Skip to main content
NEW HERE30% OFFyour first month on any monthly plan, for 72 hours after you sign up
Guides Sound

Make an AI video with sound from a photo

ai-image-to-video.io team · Updated Nov 12, 2026 · 4 min read
The photo that went inVideo
Campfire at the lake, with sound · Veo 3.1 Fast · 8 s · 720p

A clip with its own sound feels finished in a way a silent one does not. Until recently that meant making the picture first and laying sound under it in an editor. Most video models on this site now make the sound in the same run: footsteps that land when the foot lands, a voice that matches the lips, a fire that crackles when the flames jump. You write it into the prompt next to the motion, and it costs nothing extra.

In short
  1. Most models here make sound with the picture. Hailuo 2.3 and FLUX 3 video do not.
  2. Write sound the way you write motion: what makes the noise, and how it sounds.
  3. For speech, put the exact words in quotes and keep them short.
  4. Already have the audio? Lip-sync a face to it instead.

Which models make sound

Sound comes with the picture and is included in the price shown on the button. Each model card in the generator says whether it makes sound. The two that make picture only are Hailuo 2.3 and FLUX 3 video; on those, sound words in the prompt are ignored. Gemini Omni always makes sound. The rest are asked for sound on every run, and your prompt decides what it is.

If you needPickCredits
Speech and ambience, both clips on this pageVeo 3.1 Fast40 · 720P 8s
Sound at the lowest priceMiniMax H315 · 480P 5s
Characters and pets with soundHappyHorse 1.1160 · 720P 5s
Sound on every clipGemini Omni 1.2 Flash30 · 360P 5s

The full list, with sound marked on each model, is on the models page.

Ambience: describe what makes the noise

Ambience is the easiest sound to get right, because it does not have to line up with a mouth. Name each source and how it sounds, the same way you would name what moves. "Nature sounds" gives the model nothing to aim at; "the fire crackles and pops, loons call across the lake" gives it two sounds and a place.

PROMPT WE USED

The fire crackles and pops, sparks rise, loons call across the lake. One continuous shot, no cuts, natural physics, stable faces and hands. Keep the first frame’s faces, clothes and light.

A lakeside fire, camera still, flames moving. Press the sound button to hear what the prompt asked for.
  • Name the source. "Rain on the window", not "rain sounds".
  • Pick two or three sounds. A long list turns into noise.
  • Say near or far. "Distant traffic" and "a kettle right beside her" sit differently in the mix.
  • Tie sound to motion. "Sparks rise and crackle" keeps the two in step.

Speech: short lines in quotes

For someone talking to the camera, write the line they say in quotation marks, after the action that comes before it. Keep it to one short sentence. An eight-second clip fits roughly a dozen spoken words with room for a pause; longer lines get rushed or cut off. Add the room the voice is in, so it does not sound like it was recorded in a box.

PROMPT WE USED

He puts them on, nods to the beat and says: "No, these are actually good." Room sound and faint music. One continuous shot, no cuts, natural physics, stable faces and hands. Keep the first frame’s faces, clothes and light.

He puts the headphones on and talks to the camera, the way a review clip on a phone looks. Turn the sound on to hear the line.
You wantWrite it like this
One line to cameraShe looks up and says: "This changed my mornings." Quiet kitchen sound.
A reactionHe laughs and says: "No way." Crowd noise behind him.
Narration-free ambienceWaves roll in, gulls call in the distance, wind on the microphone.
Music feelFaint music from a radio in the room, she taps along.
An animalThe dog barks twice and wags its tail, birds in the garden.

Generated music is a background feel, not a track you choose. If the clip needs a specific song or a voice-over you recorded, leave sound words out and add the audio in your editor, or use lip sync to make a face follow your own recording.

When the sound is off

What you seeWhyWhat to change
No sound at allThe model makes picture only, or nothing in the prompt makes a noise.Pick a model that makes sound and name what should be heard.
The words are cut offToo many words for the length.Shorten the line to one sentence, or use a longer clip.
The wrong person speaksSeveral faces in the photo, none named.Say who speaks: "the woman on the left says…".
Sound is a wash of noiseToo many sources in one prompt.Keep two or three sounds and say which is closest.
Try this prompt on your photoVeo 3.1 Fast · 8 s · 40 credits Open in Photo to Video

FAQ

Can AI make a video with sound from a photo?

Yes. Most video models on this site make sound in the same run as the picture. Describe what should be heard in the prompt, next to the motion.

Which models do not make sound?

Hailuo 2.3 and FLUX 3 video make picture only. Every other video model here can make sound with the picture.

Does sound cost extra?

No. Sound is included in the price shown on the Generate button before you run.

How do I make the person in my photo say something?

Write the line in quotation marks after the action, for example: she looks up and says: "This changed my mornings." Keep it to one short sentence.

Can I use my own voice recording?

Yes, with lip sync. Upload a photo or a clip of a face and a line of audio up to 10 seconds, and the mouth follows your words.

Keep reading

All guides

Your photo, moving in a few minutes.

Add a photo, paste the prompt from this guide or write your own, and press Generate.