AI Podcast Generator: Turn a Topic Into a Two-Host Episode
How an AI podcast generator works: the script is written first and free to edit, then each line is voiced, so you only pay for the audio you keep. Lengths, voices and costs.
An AI podcast generator writes a conversation between hosts from a topic and voices it with text-to-speech. The part that decides whether it’s any good is the script, because a monologue read by two voices doesn’t sound like a conversation. ViralMint’s Podcast tab in Voice Studio writes the script first and lets you read it before you pay for any audio, which keeps the expensive step for a script you already like.
This guide covers how it works, what the lengths and voices are, and what an episode costs.
How an AI podcast generator works
There are two separate jobs. The first is writing: a model drafts turns between hosts with interruptions, short reactions and a beginning and end, to a set length. The second is voicing: each line goes to a voice model and the results are joined with pauses. Tools differ in how much they let you see between those two steps. In many, you press one button and get a finished file, so a wrong fact or a flat opening costs you the whole episode.
In Voice Studio the two steps are separate on purpose. The script is written free, and you edit it as ordinary text before you spend anything on audio.
How to generate a podcast
- Give it a topic. Type the subject and any direction, or paste source text the hosts should work from. The writer is told not to invent facts that aren’t in what you gave it, and you should still read the result, because a model can be confidently wrong.
- Pick a length. Choose 2, 5 or 10 minutes. The length becomes a word budget for the script.
- Read and edit the script. It arrives as lines by speaker. Fix anything, cut a tangent, or rewrite the opening.
- Choose the voices. Pick one for each host, then generate the audio.
- Download it. The finished episode lands in your Library as one audio file, ready to publish wherever you host your show.
A failure partway through doesn’t waste the money you spent: finished lines are kept, and retrying buys only the lines that didn’t finish. On the desktop, a written script is also remembered per brief until the episode is delivered, so a retry doesn’t rewrite it into something new.
Which voices to pick
| Voices | Languages | Price |
|---|---|---|
| Gemini 3.1 Flash TTS, 30 voices | Many languages | $0.12 per 1,000 characters |
| Premium (MiniMax Speech 2.8 HD), 12 voices | English only | $0.20 per 1,000 characters |
| Your own cloned voice (desktop) | Local, on your computer | 2.5¢ per 1,000 characters, capped at 10¢ per generation |
Premium voices suit longer narration where the listener notices the difference. Their English-only limit is a hard rule, because they bill far more on other scripts; a mostly non-English script is refused before any charge. For a show in another language, use the Gemini voices.
What an episode costs
Speech runs about 13 to 14 characters per second, so a 5-minute episode is around 4,000 characters:
- Gemini voices: about 50¢
- Premium voices: about 80¢
- A 10-minute episode: about double those figures
The script itself costs nothing, and the price is shown before you confirm. There’s no subscription; audio bills from your prepaid balance.
When to use Dialogue instead
The Podcast tab writes a script from a topic. If you already have a script, or you want a skit, an interview or a two-voice ad, use the Dialogue tab: paste the lines by speaker and assign a voice to each one. The desktop app can also cast up to six speakers on one episode and render an audiogram video, which the web app doesn’t do.
A note on what to publish
An AI-written conversation sounds confident whether or not it’s right, and listeners assume a podcast was fact-checked. Read the script before you voice it, and say that your show uses AI voices where your platform or your audience expects to know.
Related posts
- AI Voice Over for Videos
- Best Free AI Voice Generators
- Podcast to Shorts: Clip an Episode Into Vertical Videos
- Audio to Video: Turn a Voice Recording, Podcast or Song Into a Video
Ready to try it? Open the web app or download ViralMint for macOS, Windows or Linux.
Frequently asked questions
What is an AI podcast generator?
An AI podcast generator writes a conversation between hosts from a topic and voices it with text-to-speech. ViralMint's Podcast tab in Voice Studio writes a two-host script for free, lets you edit it, and only charges when you voice it.
How much does an AI podcast cost to make?
The script is free. Voicing bills by character: Gemini voices at $0.12 per 1,000 characters and Premium English voices at $0.20 per 1,000. A 5-minute episode runs about 4,000 characters, so roughly 50¢ with Gemini voices or 80¢ with Premium voices.
How long can the episode be?
The presets are 2, 5 and 10 minutes. The script is written to a word budget for the length you pick, so the audio lands close to it.
Can I have more than two hosts?
In the desktop app you can cast up to six speakers. The web app writes two-host episodes.
Can I use my own script?
Yes. Dialogue, another tab in Voice Studio, voices a script you already have line by line, with a different voice per speaker.
Does it make a video of the podcast?
The web app produces audio only. The desktop app can also render a 1920×1080 audiogram video of the episode.
What happens if a generation fails halfway?
Lines already voiced are kept. A retry buys only the lines that didn't finish, so you don't pay twice for the same audio.