Baidu reportedly launched a brand new synthetic intelligence (AI) video technology mannequin on Wednesday. As per the report, the MuseStreamer AI mannequin may also combine Chinese language audio within the generated movies, making it the second such mannequin after Google’s Veo 3. The tech large claims it to be the world’s first AI mannequin with native Chinese language audio technology assist. Alongside the introduction of the big language mannequin (LLM), the corporate reportedly additionally launched a brand new video content material creation platform dubbed HuiXiang. Notably, neither MuseStreamer nor HuiXiang is at present out there exterior of China.
Baidu’s MuseStreamer Can Reportedly Generate Chinese language Audio
The world of AI video technology mannequin has advanced considerably within the final two years. Now we have moved from fashions that struggled to generate individuals with a set variety of fingers to LLMs which might now precisely depict sensible physics and movement. Nonetheless, one space most AI gamers have avoided coming into was movies that additionally supported audio natively.
At Google I/O 2025, the tech large turned the primary firm to supply this functionality with Veo 3, which instantly turned discuss of the city, leaving its largest rival, OpenAI’s Sora, behind. The Mountain View-based tech large just lately expanded Veo 3 in all of the 154 international locations the place the Gemini app is accessible, highlighting the corporate’s aggressive push for this device.
Nonetheless, in accordance with a Tech in Asia report (via AI Base), Chinese language tech large Baidu has additionally entered the race with its MuseStream AI mannequin. It’s stated to generate movies with Chinese language audio, and the one mannequin with the potential to take action. Notably, Veo 3 can solely generate audio within the English language.
MuseStreamer can reportedly not solely generate dialogues which can be synced with the movies, it could possibly additionally add sound results and ambient noises within the movies. Baidu is claimed to have claimed that the mannequin achieved a rating of 89.38 p.c on the VBench I2V benchmark, rating on the high. The tech large is pitching the LLM as a content material creation device for shoppers.
Alongside the AI mannequin, Baidu has reportedly additionally launched a brand new video content material platform dubbed HuiXiang. HuiXiang is claimed to function the front-end for the AI mannequin, the place customers can share prompts and generate movies. The platform at present helps 10-second-long video generations at 1080p decision, the report acknowledged. As compared, Veo 3 can generate solely eight-second-long movies. There isn’t any readability over the default facet ratio of the video, and if customers can generate movies in several facet ratios.