Tencent Meeting's AI Simultaneous Interpretation Feature Goes Live, Supporting Voice Mimicry with Latency Reduced to Within 3 Seconds

According to Datamation Beating monitoring, Tencent Meeting has officially launched an AI simultaneous interpretation feature, initially supporting Chinese-English translation. The new feature will compress the interpretation time lag to within 3 seconds and support users to activate the "Mimic Your Voice" function, enabling the AI voice translation to replicate the speaker's voice characteristics.

Traditional simultaneous interpretation is similar to a relay race, where the translator can only start after the speaker finishes a complete sentence. Tencent Meeting's AI simultaneous interpretation adopts a real-time translation mechanism, achieving nearly synchronous speech and translation. In a multi-person meeting scenario, enabling voice mimicry can not only make the translation sound like the participant speaking in a foreign language but also help the audience easily distinguish between different speakers, avoiding confusion from a single machine-generated voice. Users can also switch back to the system voice at any time and freely adjust the original voice volume and interpretation volume ratio.

The newly launched interpretation service is deeply integrated with Tencent Meeting's existing real-time transcription and in-meeting subtitles, establishing a complete AI meeting workflow composed of bilingual subtitles, text records, and simultaneous interpretation voices. This closed-loop process can accurately capture and instantly deliver cross-language speeches, generate high-quality meeting contextual materials that can be read and accessed by AI, thereby improving the accuracy of AI summaries, action item extraction, and post-meeting Q&A in multilingual meetings.