[Tencent conference AI simultaneous interpreting online supports voice imitation, and the delay is reduced to 3 seconds] According to the monitoring of Beating, Tencent Conference officially launched AI simultaneous interpreting function, and the first phase supports Chinese English translation. The new function compresses the simultaneous interpreting delay to less than 3 seconds, and supports users to enable the function of "imitating your voice", so that AI voice translation can restore the voice characteristics of speakers. The traditional simultaneous interpreting is similar to the relay race, which requires the speaker to speak a complete sentence before the translation can begin. Tencent conference AI simultaneous interpreting adopts real-time translation mechanism to achieve almost synchronous speech and translation. In a multi person meeting scenario, enabling tone imitation not only makes the translator sound like the participant is speaking in a foreign language, but also helps the audience easily distinguish between different speakers and avoid being confused by a single machine synthesized sound. Users can also switch back to the system voice at any time, and freely adjust the proportion of the original volume and simultaneous interpreting volume. The newly launched simultaneous interpreting service is connected with the original real-time transcription and in session subtitle depth of Tencent Conference, and a complete AI conference workflow consisting of bilingual subtitles, text recording and simultaneous interpreting voice is constructed. This closed-loop process can accurately capture and instantly transmit cross lingual speeches, generate high-quality meeting context materials that can be read and mobilized by AI, thereby improving the accuracy of AI minutes, to-do extraction, and post meeting Q&A in cross lingual meetings. [Original link]