What you actually get, and how long it really takes
Two expectations are worth correcting before you pay, because both of them cost people money or patience.
| What people expect | What is actually made |
|---|---|
| A video of the whole song | About twenty seconds — one section, the part worth sending |
| Ready in a couple of minutes, like the song | Ten to twenty minutes. The middle is around thirteen; the slowest measured was close to twenty-two |
| A landscape video for YouTube | Vertical, sized for Reels, TikTok, Shorts and status |
| The photo moves to the backing track | It is driven by the isolated vocal, which is why the mouth matches the words |
That last row is the part most tools skip, and it is the single biggest reason a result looks right or looks wrong. Before anything is animated, the singing is separated out of the finished mix and the photo is driven by the voice alone. In our own side-by-side tests the motion signal from the full mix was almost flat, while the isolated vocal gave roughly ten times as much for the face to work with. Drums and bass look like noise to a lip-sync model; a voice looks like speech.
The waiting is real work, and leaving the page is fine — it keeps going and the result lands in your library.
The photo decides almost everything
You cannot improve the animation after the fact, so the photo is where the whole result is decided. The one thing that matters more than lighting, resolution or background is how much of the frame the face occupies.
We measured this directly: in a head-and-shoulders photo the mouth came out nearly twice as wide in frame as the same person shot from the waist up. The model animates what it can resolve, and a mouth thirty pixels wide cannot be made to sing convincingly no matter how good the source photograph is as a photograph.
- Head and shoulders, filling the frame. Not a full-body shot, not a group photo cropped down.
- Facing forward, eyes open, mouth closed or slightly open. A wide laugh gives the animation a strange place to start from.
- Even light on the face. Hard shadow across one side reads as geometry and the animation follows it.
- Nothing across the mouth or eyes — no sunglasses, no hand, no microphone, no heavy fringe.
- One person. Two faces in frame means the model has to choose, and it does not always choose yours.
- A phone photo is plenty. Sharp and close beats high-resolution and distant, every time.
There is no example video on this page. Every one made here is a photograph of a real person singing, which is not ours to publish however good it looks. The written-up comparison on the photo-singing page shows what the differences look like without putting anybody's face on a public page.
The words on screen, and what to do if a line is wrong
The lyrics are burned into the picture rather than attached as a subtitle track, so they survive being posted anywhere — Reels and status screens strip subtitle tracks, and a caption file that never displays is the same as no captions at all. Five to eight lines is what a twenty-second section usually carries.
The fonts travel with the video rather than being borrowed from whatever machine renders it, which is what keeps non-Latin lyrics from coming out as empty boxes. If a line still looks wrong — a mis-split phrase, a name spelled the way it sounded rather than the way it is written — that is fixable on the finished video without paying again or regenerating the animation.
Before you spend anything, get the song right. The video is made from the song you already have, so a second take of the song is free and a second video is not.
Before you pay for a video
- The song is the take you actually want — the video is built from it
- Head and shoulders, filling the frame, facing forward
- Even light, nothing across the mouth or eyes, one person only
- You own the photo, or the person in it said yes
- Ten to twenty minutes of waiting is expected — leaving the page is fine
- Vertical and about twenty seconds: plan the post around that, not a full-length video
Written and fact-checked by the Tunely team · Updated