New: 13 caption styles

Hear a conversation in more than one voice

An AI dialogue generator that writes the lines, or voices your own [Speaker] script as text to speech with multiple voices.
Script editor for the episode Launch date check. Under the chapter The new date, Alex Rivera asks if the launch moved to November 10, and Maya Chen answers that it did and asks what failed in the proofs. Each line shows Regenerate Audio.
Connect your sources

4.6out of 5 on G2

  • Zendesk
  • Salesforce
  • Columbia University
  • Virgin Voyages
  • Xerox
  • Penn State University
  • JetBrains
  • Southwest Wisconsin Technical College
  • Ohio State University
  • Healthee
  • Virgin Media
  • Kaishan USA

Jellypod writes the lines, or voices the script you already wrote

Give it a prompt and the Podcast Agent writes a conversation for your hosts. Paste each turn on its own line as [Maya] or Maya:, and those words are spoken in order.

Upload Script and Import Script both open a speaker review. You match each name to a host before any audio is made.

Each speaker gets a voice, so a question and an answer stay distinct

A podcast, Magic Video, Short, or Voiceover takes up to four hosts, chosen from 3,300+ voices across 121 languages, or a clone of a voice you have the right to use. On a short or video, the review refuses a fifth distinct speaker.

Review script dialog for two detected speakers. Alex is matched to host Alex Rivera, and the open voice menu has Leo selected. Maya is matched to host Maya Chen, whose voice is Nora. Import script is at the bottom right.

Label the speakers, pick up to four voices, then generate

Upload script dialog with a pasted two-speaker script. [Alex] asks whether the launch date moved to November 10, and [Maya] confirms it did and asks what failed in the proofs. The turns sit on consecutive lines, with Add script at the bottom right.

  1. Step 01

    Start from a prompt or a labeled script

    Describe who is talking and what they need to settle, or paste turns that each begin with [Name]. Speaker detection works most reliably when there are no blank lines between turns.

  2. Step 02

    Match every name to a host

    A saved character with the same name is preselected. Assign the rest, or remove a label that is not a speaker.

  3. Step 03

    Generate the multi-voice text to speech

    Edit a line, regenerate that line, then download an MP3 or an MP4. The Conversation template shows host avatars and live captions.

When those lines should become illustrated scenes, continue in script to video. When a PDF or deck should stay on screen while it is read, start with voiceovers.

For a show you want to publish, see the best AI podcast generators.

FAQ

Does Jellypod write the dialogue, or only read a script I bring?
Both. A prompt makes the Podcast Agent write a new script for your hosts, and Upload Script or Import Script speaks the words you already wrote, in order.
How do I mark who is speaking?
Start each turn on its own line with [Maya] or Maya:. Bold, heading, and list labels from Word or Google Docs still count, and unlabeled scripts are detected too.
How many voices can one dialogue use?
A podcast, Magic Video, Short, or Voiceover takes up to four hosts. On a short or video, the script review refuses a fifth distinct speaker.
What is the script import limit?
No script can exceed 75 minutes of narration, and a longer script is refused. A script over the episode length for your plan is refused rather than shortened.
Can I re-voice one line without redoing the whole script?
Yes, hover a speech block and choose Regenerate Audio to replace that line while the rest of the audio stays. Regenerating is free, up to a daily limit that resets at midnight UTC.
Can I download the dialogue as audio or video?
For an episode, Download Audio saves an MP3 and Download Video saves an MP4, and Conversation shows host avatars and live captions. Free video includes a small "Made with Jellypod" watermark, and voiceovers offer video only.
Are stage directions read aloud?
Cues such as [Laugh], [Sigh], and [Pause] fold into the surrounding line, not their own turn. A short header such as Audience: is set aside.
Do I need an account?
Yes, Generate opens signup, and the Free plan starts with 500 credits and no credit card. Paid plans include a 7-day free trial.

Discover more

Explore related tools

Finish the dialogue in Jellypod

Sign up

Start free with 500 credits. No credit card required. Paid plans include a 7-day free trial.

Deep Ocean gouache painting of a podcast studio under the sea

Download the dialogue as audio or video

Edit a line and regenerate only that line. Download Audio saves an MP3. Download Video saves an MP4 of the template you picked.