Tencent Hunyuan has released its new-generation speech recognition model, Hy ASR 3.0 preview, achieving low word error rates across multiple languages. The model demonstrated a WER of approximately 3% for multilingual tasks, with specific rates of 3.34% for Mandarin Chinese, 2.62% for English, and 3.12% for Cantonese. This release is part of Tencent Hunyuan's ongoing efforts in AI development, alongside other tech giants like Alibaba and ByteDance focusing on AI for office applications. AI
IMPACT Enhances speech recognition capabilities for multilingual applications, potentially improving user interaction with AI-powered office tools.
RANK_REASON This is a product release from a major tech company, but not a frontier model release.
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
