Quick Answer
If you already have the audio, you do not need voice cloning: you need lip sync. Give Percify's InfiniteTalk Fast one photo and the audio file and it renders a person speaking it at 2 credits a second, so a 60 second clip is 120 credits, about $2.53 on the Creator plan, and a 10 minute episode is 1,200 credits, about $25.30. That is why most podcasts publish clips rather than whole episodes as talking video. YouTube requires you to disclose AI that makes a real person appear to say something they did not, using the 'AI use' setting in YouTube Studio.
Turn an audio file into a talking video: the steps, what each length costs at 2 credits a second, and YouTube's own rule on disclosing AI. Checked 16 September 2026.
Related next steps
The premise of this job is that the audio already exists: a podcast episode, a recorded talk, a voice note. What is missing is a face saying it. That is lip sync, not voice cloning, and skipping that distinction is what makes most guides on this topic cost you money for nothing.
Percify numbers below were read on 15 September 2026; YouTube's rules were read on its own help pages on 16 September 2026.
The steps, with what each one costs
- Pick the face. One photo where the face fills at least a third of the frame. Creating a Percify account is free; making videos needs a plan, from $12 a month.
- Upload the audio. InfiniteTalk Fast ↗ takes an image and an audio file and renders the video. You do not record a script and you do not clone a voice: the voice in the file is the voice in the video.
- Pay by the second. 2 credits a second, so 30 seconds is 60 credits and a minute is 120, about $2.53 on the Creator plan ($26 for 1,233 credits).
- Add captions if it is going in a feed. Word timed captions burn in at 1 credit a second.
- Disclose it on upload, using the setting described below.
What length actually costs
| Audio length | Credits at 2 a second | About, on Creator |
|---|---|---|
| 30 seconds | 60 | $1.27 |
| 1 minute | 120 | $2.53 |
| 5 minutes | 600 | $12.65 |
| 10 minutes | 1200 | $25.30 |
Read that table before planning a series. Rendering a whole episode as a talking head costs real money and is rarely what people watch anyway: the common shape is three or four clips of 30 to 60 seconds cut from the episode, each $1.27 to $2.53, with the full episode staying audio. Percify's plans and packs are on pricing.
YouTube's rule, in YouTube's words
YouTube requires creators to disclose ↗ when they use AI "to meaningfully alter or generate photorealistic content". It names three cases: content that "makes a real person appear to say or do something they didn't do", content that "alters footage of a real event or place", and content that "generates a realistic scene that didn't actually occur".
A photo of a person lip synced to audio is the first case, so it needs disclosing. You do it with the 'AI use' setting in YouTube Studio; after you select it and upload, YouTube says "the content will be labeled as AI generated or altered for viewers".
YouTube also says what does not need disclosure: content that is not realistic, and edits that are "primarily aesthetic" and do not mislead the viewer about what happened.
If it is going out as a Short
YouTube's help pages say Shorts creation tools make videos "up to 3 minutes long", so a 3 minute clip is the ceiling for that format. At 2 credits a second, 3 minutes is 360 credits, about $7.59, which is another argument for cutting to the strongest 45 seconds instead.
When a different tool fits better
- You filmed yourself and want it tidied: an editor, not an avatar tool. Descript's Hobbyist plan was $16 a month billed yearly, or $24 monthly.
- The presenter should not be a real person you know: Synthesia's stock avatars start on its Starter plan, $29 a month or $18 billed yearly.
- You have no audio yet: then you do need a voice, and cloning your voice free compares the options, including ones you can run yourself.
For the wider map, see which tool for which job, five jobs and what each costs and the best AI avatar generators compared.
Ready to Create Your Own AI Avatar?
Make a talking video of yourself from one photo, in your own voice. No watermark on any plan.
Cancel anytime from your profile
Got questions?
Frequently asked
Give a lip sync model one photo and the audio file. On Percify, InfiniteTalk Fast renders the person speaking that audio at 2 credits a second, so a 60 second clip is 120 credits, about $2.53 on the Creator plan. You do not need to clone a voice, because the voice is already in your file.
No. Voice cloning makes a voice from scratch for scripts you have not recorded. When the recording exists, the job is lip sync, and cloning would only add cost.
Yes, for realistic content. YouTube requires creators to disclose AI used "to meaningfully alter or generate photorealistic content", including anything that makes a real person appear to say or do something they did not do. Use the 'AI use' setting in YouTube Studio, and YouTube labels it for viewers. Non realistic content and minor aesthetic edits do not need disclosing. Read on 16 September 2026.
At 2 credits a second, 10 minutes is 1,200 credits, about $25.30 on the Creator plan ($26 for 1,233 credits). Most people render 30 to 60 second clips instead, at $1.27 to $2.53 each, and leave the full episode as audio.
YouTube's help pages say its Shorts creation tools make videos up to 3 minutes long, read on 16 September 2026. At 2 credits a second, a full 3 minutes of talking video is 360 credits, about $7.59.
