AIMPACT Message, May 20 (UTC+8): According to monitoring by Beating, real-time speech translation systems are evolving from monotone voice translation into multimodal digital interpreters capable of understanding visual content and cloning voices. On May 19, Alibaba’s Tongyi Lab officially unveiled its next-generation real-time audiovisual translation large model, Qwen3.5-LiveTranslate, significantly upgrading real-time translation capabilities to over 3,500 language pairs and introducing, for the first time, real-time voice cloning, custom hotword injection, and visual understanding. Built on the Qwen3.5-Omni architecture, the new model now supports comprehension and generation in 60 languages and speech output in 29 languages. Unlike traditional translation software that relies solely on audio, the new model incorporates real-time visual context to eliminate semantic ambiguity. For example, when a specific mask appears in the video, the system leverages visual features to accurately distinguish between a medical mask and a masquerade mask in English, compensating for missing audio information. To mitigate transcription errors caused by noise and accents, the model also introduces a dynamic hotword injection mechanism: users can directly specify names, brands, or industry-specific terms within the translation stream to lock in accurate translations and prevent proper nouns from drifting during live translation. During cross-language translation, the model also supports real-time voice cloning, enabling it to replicate the original speaker’s tone and timbre within the live translation stream. The new model is currently available on the Qwen Omni demo platform, with future API access planned for Alibaba Cloud’s Bailian platform. (Source: BlockBeats)
Alibaba Launches Qwen3.5-LiveTranslate, Supporting 3,500 Language Pairs
KuCoinFlashShare
Summary
Alibaba’s Tongyi Lab has launched Qwen3.5-LiveTranslate, a real-time audio-video translation model supporting over 3,500 language pairs. The model enhances translation accuracy with visual context understanding, real-time voice cloning, and hotword customization. It is now available on the Qwen Omni experience platform, with API access coming soon to Alibaba Cloud’s Bailing platform. This release aligns with growing interest in real-world assets (RWA) and could support new token listings in multilingual markets.
Source:Show original
Disclaimer: The information on this page may have been obtained from third parties and does not necessarily reflect the views or opinions of KuCoin. This content is provided for general informational purposes only, without any representation or warranty of any kind, nor shall it be construed as financial or investment advice. KuCoin shall not be liable for any errors or omissions, or for any outcomes resulting from the use of this information.
Investments in digital assets can be risky. Please carefully evaluate the risks of a product and your risk tolerance based on your own financial circumstances. For more information, please refer to our Terms of Use and Risk Disclosure.