AI Podcast Generator

An AI podcast generator that lets you edit before it speaks

One-click audio overviews made AI podcasts famous: paste a document, wait, and two cheerful hosts discuss it. They are impressive the first time and frustrating the tenth. You cannot change what the hosts say, you cannot choose their voices, and if one sentence is wrong the only option is to generate the whole thing again. This AI podcast generator splits the process into two steps so you stay in control: first the script, then the voices.

The script step reads your source — a topic in a sentence, a few paragraphs of notes, or the text of a public web page — and writes an episode shaped for listening: a hook, a short welcome where the hosts introduce themselves, three to five segments with concrete points, and a sign-off. It is written for the ear, with short sentences and contractions, and it is told not to invent precise statistics or quotes. Then it stops and hands the script to you. Rewrite the jokes, cut a segment, move a point to the other host, or add a line about your own product. Writing the script costs one credit.

Voices you choose, recorded line by line

When the script reads right, pick a voice for each host from the full library of 118 voices — contrasting voices, such as a low storyteller and a bright upbeat voice, make it easy for listeners to tell who is speaking. Every voice is level-matched, so the episode doesn't jump in volume when the hosts trade lines. Recording uses the same engine as the AI voice generator: each line is its own take, so once the episode exists you can rewrite a single line and the AI podcast generator records just that one again.

Length, languages and cost

Choose 2, 5 or 8 minutes. The AI podcast generator aims for about 150 spoken words per minute, so a five-minute episode is roughly 4,500 characters and costs about 12 credits to voice, plus one for the script. Fourteen languages are available for the script and the voices, including Spanish, Portuguese, French, German, Japanese, Korean, Mandarin, Hindi and Arabic, so the same source material can become a separate episode for each audience.

What people make with it

Newsletter writers turn each issue into a short audio edition for subscribers who would rather listen. Teachers make two-voice recaps of a lesson that feel like a conversation instead of a lecture. Companies turn release notes and internal updates into a weekly five-minute show. Language learners generate dialogues on topics they care about, at a pace they can follow, with a transcript to read along. Each episode comes with SRT and VTT files, which double as a ready-made transcript for show notes and accessibility.

Add a theme tune with the AI music generator, and if you already have a recorded episode that needs a transcript, use audio to text.

Tips for a better episode from the AI podcast generator

• Give the AI podcast generator real material. Notes with specific facts produce better episodes than a one-word topic.
• For an article link, use a public page; paywalled or script-heavy pages may not be readable — paste the text instead.
• Read the script out loud once. Cut any line you wouldn't say to a friend.
• Pick voices with clearly different pitch so listeners can follow the conversation without names.
• Keep episodes short at first; two to five minutes is the sweet spot for newsletters and updates.
• Disclose that the hosts are AI voices in your show notes.