What kind of photo should I use? Can I use a group photo?

One person only. Use an evenly lit front-facing portrait without occlusions or a strong profile angle — with several faces in frame the system cannot tell which one to drive.

Should I type the script or upload audio?

Typing is quicker: the platform voices it, and you can preview voices before committing. If tone and pacing matter, record it yourself and upload mp3 / wav / m4a. Billing differs: text mode includes voice credits, audio mode is priced by your audio length.

Do I lose credits if my balance is too low?

No. In text mode the voiceover is rendered and the quote is locked before the task is paused. After topping up, press Continue on the same page — no re-rendering, no double charge.

How long does it take? What if I close the page?

Usually 40 seconds to 2 minutes. The job runs on the server, so closing the page or switching devices is fine; the result shows up in your assets.

AI Avatar

Turn one portrait and a script into a talking avatar video

Result

Fill in the form on the left, then hit generate

Text mode quotes first — if you are short on credits you can still submit and continue after topping up

Watermark Enhancer Video HD AI MV Motion Control One-Click Video Ad One-Click Trending Replicate Short Play AI Avatar

What an AI avatar actually is

It is not a photo with an animated mouth. Your audio is split into phonemes, the mouth shapes follow them, and the micro-movement of head and shoulders is matched to the same timing — which is why it reads as the person genuinely saying those words.

Built for work that has to ship every week: product explainers, course narration, multi-language channels, internal training. The value is consistent output, not one viral clip.

Three steps to a talking avatar

Upload a portrait
A single front-facing photo works best — png, jpg or webp. Keep the face clear and evenly lit; avoid profile shots or group photos.
Write the script or upload audio
Type the text and the platform voices it (pick and preview a voice), or upload your own mp3 / wav / m4a. The audio length sets the video length.
Pick quality and generate
Choose 720p or 1080p, review the exact credit cost, then submit. Most videos finish in 40 seconds to 2 minutes, and closing the page does not interrupt the job.

How billing works

You pay per second of finished video, and voice synthesis is shown as its own line. In text mode the quote already includes the voiceover; in audio mode billing follows your audio length. If credits run short the task is paused with the voiceover rendered and the price locked — top up, press Continue, and it resumes without redoing the voice. Failed generations are refunded automatically.

Frequently asked questions

What kind of photo should I use? Can I use a group photo?: One person only. Use an evenly lit front-facing portrait without occlusions or a strong profile angle — with several faces in frame the system cannot tell which one to drive.
Should I type the script or upload audio?: Typing is quicker: the platform voices it, and you can preview voices before committing. If tone and pacing matter, record it yourself and upload mp3 / wav / m4a. Billing differs: text mode includes voice credits, audio mode is priced by your audio length.
Do I lose credits if my balance is too low?: No. In text mode the voiceover is rendered and the quote is locked before the task is paused. After topping up, press Continue on the same page — no re-rendering, no double charge.
How long does it take? What if I close the page?: Usually 40 seconds to 2 minutes. The job runs on the server, so closing the page or switching devices is fine; the result shows up in your assets.