Every Audio Overview feature explained: generation, customization, Interactive Mode, sharing options, downloading, and troubleshooting common issues.
Can: Produce natural-sounding two-host conversations about your sources. Explain complex topics accessibly. Generate downloadable MP3 audio. Support interactive Q&A mid-episode. Cannot: Use custom voices. Guarantee exact topic coverage. Produce episodes over ~20 minutes. Access the internet beyond your sources.
Audio Overview is accessed via the Notebook Guide panel (magic wand icon). After clicking "Generate," NotebookLM reads all sources and creates a conversation script, then synthesizes the audio. Generation time ranges from 1–5 minutes depending on source volume and server load.
The most important feature you may be underusing. Before generating, you can enter a prompt in the customization field. This is not a full prompt interface — you have a character limit and the model will prioritize its default style — but targeted customization can significantly change focus, tone, and audience level. See the Customize Audio guide for prompt templates.
During playback, click the microphone button to activate Interactive Mode. You can speak or type questions. The AI hosts will respond and incorporate your question into the conversation. This mode ends when you click the microphone again or let it time out. Note: Interactive Mode does not extend the episode length significantly — it typically adds 1–3 minutes of responsive content before the hosts return to their planned conversation.
You can share a NotebookLM notebook (including its Audio Overview) via a share link. Recipients can listen without a Google account, but they cannot edit the notebook or generate their own Audio Overviews unless they copy it to their own account.
Click the three-dot menu next to the audio player, then "Download." The file downloads as MP3, typically 5–15MB. You can upload this file to podcast hosting platforms (Spotify for Podcasters, Buzzsprout, etc.) or embed it anywhere MP3 is supported.
Generation fails or loops: Reduce source count or try regenerating with fewer sources. Audio covers the wrong topics: Use the customization field to direct focus, or remove off-topic sources before regenerating. Audio sounds too generic: Your sources may be too broad or superficial — add more specific, dense source material. Can't hear certain details: Audio Overview deliberately simplifies — the chat interface is better for precise extraction.