Search AI Products and News

Alibaba has released EMO (Emote Portrait Alive), a portrait video generation framework that can create videos with rich facial expressions and head poses. The EMO technology outperforms existing state-of-the-art methods such as DreamTalk, Wav2Lip, and SadTalker. The EMO model could potentially be used as a tool for criminal activities.

Alibaba's EMO and OpenAI Sora's text-to-video models are gaining popularity. The EMO framework enhances the realism and naturalness of top video generation. EMO allows for the creation of dynamic, expressive AI character avatars. The EMO method supports the generation of songs and spoken audio in different languages, enabling interaction between various characters.

["Alibaba's latest launch is the audio-driven portrait video generation framework EMO","EMO can generate videos of any duration based on input audio","The team behind EMO comes from Alibaba's Intelligent Computing Research Institute"]

A Douyin influencer reveals that the frequently heard AI voices in the emo community are actually derived from real human voice sources. The influencer's husband, having heard her vocal catchphrases many times, now actively avoids them. The entertainment influencer has edited and shared these sounds on Xiaohongshu, receiving great reviews. The AI voices are actually synthesized using real human voices through deep learning, with the 'Yunxi' voice source coming from Microsoft's Azure Text-to-Speech service.