step will focus on areas such as duplex interaction, long-term personalized memory, and foreground/background Agent interaction to continuously improve and bring users a better interaction experience.
The current version can be experienced through the latest version of the Yuanbao App by selecting the "Chuyu" voice in the voice call feature; some capabilities can also be experienced via the web Demo.
- https://rtcube.cloud.tencent.com/component/hunyuan/index.html
IT之家 attaches the official introduction as follows:
Hy Realtime performs excellently in understanding user intent and following instructions. In real-world communication, user interactions may be slurred, cut off mid-sentence, or self-corrected midway; the model can combine conversational context to reconstruct the intent of such incomplete input, with significantly improved understanding accuracy. At the same time, the model's multi-turn instruction following and context management capabilities are further enhanced, and it can still stably understand user intent and maintain conversational fluency in longer continuous dialogues. Beyond basic conversational functions, Hy Realtime also has tool calling capabilities; it can accurately determine whether to perform a web search based on conversational context and user intent, unifying search decisions, tool calls, and answer generation. In addition to the major improvement in "listening comprehension," Hy Realtime's spoken expression capabilities have also been further upgraded. It is comprehensively superior to the previous generation model across dimensions including sound quality, pronunciation, prosody, and emotion; emotional changes and progression within a conversation are more natural, prosody and intonation handling are more diverse, and empathy with users is further enhanced. Relying on deep Co-Design with Tencent Music TME Lingyin AI team, Hy Realtime adds singing capability, integrating natural-language song requests and expressive singing into conversations, taking interaction from "able to speak" to "able to sing." Users can request songs by combining singer, album, genre, language, release year, and popularity, and can also continue selecting based on prior context. For example, when a user says "play an English folk song from between 2010 and 2020," the model can combine multi-turn context to understand the song request intent, match the song, and sing it. In terms of singing performance, the Hunyuan team has continuously refined expressiveness, naturalness, and musicality, making singing emotions more layered, articulation and breath transitions more natural, and melody and rhythm expression more coordinated. In internal expert A/B Test evaluations, Hy Realtime shows significant advantages in core experience dimensions. ▲ In application scenarios mainly involving Chinese / mixed Chinese-English, Hunyuan speech has built an experience-oriented real product evaluation system. In the results of internal expert blind evaluations, there has been significant improvement in basic search-and-answer experience. ▲ On the internally built foundational intelligence test set covering multiple dimensions such as mathematics, science, logical reasoning, knowledge Q&A, complex instructions, text understanding, and multi-turn, the non-thinking mode performs well in Chinese. ▲ On multiple public Benchmarks in the speech interaction field, its comprehensive objective performance is at a leading level.
Hy Realtime performs excellently in understanding user intent and following instructions. In real-world communication, user interactions may be slurred, cut off mid-sentence, or self-corrected midway; the model can combine conversational context to reconstruct the intent of such incomplete input, with significantly improved understanding accuracy. At the same time, the model's multi-turn instruction following and context management capabilities are further enhanced, and it can still stably understand user intent and maintain conversational fluency in longer continuous dialogues.
Beyond basic conversational functions, Hy Realtime also has tool calling capabilities; it can accurately determine whether to perform a web search based on conversational context and user intent, unifying search decisions, tool calls, and answer generation.
In addition to the major improvement in "listening comprehension," Hy Realtime's spoken expression capabilities have also been further upgraded. It is comprehensively superior to the previous generation model across dimensions including sound quality, pronunciation, prosody, and emotion; emotional changes and progression within a conversation are more natural, prosody and intonation handling are more diverse, and empathy with users is further enhanced.
Relying on deep Co-Design with Tencent Music TME Lingyin AI team, Hy Realtime adds singing capability, integrating natural-language song requests and expressive singing into conversations, taking interaction from "able to speak" to "able to sing."
Users can request songs by combining singer, album, genre, language, release year, and popularity, and can also continue selecting based on prior context. For example, when a user says "play an English folk song from between 2010 and 2020," the model can combine multi-turn context to understand the song request intent, match the song, and sing it.
In terms of singing performance, the Hunyuan team has continuously refined expressiveness, naturalness, and musicality, making singing emotions more layered, articulation and breath transitions more natural, and melody and rhythm expression more coordinated. In internal expert A/B Test evaluations, Hy Realtime shows significant advantages in core experience dimensions.
Advertising Statement: External redirect links contained in the article (including but not limited to hyperlinks, QR codes, passcodes, and other forms) are used to convey more information and save selection time; the results are for reference only. All articles on IT之家 containing external links include this statement.
Related Articles
- Tencent 李锐章 talks about Yuanbao search business model: does not exclude advertising, but users should not have to look for answers in ads
- Tencent's brand-new search product "Yuanbao Search" officially launches: can search WeChat Official Accounts and Channels, with ads and answers clearly distinguished
- Tencent Hunyuan: Hy4 calls surge, WorkBuddy has expanded tenfold, Hy3 free trial extended
- Tencent Hunyuan Hy4 official version released, claimed to rank among the strongest open-source models for the first time and built for real productivity
- Sources say Tencent will launch an AI search engine for C-end users: minimalist design, strengthened ranking of authoritative sources
- Tencent Hunyuan releases Hyra 2.0: building reusable tools for model R&D