About this tool

Seed Audio 2.0 is an upcoming AI audio generation model built for text-to-audio, video-aware dubbing, and multi-stem scene rendering. It supports up to six minutes per scene, six reference voices, thirty languages, and separate stems for dialogue, music, ambience, and effects. Early-bird pricing is available ahead of launch, and the current Seed Audio 1.0 studio can be used to rehearse prompts.