Write the intro, the ad read or the whole episode, pick a voice from OpenAI or Gemini, and tell it how to sound in plain words. Then put it on a cover image or footage in Studio and export a video podcast — without booking a booth.
1,000 premium characters to try, on the house. No card, no trial clock.
Up to 5,000 characters per generation — roughly five to six minutes of speech. Longer episodes are voiced in parts and joined in Studio; numbers, dates and abbreviations are rewritten into speakable words first.
“Relaxed, like talking to a friend”, “late-night radio host”, “clear and measured for a news recap”. The same direction works on both engines, so you can hear two takes before you choose.
MP3 for every plan; paid plans add lossless WAV. Drop it into your podcast host or your editor as it is.
Studio’s Podcast preset renders 16:9 at 1080p with 256 kbps audio, normalised to −16 LUFS, over a cover image, a slideshow or your own footage — with burned-in subtitles if you want them.
Not a replacement for a host with opinions. A voice for the parts that should sound the same every week.
The same voice, the same delivery, every episode. Change a sentence, regenerate it, and the rest stays as it was.
A sponsor’s copy changes; the read does not have to wait for the next recording session.
History, true crime, explainers and newsletters read aloud — a single narrator, directed line by line.
Generate each speaker’s lines with a different voice, then place up to eight voice clips on one Studio timeline.
Paste the translated script and voice the same episode in another of 59 languages.
Cut a moment, add bold subtitles, export 9:16 for Shorts and Reels with the same voice.
Better to know before you sign up than after.
| ProsodyAI | |
|---|---|
| Turn a document into a two-host conversation | No — it voices the script you write; it does not invent the dialogue |
| Your own recorded voice | No — you choose from the built-in OpenAI and Gemini voices |
| Record or edit a live recording | No — for multitrack editing of real conversations, use a full audio editor |
| Publish to your podcast host | No — you download the file and upload it yourself |
Yes. The audio you generate is yours, and paid plans include the commercial licence — sponsored episodes and ad reads included. The free trial is for listening, not publishing.
A minute of speech is roughly 800–900 characters. Starter’s 100,000 premium characters a month is about two hours of audio; Creator’s 400,000 about eight. Studio editing never draws on the character allowance.
You write both parts. Generate each speaker’s lines with a different voice and place them on the Studio timeline in order; up to eight clips go on one video. There is no button that writes the conversation for you.
Some will, some will not; the delivery is directed rather than flat, which is most of what gives a synthetic voice away. Tell them anyway — say in the show notes that the voice is generated.
Yes. A new account starts with a one-off 1,000 premium characters, no card — enough to hear your intro on both engines.
Paste it, hear it on both engines, keep the one that sounds like your show. 1,000 premium characters to try, no card.