Give it a subject, a feeling, or a half-finished line, and get back a full song: labelled verses, a chorus built to repeat, and a bridge that turns the story. It writes to a consistent meter so matching sections can be sung to the same melody, and it comes with a suggested key, tempo, and rhyme scheme.
Songs come out best from a specific scene rather than a big abstract feeling. Compare "heartbreak" to any of these.
Most weak lyrics fail for the same two reasons: they name feelings instead of showing them, and their second verse repeats the first.
The lyrics are yours to use, including commercially. Bear in mind that AI-generated text may not be copyrightable in some jurisdictions, which affects your ability to stop others using it. If the song matters to you, treat the output as a strong draft and make it your own.
No, this writes lyrics. It suggests a key and tempo to aim at, and keeps the meter consistent so matching sections fit the same melody, but you or your producer write the music. The output pastes cleanly into tools like Suno or Udio if you want a demo.
Because specific scenes produce better songs. "Heartbreak" gives the model nothing to hold; "packing the last box at 2am" gives it objects, a time, and a room, and concrete images are what make a lyric land.
It will follow a genre and a mood, and you can describe the feel you want. It won't imitate a named living artist's voice, and copying a specific artist's style closely is a good way to end up with something you can't release.
They mark the song's arrangement: [Verse 1], [Pre-Chorus], [Chorus], [Bridge] and so on. Anyone recording the song reads those to know the structure, and they're the format tools like Suno expect.
Yes, and it helps. Paste your line or your chorus into the box and describe where you're stuck, and it will build around what you have rather than starting over.