Upload a portrait with one clear face to an AI talking-photo tool, add the song as the audio, then generate the video and download it. Magic Hour takes a song as its audio input and syncs the lips and face to it. Akool bills its talking photo as talking and singing. Start with one visible face.
From still photo to singing video
This walk-through follows Magic Hour, because, according to its own page, it's the one that sets out the whole route with a song as the audio.
Choose the photo. One person, face clearly visible. Magic Hour supports making one face talk per photo, so a group shot or a couple will give you one singer at most.
Upload it. Drop the portrait into the talking-photo tool.
Add the song. The audio can be a song, speech or laughter, or one of its presets if you don't have a file to hand. The lips and face are synced to whatever audio you give it.
Generate, preview, download. Click Generate, preview the animation when it's ready, and download it once you're happy with how the mouth follows the words.
If you'd like to try a spoken line before a whole song, our guide to free ways to make a photo talk covers the talking side of these tools.
Three tools, three different starting points
Only one of these pages describes adding a song directly, so the table sets out what each one takes as its input and what free route it mentions, as of Nov 2026.
Want AI doing the heavy lifting in your marketing?
I build the systems that handle the boring 80 percent, so you get your week back. Done properly, with the human kept in.
| Tool | What you give it | Free route on its page | The limit to plan around |
|---|---|---|---|
| Magic Hour | A portrait plus a song, speech, laughter or a preset | Clips up to 3 seconds, no account | One face per photo; free results carry a watermark |
| Akool | A photo, with voice clone, lip sync and emotion controls | A free trial | Its feature list is about voices, not uploading a track |
| VEED | Any photo, artwork or drawing, plus a text description | Not set out on its talking-photo page | Movement is directed in words, not by a song file |
Magic Hour
Magic Hour is the direct route. You upload a portrait, add audio and it builds a talking video with lip and facial movement synced to it, and its page names a song as one of the audio types it takes. The free route, with no account, makes clips up to 3 seconds, and free results include a watermark. Three seconds is a line, not a chorus, so treat the free version as a test of whether your photo works. It animates one face per photo, which makes it the wrong pick for a duet or a band shot. The detail you wouldn't guess: it offers audio presets, so you can see a result before you've found the song.
If you're making a real person's photo sing for a laugh, this is where I'd start. You'll know within minutes whether the photo is usable.
Akool
Of these three, Akool alone puts singing in its headline: its Talking Photo tool is billed as talking and singing. The features it lists are voice clone, lip sync, customisable emotions and a voice library, so its strength is control over the voice and the expression. Voice clone is the detail that sets it apart. What its feature list doesn't spell out is a step for uploading a finished song, so check the audio options on the free trial before you plan a clip around a particular track. For a quick three-second joke, it's more tool than you need; for a performance where the voice and the mood matter, it's the one to try next.
VEED
VEED comes at it from a different angle. You upload any photo, then write a text description of how you want the image to speak and move, so you're directing it in words. The step its page describes is that text description, not an audio upload, which makes it the wrong choice if lip-syncing to a specific song is the whole point. Where it earns its place is the subject: VEED says its animation works with portraits, artwork, drawings and other image types. If your singer is a painting or a cartoon rather than a photo of a person, it's the one to open. Its talking-photo page doesn't set out a free allowance.
Which one to open first
For a song on a real photo, open Magic Hour, because its page is the one that says it takes a song. If the voice itself matters and you want emotion controls, move to Akool and confirm on the trial that your track goes in. If the subject is artwork and you're happy to describe the movement rather than sync to a file, use VEED.
For more tools built around matching mouths to audio, see our round-up of AI lip sync tools, and if you'd rather build a reusable character than animate one photo, our guide to AI avatar generators is the next step.
