Photo to Talking Video: One Portrait, One Clip

You can turn one photo and one voice track into a talking video with GoodAvatar. Upload a clear portrait and an MP3 or WAV track; the portrait speaks the audio in a single generated clip.

This is useful when you have a good still image but no recording of the person speaking.

What makes a good photo

• Front-facing or slightly angled
• Even lighting; no strong shadow across the eyes
• Face large enough to read expression
• No heavy beauty filters or extreme wide-angle distortion
• JPG or PNG

Generate the portrait at the size you need rather than photographing one: the studio renders a clean, front-facing subject, and you can keep the face large in the frame without cropping afterwards. Only AI-generated images are accepted; files that carry camera metadata are rejected before generation.

What makes a good audio track

• Clean voice, no loud background music
• MP3 or WAV
• 15-30 seconds for the first test
• Within the current 120-second single-track limit

The complete track goes into one generation pass. You do not need to split it into sentences or merge video segments afterwards.

Crop and aspect ratio

The photo's shape should match the place you will publish:

• 9:16 for Shorts, Reels, TikTok, and vertical landing-page embeds
• 1:1 for feed posts and square cards
• 16:9 for a landing page, a course player, or a presentation

Generate the portrait in the same shape as the place you will publish. A vertical portrait dropped into a 16:9 frame will either be letterboxed or cropped by the player; neither looks intentional. Leave a little space above the head and keep the eyes in the upper third of the frame.

The create flow can also choose an aspect ratio for the output. If you upload a portrait that is already close to the target ratio, the result is less likely to feel tightly cropped.

A 20-second script for a photo

A 20-second clip needs about 40-55 words. Use three beats:

1. Who and why (5s): name the speaker and the reason for the clip.
2. One point (10s): the single thing the viewer should remember.
3. Next step (5s): where to click, read, or reply.

Example: "Hi, I'm Dana from the onboarding team. Your first week is about one thing: getting your workspace set up. The checklist below takes about ten minutes, and it unlocks the rest of the course. Start with step one."

Read it aloud. If you stumble, cut a clause. The model can reproduce a clean read; it cannot rescue a sentence that was never meant to be spoken.

What the output can and cannot do

The result is a head-and-shoulders performance. The model animates the face and mouth to the audio; it does not turn a still photo into a full-body actor, add hand gestures, or invent a new scene.

That is enough for a presenter clip, a course intro, or a sales follow-up. It is not enough for a demonstration that depends on hands, props, or a changing environment. If the viewer needs to see a product in use, record the screen or the product separately and use the avatar clip as the intro or recap.

Before vs. after

Portrait policy

GoodAvatar accepts AI-generated portraits only. Every upload requires you to confirm the portrait was produced by an image model, and files that carry camera metadata are rejected outright. Real photographs — including your own, a client's, or a public figure's — are not accepted.

The terms of service prohibit impersonation, non-consensual imagery and deceptive use, and a maintained public-figure blocklist rejects prompts that name real politicians or celebrities. If a generated portrait turns out to depict a real person, report it to support@goodavatar.dev and we will remove the content and the account.

Step by step

1. Upload the portrait.
2. Upload the audio, or generate a voiceover with text to speech.
3. Review the estimated duration and credit reservation.
4. Submit the task and follow its progress.
5. Preview and download the finished MP4.

A 30-second clip usually takes about 10-15 minutes because generation runs at roughly 20-25 times the video length.

Common failure cases

• The face is too small. Crop closer before uploading; a full-body photo usually gives the model too little face detail.
• The photo is a strong side profile. A front-facing or slight angle is easier to animate than a near-profile.
• The eyes are hidden. Sunglasses, hair across the face, or a heavy shadow make the result less convincing.
• The audio is noisy. Background music and clipping compete with the voice; use a clean track.
• The script is too long for the clip. Cut the words before generating, not after.

If the first result is off, change one input at a time. A different crop often fixes framing; a cleaner audio track often fixes sync.

Where this format fits

• A founder or instructor intro from a headshot
• A course module summary
• A sales follow-up with a personal opening line
• A help-center answer with a consistent presenter
• A product launch note where the product itself is shown separately

It does not replace a screen recording, a product demo, or a full lecture. It replaces the talking-head layer of those videos.

Privacy and consent

Upload only images and audio you have the right to use. If the photo shows another person, get their permission before generating a video with their likeness. Do not use the tool to impersonate a public figure, create sexual content, or imply an endorsement that does not exist.

You can delete individual tasks and their files from the service. For data requests, contact support@goodavatar.dev.

Cost

GoodAvatar uses 5 credits per second. A 30-second video costs 150 credits, about $1.69 at the Starter pack rate. New accounts get 100 credits for a first test. Credits are valid for 180 days, and failed generations return the reserved credits automatically with no exceptions.

For the full pack comparison, see pricing. For the broader workflow, see the AI avatar video generator page. For the format itself, read what is a talking head video. For usage rights, see commercial use.

Try a short clip first

Start with 15-30 seconds and a clean portrait. Create your first video with the 100 starter credits, then decide whether a pack fits your production volume.