A photo of your nan can sing her favourite song. Your dog can perform a power ballad. Your kid's drawing can front a band. Making a picture sing used to be a party trick with rubber-banded mouth flaps; now it is a proper lip-synced performance, and it takes about five minutes.
What "making a photo sing" actually means
You give the tool two things: a picture with a face in it, and a song. The AI detects the face, works out how the mouth, jaw and head should move for the vocal, and animates the photo so the person (or pet, or drawing) performs the track. Done well, the eyes blink, the head moves with the phrasing, and the mouth shapes actually match the words.
There are two levels of this:
- Lip sync: one continuous close-up shot of the photo singing. Quick, cheap, and perfect for a birthday message or a meme.
- Full music video: the singer from your photo performs across multiple generated scenes, cut to the lyrics, in a visual style you choose. This is the one that looks like an actual music video rather than an animated photo.
Muzie does both. You choose per video.
Step by step
- Pick your photo. Front-facing, decent light, face clearly visible, nothing covering the mouth. A sharp phone photo is ideal. It does not have to be a person: pets, drawings, dolls and statues all work, anything with a face the model can find.
- Pick the song. Upload any audio file, choose from the built-in library, or describe a song and have an original generated with vocals. Then trim to the section you want, usually the chorus.
- Choose the mode. Lip sync for a single-shot clip, full music video for a multi-scene performance.
- Choose a style (full video mode). Anime, claymation, 35mm film, or type your own direction.
- Generate, check, share. Review the result, regenerate anything that looks off, then download or share straight to TikTok, Reels or Shorts. It renders in portrait, ready to post.
Making it look good
- The photo is 90% of the result. Bad light in, uncanny video out. If a result disappoints, change the photo before anything else.
- One face per photo works best for a solo performance. Group photos are their own feature: each face can be detected and animated as part of a duet or group number.
- Short beats long. A fifteen-second chorus hits harder than three minutes of it, costs less, and fits where you will post it.
- Lean into the absurd. The gap between subject and song is the joke. A serious cat singing a dramatic ballad outperforms a person singing it straight.
What people use it for
Birthday and anniversary messages where the birthday person sings their own greeting. Pet videos, which are their own genre at this point. Memorial tributes set to a loved one's song. Kids' drawings brought to life. And for musicians, the serious use: turning a press photo into a performance video for every release.
Whose photo can you use?
Yours, and anyone's who has said yes. Do not make singing videos of people without asking, and keep well away from using someone's face to mislead. Pets file no complaints.
Try it with the photo already on your camera roll: upload it, pick a song, and your first singing video is free.