Generating Episode Audio

Overview

Once the Podcast Agent finishes writing your script, it generates audio for every speech block automatically. Each line assigned to a host becomes its own audio segment, produced with that host's voice, pacing, and personality. Script writing and audio generation happen together in one step.

Credits are only consumed when you publish or download the episode.

How Audio Generation Works

When the Podcast Agent creates your episode, it writes the script and generates audio in a single pass:

1. The agent finishes writing the script.
2. Audio generation begins automatically. While it runs, the script area shows Loading script... and a Generating audio... toast appears.
3. When generation completes, the script appears and a toast reads Audio generated successfully (a single segment) or Audio generated for all segments. when generating audio for the whole episode at once.

Each speech block in the script now corresponds to an audio segment on the timeline. Press play to hear your episode.

Before the Episode Opens

A new episode's editor isn't enterable until this first generation succeeds. While it runs, you watch progress from the episode's row on the podcast page instead; clicking the row shows a reminder rather than opening the editor. If generation fails or stalls without producing audio, clicking the row opens a Generation failed dialog with a Retry generation button, which reruns generation from scratch. The episode opens automatically to the Script Editor once a generation succeeds.

This gate only applies to the first generation of a podcast episode. Slide Voiceovers always open into their editor while generating, since their creation flow lives there.

Regenerating After You've Opened the Episode

Once you're inside the editor, script edits and Take fixes work differently from the initial generation described above.

When the Episode editor agent applies a script edit, Jellypod saves the updated script before regenerating audio:

• A small edit with stable speech-block identity can regenerate only the affected Takes.
• Adding, removing, or reordering blocks, changing more than half of the spoken words, or finding an ambiguous Take mapping regenerates every Take.

Jellypod determines the scope on the server from the saved script and timeline rather than asking the agent or browser to choose. The update keeps running if you leave Studio. A banner above the script shows the current status: a spinner while audio or visuals are updating, or a notice to retry when the last update didn't finish. Editing a speech block yourself doesn't raise a banner; the block's own Regenerate Audio button stays visible until its audio matches the new text. If any required Take fails, Jellypod leaves the previous narration untouched and shows a Retry Audio Update action, which asks you to confirm before it regenerates audio for every segment.

For Magic Video, final narration timing supersedes older visual work. Unchanged narration anchors can reuse their existing generated media; visuals anchored to changed spoken content regenerate. Download and publish controls stay visible but inactive until narration and visuals are current. Clicking either one shows a message explaining what's still in progress, such as narration finishing or a new video style being applied.

For audio regeneration without an agent edit, click Regenerate Audio on one speech block to regenerate exactly that Take, or use the script editor toolbar to regenerate the full Episode.

When Credits Are Used

Credits are consumed at the point of final output:

• Publishing: 60 credits per minute when publishing to your RSS feed, Spotify, Apple Podcasts, or other platforms.
• Downloading: 60 credits per minute when downloading audio (MP3) or video (MP4) files.

You can generate, listen, edit, and regenerate as many times as you want before then. Credits are deducted only when you publish or export.

If you do not have enough credits to publish or download, an upgrade dialog appears. You can upgrade your plan or purchase additional credits at any time.

Limitations

• You cannot generate audio for a published episode. Unpublish first if you need to regenerate.
• Episodes have a maximum duration limit based on your plan. If the estimated duration exceeds this limit, generation is blocked with an error message.

Frequently Asked Questions

Does audio generation cost credits?

No. Credits are only consumed when you publish or download the final episode, at a rate of 60 credits per minute.

Can I regenerate audio for a published episode?

No. You must unpublish the episode first, which reverts it to draft status. Then you can make edits and regenerate audio before republishing.