Song AI generates complete songs from a text description – melody, arrangement, synthesised vocals and mixing – in a stated 15 to 30 seconds. Around the generator sit lyrics-to-song for words you wrote yourself, an AI song cover that restyles an existing track, a lyrics generator, vocal removal and stem splitting, a MIDI editor and audio-to-MIDI, across 50-plus genres and multiple vocal styles and languages, exporting WAV, MP3 and FLAC.
The disclaimer is worth leading with. The site states plainly that it is not affiliated with, endorsed by or sponsored by Suno, Udio or any third-party AI provider – which is the opposite of the pattern this directory has been rejecting all week, where products adopt a model’s name and let the reader draw the conclusion. Saying it unprompted costs the vendor the halo and is the reason to trust the rest of the page.
Shipping a MIDI editor alongside generation is the substantive feature, for the same reason it mattered on Lacuna: generated music is wrong in one specific bar, and without note-level editing the only remedy is regenerating and losing what worked. There is a free tier with initial credits for personal use, paid plans carrying a commercial licence, and no figures published. No underlying models are named, no company entity appears, and generated vocals remain the giveaway in AI music whatever the genre.
