
1. Words
Lyrics are split into syllables and stressed beats.
Song AI works in four stages: words, melody, band and voice. Each stage learns patterns from music, then follows your lyrics and style to build a finished track.
Every song AI request passes through these stages in seconds.

Lyrics are split into syllables and stressed beats.

A tune is shaped so key words land on strong notes.

Drums, bass and chords follow the genre and tempo.

A singer performs the melody with your words.
Both tracks sing identical lyrics. Only the Style line changed.
Synth pop, female vocal
Airy voice over shimmering pads.
Acoustic ballad, male vocal
Warm baritone with piano and upright bass.
Use AI for songs with these four inputs.
| Stage | Your input | Where |
|---|---|---|
| Words | Lyrics with section tags | Lyrics field |
| Melody | Mood and tempo | Style field |
| Band | Genre and instruments | Style field |
| Voice | Male or female, tone | Style field |

Split any finished song to study the vocal.
Lead vocal and backing vocals apart.
Two voices on separate tracks.
Each singer on a track.
Paste words, describe the sound and listen.

Open the Music Generator in Custom and paste lyrics with [Verse] and [Chorus] tags.

Describe the genre, instruments, tempo and voice in Style, choose a length and press Create for two sung takes.
Yes. You can start free in your browser and get two complete sung takes from every idea. All the songs AI writes for you are yours to keep and compare.
It splits each line into syllables and places stressed syllables on strong beats, then shapes the tune around them. Short, even lines sing most clearly.
Yes. Name the voice in Style, such as warm male baritone or airy female vocal.
Each take explores a fresh melody and arrangement from the same input, so you can pick the one that fits best.
Yes. [Verse], [Chorus] and [Bridge] tags set the order, and the chorus returns where you place it.
Yes. The Stem Splitter separates the voice from the band.
Paste a verse and a chorus and compare two finished takes.