AI Avatar Video Generator: Turn a Script Into a Talking Video
Pick a presenter from the catalogue, pick a voice to go with them, and paste in the script you want spoken. A few minutes later a video comes back of that presenter saying your words.
1,000+
presenters to choose from
1,000+
voices and accents
Up to 1 min
of speech per video
3 a day
free units, a long script uses two
~3 minutes
per generation
Free with an account
sign in to generate
What an AI Avatar Video Generator Actually Does
It turns a written script into a video of a presenter speaking it out loud. There is no camera, no recording booth and no editing timeline: the presenter, the voice and the words are three separate choices, and the generator renders a video that puts all three together.
You pick a person from a catalogue of stock presenters, filtered by look and by setting, rather than filming or animating anyone yourself.
A separate catalogue of voices and accents supplies the sound. The presenter you see and the voice you hear are chosen independently, so you can match a look to a tone that fits it.
Paste the words you want spoken, up to about a minute's worth, and the generator lines the presenter's delivery up to the voice reading that exact script.
Choosing a Presenter and a Voice
Both catalogues are large enough that the harder part is narrowing them down, not finding something usable, and the tools for narrowing them are built into the picker itself.
Filter presenters by look and by setting before you commit to one, instead of scrolling a flat list and judging thumbnails one at a time.
Preview a presenter and preview a voice separately before you generate anything, so the combination you land on is one you already heard and saw, not a guess.
The finished video speaks whatever language the script itself is written in, so pick a voice and accent that actually fits that language rather than assuming a default will cover it.
Voice speed and volume are both adjustable, which is often the difference between a delivery that sounds rushed or flat and one that sounds like it was recorded on purpose.
How to Make an Avatar Video
Four steps between a blank script and a finished video.
An account is required to generate. Signing in also gives you a daily allowance of free avatar videos and a library of everything you have made.
From the Studio menu, open Avatars. This is a dedicated page rather than a chat, because picking a presenter and a voice is closer to filling in a form than holding a conversation.
Filter the presenter catalogue by look and setting, filter the voice catalogue by accent, gender and age, and listen to or watch each before you settle on it.
Write or paste the script, up to 900 characters, about a minute of speech, then send it. The video is ready in about three minutes.
What You Control in an Avatar Video
A handful of settings decide how the finished video looks and sounds.
The script is the only script of the video: whatever you paste is exactly what the presenter says, read by the voice you picked.
A shorter, cleaner script tends to read more naturally than one packed with abbreviations or numbers a voice has to guess how to say.
Speed can be nudged from 0.7x up to 1.2x, so a script written for calm narration does not have to sound rushed, and a fast-paced one does not have to drag.
Volume is adjustable on its own, separate from speed, for a delivery that sits at the level you actually want in the finished video.
Turn the background transparent to layer the presenter over your own footage, slides or graphics instead of whatever setting came with the catalogue shot.
This is the setting that turns a stock presenter clip into something you can drop into a larger edit rather than use as the whole video.
Every finished avatar video lands in your Generations library, so you can come back to it later without hunting through a chat history.
Sending the same presenter, voice, script and settings again returns the video you already made instead of rendering a duplicate.
What People Use Avatar Videos For
A presenter and a voice cover a lot of ground once the script is the only thing you have to write.
Product explainers and ads: a presenter walking through what something does reads as more direct than text on a slide, and you can try several scripts before picking the one that lands.
Course and onboarding clips: a consistent presenter across a series of short lessons gives the material a face without booking a person for every update.
Social posts in another language: write the script in the language you want spoken and pick a voice and accent that match it, rather than dubbing a video after the fact.
Testing a script before a real shoot: running a few versions through an avatar video is a cheap way to hear how a script sounds out loud before committing a real presenter's time to it.
Questions About the AI Avatar Video Generator
What people ask before their first avatar video.