Finding 01 · Vocals
Do AI music creators choose the singer, or let the model decide?
They let the model decide — by a landslide. Of all 3,428 tracks, 74.5% were generated with the voice left on "AI decides." Only a quarter of creators specified a vocal at all — and when they did, they chose a male voice 2.6× more often than a female one.
The AI music generation isn't just writing the song — for three out of four creators, it's casting the singer too.
The male-skew (2.6:1 among creators who set a voice) mirrors a known pattern in default AI voice output and creator preference. It's one of the clearest signals in the dataset that "hands-off" is the default posture.
Finding 02 · Genre
What genres do AI music creators actually make?
Pop leads — but the field is remarkably even. Ranked by the share of creators who made at least one track in each genre, Pop, Electronic, and Hip-Hop cluster at the top, with R&B, Lo-Fi, and Rock close behind. AI music isn't one sound; it's a broad, mainstream spread.
Share of the 739 creators who made ≥1 track in each genre (creators can appear in more than one). Ranking creators — not raw track counts — keeps a few high-volume users from distorting the picture.
Finding 03 · Language
What languages is AI music made in?
AI music is global. Creators wrote songs in 36 different languages. English dominates at 95.5% of creators, but nearly 1 in 11 creators (8.8%) made at least one non-English track — from Spanish and Korean to Arabic, Ukrainian, and Bislama.
Non-English creators ranked by number of distinct creators. English: 706 creators (95.5%).
The long tail is the real story — songs were also made in:
Finding 04 · Influence
Which artists do AI music creators model their songs on?
One in three creators (31.7%) named a specific artist as an influence when making a track. Ranked by how many distinct creators cited each one, The Weeknd leads, followed by Billie Eilish, Drake, Taylor Swift, and Bad Bunny — a lineup that reads like the actual 2020s pop-culture canon.
Number of distinct creators who named each artist as an influence when generating a track. These figures reflect what creators typed as inspiration — they are not affiliations with, or statements about, the artists themselves.
Finding 05 · Workflow
Do creators want AI to write the whole song, or help with their own?
They want a collaborator, not a ghostwriter. Among creators who chose how to handle lyrics, 56% asked the AI to enhance a draft they'd written, 30% locked in their exact words, and only 14% asked it to write from a loose idea. And AI music isn't disposable: more than a quarter of sessions produced a variation or a full album, not a one-off single.
The winning workflow of 2026 isn't "AI, write me a song." It's "here's mine — make it better."
Lyrics-handling among the 1,344 tracks where a mode was set. Separately, of all 3,428 tracks: 72.9% single, 13.7% variation, 13.4% full album — so 27.1% extended beyond a single track.
Methodology
How we measured this
What's in the dataset
- 3,428 AI music generations created by 739 distinct creators on HookGenius between April and July 2026.
- Anonymized + aggregate. We analyze only the creative metadata of a generation — genre, vocal setting, language, artist influence, and workflow mode. No names, no lyrics, no personal data appear in this study.
- Internal and test accounts were excluded so the numbers reflect real creators.
How we kept it honest
- Distinct-creator weighting. Rankings (genres, artists, languages) count how many different creators did something — not raw generation counts — so a handful of high-volume users can't distort the trend.
- Stated denominators. Where a field is optional (genre, lyrics mode), percentages are taken against the tracks that specified it, and we say so. Vocal, language, and generation-mode figures are against all 3,428 tracks.
- Primary source. Every number here is computed directly from HookGenius's own generation data — this is first-party research, not a survey or an estimate.
Cite this study
Free to reference with attribution (CC BY 4.0). Copy the citation:
Make music that sounds like the artists above
This study is built from real HookGenius data. HookGenius writes the full lyrics + a Suno-ready style prompt that actually sounds like the sound you're going for — in seconds.
Generate your first track free →