Tencent's Hy ASR 3.0 Understands What You Meant, Not Just What You Said
Tencent's new ASR model plugs directly into Hy3's language brain, hitting ~3% WER across Mandarin, English, and Cantonese while understanding context, not just words.
- Hy ASR 3.0 preview launches on Tencent Cloud with API access for customer service, meeting transcription, and voice search.
- ~3% multilingual WER: Mandarin 3.34%, English 2.62%, Cantonese 3.12% on open benchmarks.
- Powered by Hy3 LLM: the ASR layer inherits Hy3's 295B MoE semantic reasoning for context-aware transcription.
- Four key upgrades: general accuracy, homophone correction via context, hotword injection, and noisy/whispered audio robustness.
- Already live in Yuanbao (Tencent's consumer AI assistant) with dialect recognition and context correction free for users.
- Architecture: MoE design with a proprietary unsupervised speech encoder trained on tens of millions of hours of unlabeled audio.
Most speech recognition systems do one thing: turn audio into text, word by word, and stop there. Tencent Hunyuan's Hy ASR 3.0 preview takes a different approach. By wiring its ASR pipeline directly into the language understanding capabilities of Hy3 -- Tencent's latest large language model -- it produces transcripts that reflect what you meant, not just what you said.
Hy ASR 3.0 preview leverages the language comprehension capabilities of Hy3, combining high-precision recognition with deep semantic understanding. The practical result: it can provide more accurate and coherent transcriptions by combining context, achieving an evolution from word-by-word transcription to semantic understanding.
The numbers
On open benchmarks, word error rates reach 3.34% for Mandarin, 2.62% for English, and 3.12% for Cantonese, achieving around 3% overall with impressive multi-dimensional performance. That puts it in competitive territory with the current open-source ASR field, where NVIDIA's Canary Qwen 2.5B currently tops the Hugging Face Open ASR Leaderboard with 5.63% WER -- though that benchmark is English-focused and not directly comparable.
Tencent also reports strong results on its internal evaluation sets, covering general recognition, dialect recognition, context understanding, domain-specific vocabulary, and complex acoustic scenarios like high noise and whispered speech.