How to Animate a Photo with AI with Sound and Voice: 2026 Guide

Animating a photo with AI with sound means turning an image into video: you upload the photo as the start frame of a video model, describe in words what should move, and the audio is generated together with the clip. The simplest way is Kling 3.0 or PixVerse V6: both have sound on by default and make clips from 1–3 up to 15 seconds. To add a voice that says your text, upload the portrait and a voice-over to Kling Avatar or OmniHuman 2.0; to animate a photo to music, use Seedance 2.5 with the track passed as the @Audio1 reference. A picture or a drawing comes to life the same way. Each takes a few minutes in the IIshki studio, paid with tokens.

Which AI model to pick for animating a photo

If you don't know where to start, take Kling 3.0: it is the most predictable option for a single photo, and the sound comes in the same clip.

If you need to repeat the motion from an existing clip, Kling Motion Control is a proven, working option, but for a better result use the Swap mode: upload a 4 to 30 second clip and a photo, and the character repeats the poses, facial expressions and lip movements, while the original sound is kept. No prompt needed.

How to animate a photo with AI: 5 steps

1

1. Prepare the photo

The model works with what it sees. The cleaner the source, the fewer artifacts:

• Upload the original, not a screenshot or a copy recompressed by a messenger.
• The face should be well lit and take up a noticeable part of the frame.
• One person in the frame animates more reliably than a group.
• Restore an old or small photo first: see the guide on how to improve photo quality.

2

2. Pick a model and upload the photo as the start frame

Open the model in the studio and upload the photo into the start frame slot. In Kling 3.0 and PixVerse V6 it becomes the first frame of the video; in Seedance 2.5 the photo can also go in as the @Image1 reference that you mention in the prompt.

3

3. Describe the motion, not the picture

The classic beginner mistake is describing what is already in the photo. The model sees it. The prompt should describe what changes:

Portrait: the person slowly smiles and turns the head slightly toward the camera, natural facial expression, soft light, static camera.

Formulas that work:

• Portrait “slowly smiles”, “blinks and looks into the camera”, “the wind gently moves the hair”.
• Landscape or city “cars drive by, people walk”, “clouds drift slowly, the water glitters”.
• Camera “slow push-in”, “static camera”, “light parallax”. One camera move per clip.

For sound, add one sentence about the audio: “street sounds”, “soft rain”, “background music without lyrics”.

4

4. Set the duration and format

For Stories and Reels use 9:16, for YouTube and slides 16:9. Keep the duration short: 5 seconds for a portrait is almost always better than 10 — on longer clips the chance of distortion grows, and so does the cost. The generation cost is shown on the model card before you start.

5

5. Run it and download the result

Generation takes from one to several minutes. The finished video is saved in “My generations” and, if Telegram is linked, also arrives as a file in the bot. If you don't like the result, edit the prompt, not the photo: usually it is enough to soften the motion or drop an extra detail from the description.

How to animate a photo with sound, a voice or to music

“Animate a photo and add a voice” means one of three things: you need the live sound of the scene, you need specific words from the person in the photo, or you need the motion to hit the music. Each has its own route.

Route 1. The video model generates the sound

The fastest option: one generation, nothing to stitch. In Kling 3.0 and PixVerse V6 the Sound setting is on by default — just check you didn't switch it off. Put the line in quotes and keep it short: the model fits the speech into the clip length.

The woman in the photo looks up at the camera, smiles and says: “Hi, this is me twenty years ago.” Soft café noise in the background, static camera.

Here the model picks the voice: you can't set the timbre exactly, only describe it — “a low, calm male voice”.

Route 2. A voice with your text

When the exact words and a specific voice matter — a greeting, an announcement, a host — split the work into two steps:

1. Voice the text in MiniMax: it has Russian voices and a speech speed setting. Your own phone recording works too. Download the audio file.
2. Open Kling Avatar (audio from 2 to 300 seconds) or OmniHuman 2.0 (up to 5 minutes) and upload the portrait and that file. The clip length follows the track, and lips and expression adapt to the speech.

Making a series? Use the same voice in every voice-over, otherwise the character will sound different from clip to clip. A detailed breakdown with prompts is in the guide how to make a talking photo.

Route 3. Animate a photo to music

For a music clip, a Story set to a track, or a greeting with a song:

1. Take a track: your own or a generated one. In Seedance 2.5 one audio reference is up to 30 seconds, so cut the part you need in advance.
2. In Seedance 2.5 switch to the References mode, upload the photo and the audio, and mention them in the prompt:

The couple from @Image1 slowly turns in a dance to the beat of @Audio1, a light breeze in their hair, warm evening light, the camera smoothly circles around them.

Seedance 2.5 holds up to 30 seconds — enough for a verse or a chorus. If the person in the frame sings and the lips must match the words, use OmniHuman 2.0: it builds the facial motion from the song itself.

How to animate a picture or a drawing with AI

A picture, a drawing, an illustration or AI art comes to life the same way as a photo: the image goes into a video model as the start frame. The difference is in the prompt and in how the model treats the style.

Three rules for drawn images:

• Name the style at the start of the prompt. Otherwise the model pulls the picture toward realism, and the drawing turns into an “almost photo”.
• Less motion than for a photo. Drawn faces have no “correct” anatomy, and a sharp head turn breaks the features faster.
• You can animate the background instead. For landscapes and postcards it often works better to move the world around the character: water, clouds, leaves.

Watercolor drawing, keep the style: a girl with a balloon slowly lifts her head to the sky, the balloon sways in the wind, clouds drift by, static camera. Sound: a light wind and distant birds.

Common mistakes and how to fix them

What's next

An animated photo is usually the first step. If you want the person in the photo to talk, see the guide how to make a talking photo; for a dance from a photo — AI dance video from a photo; and a comparison of video models by quality and sound is on the page best AI tools for photo animation.