88Z AI, 20 May — Alibaba Tongyi Qwen releases the real-time simultaneous interpretation model Qwen3.5-LiveTranslate, supporting multi-language audio input and output with low-latency translation, voice cloning, and industry terminology optimization.
At a glance
- Supports audio input in 60 languages and audio output in 29 languages, with end-to-end latency as low as 2.8 seconds
- Features real-time voice cloning and includes optimization for 1,000 industry-specific hot words
The update
Alibaba Tongyi Qwen has officially launched the real-time simultaneous interpretation model Qwen3.5-LiveTranslate. Built on the Qwen3.5-Omni architecture, this model enables efficient cross-language communication. It supports audio input in sixty languages with text output, as well as direct audio output in twenty-nine languages. The end-to-end average latency per character is as low as 2.8 seconds, ensuring real-time interaction. Furthermore, the model features real-time voice cloning capabilities, preserving the original speaker's vocal characteristics to make the translated speech more natural and recognizable. To improve translation accuracy in professional scenarios, the model includes an optimization mechanism for one thousand hot words and industry-specific terminology, better handling specialized vocabulary in specific fields.
Sources & editorial note
This article is an AI-assisted synthesis by 88Z AI based on the public sources below. Product claims and availability are attributed to their publishers. It does not represent independent testing or first-hand reporting.
- 通义实验室Source published
Source dates above refer to the original announcements; the date under the headline is this article’s publication on 88Z AI.