How to make a talking head video without filming
Recording a piece to camera means finding a quiet room, decent light, a microphone, and the nerve to do fourteen takes. If what you actually need is a face saying some words, you can skip all of it: upload one photo, paste the script, and the voice is synthesised and lip-matched for you.
Open the AI ActorStep by step
- 1
Pick a portrait that will hold up
One person, facing the camera, whole face visible and in focus. Sunglasses, a hand near the chin, a second face in the background, or a heavy side angle all give the model less to work with and it shows in the mouth.
- 2
Write the script the way it should be spoken
The text is read literally, so write contractions, keep sentences short, and put a comma where you want a breath. Stage directions or bracketed notes get read aloud — leave them out.
- 3
Check the estimate before you generate
The workspace shows the projected length and credit cost as you type. Speech runs at roughly six characters a second, so a 300-character script is about 50 seconds. Trim the script if the number is higher than you want to spend.
- 4
Generate, then come back
Long scripts are split into segments, generated separately and joined, so this takes minutes rather than seconds. You can queue other work or close the tab — the job keeps running and lands in your task list.
What makes the difference
- Start with two or three sentences before committing to a long script. If the face does something odd, you would rather find out after 60 credits than after 1,200.
- The video is exactly as long as the script makes it. There is no way to generate 'a 30-second video' directly — you control the length by controlling the word count.
- Every result lands in your asset library, so a portrait you are happy with can be reused for the next script without uploading it again.
- Use your own likeness, or one you have permission to use. A face is a person, and this tool makes it say whatever you type.
Questions
Do I need a microphone?
No. The voice is synthesised from the script. If you would rather use your own recorded audio, use the Lip Sync tool instead — it takes an audio file and matches the mouth to it.
How much does a one-minute video cost?
AI Actor is 20 credits per second, so roughly 1,200 credits for a minute. The workspace estimates the cost from your script before you generate, and nothing is charged if the job fails.
Can I choose the voice?
Yes. Leave the voice field empty for the default, or paste a voice ID to use a specific one. Speech speed is adjustable too.
Why does it take several minutes?
The script is broken into segments that are generated one by one and then joined, with the audio synthesised once up front so the voice does not drift between segments. That is slower than a single short clip, but it is what keeps a long piece sounding like one person.
Can I put someone else's face on my video?
Not with this tool. AI Actor animates the portrait you upload; it does not transplant a face onto existing footage. Use a photo of yourself or someone who has agreed to it.