> ## Documentation Index
> Fetch the complete documentation index at: https://docs.percify.io/llms.txt
> Use this file to discover all available pages before exploring further.
# Make a two-person AI podcast video from a script
> Percify Podcast Studio turns one photo of two people and a written dialogue into a lip-synced podcast video, up to 3 minutes long.
Percify Podcast Studio makes a podcast video from one photo of two people and a script you write line by line. Each speaker gets a voice, both people in the photo are animated and lip-synced to their lines, and the finished video lands in your library. Podcasts run from 15 seconds to 3 minutes.
## What you need
* A computer. Podcast Studio runs in the desktop web app at [app.percify.io/home/podcaststudio](https://app.percify.io/home/podcaststudio). You can also open it from **Podcast Studio** under Creative tools on the Explore page.
* One photo with both people in it, side by side and facing the camera. Images can be up to 10 MB.
* A script where both speakers have at least one line.
* Credits for the length of the finished audio (see below).
## How to make a podcast
Click **Upload** next to "Add one photo with both people in it", or drop the photo on the stage. Speaker one is the person on the left of the frame and Speaker two is on the right.
Under each speaker, open **Voice** and choose a voice. There are 10 preset voices: Puck, Kore, Charon, Aoede, Fenrir, Leda, Orus, Zephyr, Gacrux and Sulafat.
Under **The conversation**, write each line in its own box. New lines alternate between the speakers, and the speaker chip on a line switches who says it. Click **Add line** for more, or **Start from an example** to load a starter script.
In **Delivery**, describe how it should be read, for example "relaxed, like two friends catching up".
Check the estimate ("about N for roughly Ns") and the expected wait, then click **Generate podcast**. You can close the page. You get a "Your podcast is ready" notification, and the video appears in the History tab and in your library.
## How much does it cost?
Podcast Studio charges 150 credits per minute of finished audio, counted by the second. The price on the page is an estimate from your script; the charge is worked out from the voices once they are generated, before the video renders.
| Length | Credits |
| -------------------- | ------- |
| 15 seconds (minimum) | 38 |
| 30 seconds | 75 |
| 1 minute | 150 |
| 3 minutes (maximum) | 450 |
The charged length is the generated audio plus one second, rounded up to a whole second. If the render cannot start or fails later, the credits are refunded and you get a "Your podcast failed" notification saying how many were returned. See [how credits work](/percify/credits).
## Limits
| Limit | Value |
| ------------------- | ---------------------------------------------------------------- |
| Length | 15 seconds to 3 minutes of audio |
| Lines | Up to 40 |
| Characters per line | Up to 1,200 |
| Delivery note | Up to 300 characters |
| Speakers | Exactly two, from one photo |
| Podcasts at a time | One. Wait for the current one to finish before starting another. |
| Output resolution | 480p |
## How long does it take?
The page shows an estimate next to the button, based on the length. Rendering takes roughly 10 to 30 seconds per second of audio, so a 1 minute podcast is about 10 to 30 minutes and a 3 minute podcast can take up to about 90 minutes.
## Tips
* Use a photo where both faces are clear, side by side and facing the camera.
* Short, natural lines read better than long paragraphs. Give each speaker a real reply.
* Start with a 20 to 30 second script to check the voices and the photo before paying for 3 minutes.
* If the button is greyed out, the reason is shown next to it: "Add a photo of both speakers", "Both speakers need a line" or "Your last podcast is still rendering".
Not in Podcast Studio. It uses the 10 preset voices listed above. To make a video in your own cloned voice, use [Clone Yourself](/create/clone-yourself).
Podcasts are capped at 3 minutes of audio. A script that runs longer is refused with a message saying the cap and how long your dialogue runs, before any credits are taken. Shorten or remove lines and generate again.
No. Both people have to be in the same photo, the left person as Speaker one and the right person as Speaker two.
## Related
One person, one voice, lip-synced to your audio.
Record your voice once and reuse it.
Plans, monthly credits and refunds.
What to do when a render does not finish.