# AI avatar > How to make the person in a photo say your text in the chosen voice and add the video to the reel with captions: photo, text, voice, price and wait time. Source: https://docs.slicer.media/en/guide/ai/avatar/ An AI avatar brings a photo to life: the person in it says your text in the chosen voice. The finished video goes into the reel as a new clip — with captions. An avatar comes in handy to add an intro or a call to action at the end, voice an idea you didn't film, or present a product on behalf of a host. **Demo (Web):** 1. Put the playhead at the end of the second clip 2. With nothing selected, press “AI avatar” — the sheet opens 3. Pick a photo and type “Today I'll show three tricks that will speed up your editing” 4. Pick the “Confident, male” voice — the button shows the price and length “≈5 s” ## Creating an avatar ### 1. Place the playhead The avatar becomes a new clip **after the clip at the playhead** — the place is fixed when you start. ### 2. Open “AI avatar” Clear the selection and press **AI avatar** in the main toolbar. ### 3. Pick a photo The **Photo** section shows the project's photos. Click the one you need, or “+” to upload a new one. **Web** “+” opens a file picker. A photo is required: if you pick a video, you'll see **A photo is needed — a video will not do**. **iPhone and iPad** “+” opens your photo library: pick a photo from the library. ### 4. Write the text In **What to say** — what the avatar will say: 3 to 500 characters, with a counter under the field. 500 characters is about 35 seconds of speech. ### 5. Pick a voice **Warm, female**, **Soft, female**, **Confident, male** or **Deep, male**. ### 6. Create The button shows the price and the approximate speech length: “Create · ≈ N tokens (≈S s)”. Press it. The hint under the button: **It takes 2–5 minutes — keep editing meanwhile. Tokens are charged only if the avatar works out.** While the avatar is being made, you see **AI avatar is being created…** above the preview. When it's ready, the message **AI avatar added to the video** appears, and a new clip with captions from its words shows on the timeline. ## Price An AI avatar is charged **at cost** — exactly what the AI services charge for the avatar video and the speech synthesis, with no markup. You pay for each started second of speech, and no less than for 2 seconds. The length is estimated from the text — about 14 characters per second — and shown on the button together with the price before you start. The sheet explains: **Costs exactly what the AI service charges, no markup. The actual price is charged — never more than shown.** While the avatar is being made, the price is reserved on your balance, rounded up to a whole token — so you need at least one free token. The actual price is charged, never more than shown, and only if the avatar works out. If not, you see **The AI avatar did not work out — tokens returned**. How at-cost pricing works — [Tokens, packs and the free reel](https://docs.slicer.media/en/guide/pricing/#at-cost). ## What you get - The avatar video becomes a new project source “ИИ-аватар.mp4” — you can see it in the project's **Media** section. - The avatar clip is a regular clip with speech: you can trim it by words, reorder it, add a transition, effect, color, mask or [remove the background](https://docs.slicer.media/en/guide/ai/remove-background/). - Captions are built from the avatar's recognized speech, in the reel's style. - The photo is prepared automatically: a very elongated one is cropped to proportions no narrower than 1 : 2.5. > **Important.** Stay in the editor while the avatar is being made: that way it goes into the reel by itself. If you leave, the finished video still appears in the project sources — add it with “+” → **Source clip** and drag the clip's handle out to its full length ([Clips](https://docs.slicer.media/en/guide/editor/fragments/)). ## Tips - A portrait where the face is clearly visible and looks at the camera works best. - Write the text the way people speak: short sentences, no lists or abbreviations. - One avatar is made per project at a time — start the next one when the previous is ready. - Platforms ask you to label videos with an AI avatar: turn on **Made with AI** when publishing ([Publishing](https://docs.slicer.media/en/guide/publish/platforms/)). - Use photos of people who agree to have their image speak for them — for example, your own. ## What's next - [AI voiceover](https://docs.slicer.media/en/guide/ai/voiceover/) — narration for the whole reel without a face on screen. - [Clips: trim and reorder](https://docs.slicer.media/en/guide/editor/fragments/) - [Operation prices](https://docs.slicer.media/en/guide/help/prices/)