Import NotebookLM audio
Turn a NotebookLM Audio Overview into an editable episode.
Import from NotebookLM transcribes an audio file into a draft Episode with an editable script, voiced by two of your Hosts. Any audio file works, not only NotebookLM overviews. For the other ways to start an Episode, see AI Podcasts.
Choose what to bring into Jellypod
| Starting material | Workflow | What you can edit |
|---|---|---|
| Original documents, notes or supported links | Add them as Sources and create a new episode | A newly written script and narration |
| A downloaded Audio Overview | Import an audio file | A transcribed script and new narration by two Hosts |
| A finished recording to keep unchanged | Upload your own episode | Title, description and cover art only |
Importing rebuilds the speech in your Hosts’ Voices. The original voices, music and sound effects are not kept, and transcription can mishear names, words and speaker changes. Notebook Sources, citations, notes and settings don’t come across either. Upload only material you may publish; see content ownership and rights.
Create a new episode from the same sources
For a different explanation, audience or structure, add the original documents, notes or supported links as Sources (a notebook share link doesn’t replace them). Then create an episode and describe the audience and focus, for example: “Use only the attached lesson notes. Explain the three causes of the conflict for students preparing for tomorrow’s seminar.” Generation uses Credits.
Import an audio file
Download the Audio Overview first; Google’s Audio Overview help explains how. Share links are not accepted.
- Open Podcasts in the sidebar and select your Podcast.
- In Create New Episode, select Import from NotebookLM.
- Choose a First Host and a different Second Host. They default to the Podcast’s regular cast, which needs at least two Hosts.
- Drag the audio file into the upload area, or click Browse Files.
- Click Import Audio and keep the dialog open while it uploads, transcribes and imports.
The script editor opens on the new draft. Detected speakers map onto your two Hosts in order of first appearance, and extra speakers cycle back onto the same two. Non-speech sounds such as [laughter] are dropped; add audio tags afterward to direct delivery.
Narration starts right away and uses Credits. Without enough Credits the script is still saved, and you can generate audio from the editor later.
Review and publish
- Compare the script with the recording: names, quotations, facts and which Host speaks each block. Reassign blocks in the script editor.
- After changing text or a Host, regenerate the affected lines. Text edits don’t replace saved audio on their own.
- Check the title, description and cover art, then publish the episode. For a reviewer link first, use Share.
Limits and troubleshooting
Files can be MP3, WAV or M4A, up to 30 minutes and 250 MB. The script is cut at your plan’s episode length.
- “Audio is too long”, “File too large” or “Please upload an audio file”: the file breaks one of the limits above.
- “Episode script truncated”: the audio ran past your plan’s episode length. The truncated draft is still created and an upgrade prompt opens.
- “We couldn’t detect any speech in this audio”: try a different file.
- Hosts unavailable or fewer than two: add at least two active Hosts to the Podcast.
For anything else, contact support.