How to Make an Old Photo Talk
One portrait, one recording, and a minute of AI — and they are speaking again.
A photograph holds a face; a recording holds a voice. Bring the two together and you can make an old photo talk — a lip-synced video where the person in the picture moves their mouth in time with real words. This guide walks through the full workflow with our talking photo tool: restoring the photo, pairing it with audio, generating the video, and doing all of it ethically.
What a Talking Old Photo Requires
A front-facing portrait in JPG, PNG or WEBP (up to 10MB), with the eyes and mouth clearly visible. Very old or damaged photos should be restored first — the better the face reads, the more convincing the speech.
An uploaded audio file (MP3, WAV, M4A or AAC, up to 10MB, 5 seconds to 10 minutes) or a fresh microphone recording. Any language works — the model syncs lips to the sound of speech itself.
How It Works: 3 Steps
Restore the photo first
A talking old photo starts as a readable face. Scratches, fading and blur confuse lip-sync models just as much as they confuse us. Run the picture through AI Photo Restore, and if it is small, upscale it — then the mouth and eyes have enough detail to animate.
Upload the photo and the voice
Open the talking photo generator and upload a front-facing portrait (JPG, PNG or WEBP, up to 10MB). Then add the audio: upload a saved recording (MP3, WAV, M4A or AAC) or record straight from your microphone — anything from 5 seconds to 10 minutes, in any language.
Generate and download the video
Click generate and the AI animates the face in sync with the voice — mouth movements matched to the words, natural expressions, subtle head motion. A couple of minutes later your talking photo is ready to download as a video.
Why People Make Talking Photos
An old portrait paired with a saved voicemail or family video audio becomes a short clip of them speaking — the voice real, the motion generated.
Read a letter or diary entry aloud, then let the photograph of its author deliver it. A striking way to share family stories with younger relatives.
At memorial services, some families play a short talking photo alongside the photo slideshow. If you go this route, keep it short and check with close family first.
The same tool works for creators: narrate a story in your own image, build a talking greeting card, or prototype a presenter clip without a camera.
Tips for a Lifelike Talking Photo
- Pick the most frontal photo you can find — eyes and mouth fully visible. Profiles and tilted heads lip-sync noticeably worse.
- Use clean, even audio. Background noise, music and heavy reverb all degrade how well the mouth matches the voice.
- Restore before you animate. A faded, scratched face animated in motion magnifies every flaw; a restored face looks strikingly alive.
- Short clips work best emotionally. A 20–60 second clip of a treasured recording usually lands better than a long one.
- The optional prompt field can steer the mood (for example, a gentle, warm expression). Keep it simple.
Use It Responsibly: Consent and Impersonation
A tool that makes any face say anything is powerful, and it comes with real responsibility. The line between a moving tribute and a harmful deepfake is consent and intent.
- Do not make a living person say anything without their clear permission — even family, even as a joke that gets screenshotted.
- Never impersonate a real person to deceive anyone — no fake statements, no fabricated endorsements, no “messages from beyond” presented as real.
- For a deceased relative, keep the project within the family and let close relatives decide together whether and where it is shared.
- Be ready to say what it is: an AI animation based on a real photo and a real voice. Honesty protects both the memory and the people who loved them.
Related Guides
Frequently Asked Questions
What do I need to make an old photo talk?+
Two things: a photo and a voice. The photo should be a clear, front-facing portrait (JPG, PNG or WEBP, up to 10MB). The voice can be an uploaded audio file (MP3, WAV, M4A or AAC, up to 10MB, between 5 seconds and 10 minutes) or something you record directly in the tool. The AI handles any language, because it syncs to the sound of speech rather than to specific words.
Can I use a voice recording of someone who has passed away?+
Yes — that is one of the most common and meaningful uses. Many families pair a preserved voicemail or home-video audio with a restored portrait. Because the person in the photo and the voice belong together, the result feels like a genuine keepsake rather than an imitation. Just keep the use private and respectful, and consider how close family members will feel about it.
Should I restore the photo before making it talk?+
Strongly recommended. Lip-sync animation relies on reading the mouth, eyes and face outline. Tears, stains, fading and blur all reduce that readability and increase artifacts in the video. AI Photo Restore fixes the damage first, and AI Photo Upscaler helps when the original scan is too small.
Is it ethical to make a photo talk with AI?+
It depends on who and why. The rule of thumb: use your own image, a family member's image with their blessing, or a deceased relative's image in a respectful remembrance context. Do not make living people say things they never said without their consent, do not impersonate anyone, and never use a talking photo to deceive. Handled honestly, it is a memorial medium, not a trick.
How long can the audio be?+
From 5 seconds up to 10 minutes, with a maximum file size of 10MB. Credits are charged based on audio duration — 5 seconds of audio costs 2 credits — and new accounts get 2 free signup credits, enough for one short test clip.
Does the talking video look like the real person?+
The face in the video is your uploaded photo brought into motion, so the likeness is real — but the performance is generated. Micro-expressions will not perfectly match how that person actually spoke. Treat the result as a loving recreation, not a recording of them.
Let Their Photo Do the Talking
One photo, one recording, one video. New accounts get 2 free signup credits — enough for one short clip.