Somio generates original songs from text, supplied lyrics, reference tracks or a described idea. It produces acapella and instrumental separately, offers a My Voice option for custom vocals, and edits through remixing, extending and replacing individual sections. Beyond audio it makes music videos, singing and dancing videos and lyric videos, and handles vocal removal and stem splitting, across multiple languages.
Generating without a length limit is the constraint most competitors do not lift, and it is the one that decides whether the output is usable. The standard AI song arrives at two or three minutes and then stops because the model stops, which is fine for a demo and useless for anything scored to a video or a set. Full-length generation means the song ends where the song ends.
Section replacement is the natural companion to that: on a long track, one bad verse should not cost the whole render. New users get 10 credits free, paid plans exist without published figures, and generated music is described as copyright-free for personal and commercial use. Treat that last claim carefully – it is the vendor’s assurance about its own output, not a settled question about what the model was trained on, and the risk on a commercial release sits with the person publishing it. No company entity is named either.
