Synthesia AI Avatars: How to Create Stock, Photo and Studio Avatars

Synthesia gives you four ways to put a presenter on screen: a stock avatar, a character you design in Avatar Builder, a personal avatar of yourself, and a filmed Studio avatar. For your own face, the practical route is a personal avatar from a photo: one picture plus a short live consent video, typically ready within minutes, on paid plans only. Starter includes 3 personal avatars and Pro 5. Stock avatars come with every plan, from 9 on Basic to 240+ on Enterprise. A Studio avatar is a $1,000-a-year add-on for annual plans only and can take up to 10 days to process.^[1] Pick the type first. It decides gestures, backgrounds and API use.

Choose the avatar type before you write a script

Stock avatars cost nothing extra on any plan, but the library grows with the plan: 9 on Basic, 125+ on Starter, 180+ on Pro and 240+ on Enterprise. Avatars outside your plan show an upgrade icon, so a face you like while browsing may not be yours to use. Synthesia says its stock actors consented and are paid per video made with their likeness. The library mixes several technology generations, and the generation matters more than the face, because it sets what the avatar can actually do on screen.

Two practical consequences follow from that grid. If you plan to generate videos through the Video API, Express-1-based avatars work there (Express-1 stock, legacy stock and Studio avatars); Express-2 avatars do not. If you want a different background, Express-2 stock avatars keep a fixed one that cannot be removed, which is why the customizable avatars below matter. Express-3 is still rolling out. Confirm it in your workspace first.

Customize a stock avatar or build a new one

Customizable avatars let you change outfit, space and pose by prompt, on every plan. The feature works only on Express-2 customizable stock avatars and on personal avatars made from a photo.^[2] A generation takes around 2 to 4 minutes and returns several variations; poses are standing or seated, placed left, middle or right. You can also remove the background or swap it for a solid color or gradient. Brand-kit colors and a logo on the outfit or set are Enterprise only, and so is uploading a reference image.

Action b-roll goes one step further. From a preset or a text prompt, a customizable or photo-based avatar performs an action in a short clip that you drop between talking scenes. Each such clip costs 230 credits, so budget it against the month's balance. Credits reset every billing period and do not roll over, as our Synthesia pricing guide explains.

Avatar Builder is the route when no stock face fits the brief. You describe a concept, or use the panel controls, and it generates realistic or stylized options, including non-human characters. Photo uploads are not accepted here; that is what personal avatars are for. Each generated option uses 14 credits, and the default batch of two uses 28. Builder runs on every plan but needs a desktop browser, like the video editor itself. Its avatars also work in Syren, the AI video agent launched in October 2026 and still labeled beta.

Create a personal avatar from your photo

A personal avatar is your digital twin, and it needs a paid plan: Starter includes 3, Pro 5, and Enterprise sets no fixed cap, subject to reasonable consumption and compute. Starter is $19 a month billed monthly, or $14 a month billed yearly ($168 a year). Synthesia recommends the photo route over the older video route, and it runs on Express-2 technology, so outfits, backgrounds, poses and action b-roll all stay available.^[3] The steps are short:

1. Upload one photo of yourself; this works from a mobile browser too, though there is no app.
2. Record a short live consent video in which you read aloud the passcode shown on screen.
3. Add a voice sample if you want a clone; otherwise the avatar speaks with a Synthesia voice.
4. Wait for the identity and consent checks; the avatar is typically ready within minutes.
5. Customize outfit, background, pose and action b-roll as described above.

Synthesia creates a personal avatar or voice clone only with a live consent recording from that same person, who must be of legal age; it refuses celebrities, public figures, deceased people and anyone who has not consented, and you cannot create a photo avatar of someone else even with their permission. Creating an avatar processes biometric data such as facial geometry and voiceprint. Synthesia says it will not sell that data, and owners can request deletion at any time.

Set expectations on realism before you show it to colleagues. Photo avatars use a speech-driven animation model, so body language stays neutral and motion does not follow how you really move. Synthesia itself warns the result may not perfectly match the person. A new version means recording consent again. The legacy video route needs one continuous take of 1 to 5 minutes (.mp4, .mov or .webm up to 2 GB), takes about one business day, clones your voice from the same footage, and cannot be customized afterward. Deleting a personal avatar frees its slot.

Clone the voice that goes with it

Voice cloning is on every plan, Basic included. You either record yourself or upload 1 to 5 minutes of audio, then read a randomly generated passcode aloud as consent.^[4] Silent recordings are rejected. Keep the consent recording under 60 seconds. A clone is typically ready in 5 to 10 minutes and works on any avatar your plan includes, not only your own.

Language coverage is the limit to check. A clone speaks 32 languages on Basic, Starter and Pro, and Enterprise unlocks about 47 more for both creating and using clones. Stock voices for avatar videos cover 140+ languages and accents on every plan, so multilingual projects can pair your clone with stock voices where your language is missing. Synthesia builds clones on ElevenLabs instant voice cloning, and it notes a clone may not keep a non-native English accent. Importing an ElevenLabs Professional Voice Clone is Enterprise only.

Decide whether a Studio avatar is worth $1,000 a year

A Studio avatar is a filmed Express-1 avatar sold as a paid add-on at $1,000 a year, for annual-plan users only. The yearly pricing view lists this add-on on the Starter, Pro and Enterprise columns. Processing can take up to 10 days. The footage is demanding: three takes plus a consent recording, MP4 in H.264, 4K preferred and 1080p at minimum, shot on green screen, with a performer aged 18 or older. You can film it yourself or in Synthesia's partner studios in London and New York.

The result has no background, follows the script with facial expressions, and has no hand gestures. Its appearance is fixed after filming. So why pay? API compatibility is the clearest reason. Studio is the only custom avatar type that works with the Video API, which runs on Pro (up to 360 minutes a year, deducted from plan limits) and Enterprise. If you generate videos only in the editor, a photo avatar gives you outfits, backgrounds and poses on a plan you already pay for.

Check the rules before your avatar goes public

Every video passes automated moderation plus human review at generation. Stock avatars are limited to neutral, factual, educational content, so scripts on medical, legal, financial, news or political topics, or naming real people and companies, can be ruled ineligible for a stock avatar. Several of those topics need a personal or Studio avatar, and some need Enterprise. Misinformation, guaranteed returns, gambling and prescription-drug promotion are banned on every avatar. Rejections can be appealed once, on paid plans only, with a decision within 48 hours.

Advertising is the other trap. Stock avatars may not appear in paid TV, paid social, programmatic or any paid promotion without Synthesia's written consent, and they cannot endorse products or act as spokespeople. Organic posts and training videos are fine. Personal and Studio avatars are not bound by that paid-promotion rule; their use follows your agreement with the person depicted. Watch the output side too. Basic videos carry a watermark and are shared by link or embed, with no MP4 download; that starts on Starter.

Avatars can also hold live conversations. Interactive Avatars have been on all plans since October 2026, still labeled beta, and every plan gets a one-time 60 minutes at sign-up. Built in Studio with a knowledge base and deployed as a website widget, they use 20 credits a minute. An interactive clone of yourself also uses 20 credits a minute, with conversations capped at 2 minutes. For Interactive Avatars on your website or app, stock actor avatars such as Ryan and Ada cannot be used; the avatar must be synthetic or personal (Roleplay Sessions have their own stock-avatar list). If you are comparing digital twins across platforms, our HeyGen review covers its footage-based approach.

The bottom line

Start with a stock avatar to test the script, then move to a personal avatar from a photo once the series is worth your own face: it is the fastest custom route, it keeps outfits and backgrounds, and it fits inside any paid plan. Clone your voice early, because the language list sets your multilingual plan. Save the Studio avatar for teams that need the Video API, and read the stock-avatar topic and advertising rules before the first script, not after a rejection.