Upload your course notes to NotebookLM, click Audio Overview, and a few minutes later you have a two-host podcast discussion about your material. The hosts are AI-generated, the conversation is grounded entirely in what you uploaded, and the result is a polished audio file you can share with students or use as a marketing preview. Since Google launched the feature in 2024, over 100 million Audio Overviews have been played — it became the fastest-adopted feature in NotebookLM's history.
What you’ll walk away with:
- A podcast-style discussion of your course material — for free
- Supplementary audio that auditory learners can use on a walk
- A marketing preview for your course landing page
Why NotebookLM Audio Overview
Most course creators think about producing a podcast and immediately picture hours of recording, editing, and post-production. Audio Overview skips all of that. You upload your existing course content — lesson notes, slide decks, PDFs, even a Google Doc draft of a module — and the tool generates a conversation between two AI hosts who discuss your material as though they are genuinely interested in it. They ask each other questions, build on points, and occasionally simplify complex ideas for an imagined listener.
Three things make it worth trying. First, it is free — included with any Google account, no subscription required. Second, everything the hosts say is grounded in the content you uploaded. They will not invent facts or drift into topics you did not include. Third, the conversational format is genuinely engaging in a way that a text summary is not. Listening to two people discuss your course topic activates a different kind of attention than reading bullet points, which is useful for students who learn better by listening or want to review material during a commute or a walk.
Step by step: Creating an Audio Overview
Upload your course materials
Go to notebooklm.google.com, sign in, and create a new notebook. Upload the materials you want the podcast to cover. This could be your lesson notes for a single module, a set of slides, a chapter from a workbook you have written, or a combination. NotebookLM accepts PDFs, Google Docs, Google Slides, web page URLs, YouTube URLs, and pasted text. The key decision here is scope: the more you upload, the broader the conversation. For a focused, useful podcast episode, one module's worth of material tends to work better than an entire course.
Open Audio Overview and optionally add a prompt
In the notebook, look for the Audio Overview panel (sometimes labeled "Notebook guide" depending on the interface version). Before you generate, you can add a short prompt to steer the conversation — something like "Focus on practical applications" or "Explain this for someone new to the topic." This is optional but helpful. Without guidance, the AI hosts will cover everything in your sources with roughly equal emphasis, which is not always what you want.
Generate and wait
Click "Generate" and give it a few minutes. Generation time varies — typically between two and five minutes depending on the amount of source material. You can leave the tab open and do something else. The tool will notify you when the audio is ready.
Listen and evaluate
Play the result all the way through before sharing it. Listen for accuracy: are the hosts representing your ideas correctly? Are they spending time on the parts that matter most to your students? Note any points where they oversimplify, misinterpret a specialized term, or give equal weight to a minor detail and a central concept. This evaluation step matters — the output is good, but it is not perfect.
Download the audio
If the result meets your standard, download the audio file. You can use it as-is or bring it into a tool like Descript to trim the intro, cut a section that missed the mark, or add a brief spoken intro of your own before the AI conversation begins.
Share as a supplementary resource
Upload the audio to your course as a supplementary resource alongside the written lesson. Label it clearly — "Audio discussion: [Module name]" — so students know what they are getting. You can also publish it as a free preview on your course landing page, share it in a community space, or post it on social media to give prospective students a taste of your material.
The human layer
The AI hosts do not know what matters most in your course. They treat every piece of your uploaded material with roughly equal interest, which means they might spend two minutes on a foundational concept and three minutes on a tangential example. They do not understand your pedagogical priorities — which ideas are prerequisites for others, which points students typically struggle with, which frameworks are the backbone of your entire curriculum.
That means Audio Overview works best as a supplement, not a replacement for your teaching. The generated podcast can introduce a topic, spark curiosity, or give students a different angle on material they have already encountered in your lessons. But the sequencing, emphasis, and depth decisions still need to come from you.
Course creator tips
Use it as a lesson preview students can listen to on a walk
One of the most effective uses is giving students an audio preview before they engage with the written lesson. A student who listens to a ten-minute podcast discussion of next week's module while walking the dog arrives at the actual lesson with context already in place. That priming effect — hearing the key terms and ideas in a casual conversation before encountering them in a structured lesson — can meaningfully improve comprehension.
Upload one module at a time for focus
If you upload your entire course into one notebook and generate an Audio Overview, the hosts will try to cover everything, which produces a scattered, surface-level conversation. Instead, create a separate notebook for each module or lesson and generate individual episodes. This gives you a focused discussion for each topic, and it mirrors how podcast seasons actually work — one episode, one subject.
Use the interactive transcript as a study guide
NotebookLM generates a transcript alongside the audio. That transcript is clickable — students can jump to any point in the conversation. Some course creators share both the audio and the transcript, letting students listen first and then use the transcript as a review tool. For students who prefer reading, the transcript itself can serve as an alternative summary of your lesson content in a more conversational tone.
What it gets wrong
The biggest limitation is emphasis. The AI hosts treat all uploaded content as equally important, so they will spend the same amount of time on a minor supporting detail as on a concept that is central to your course. You can partially address this by adding a steering prompt before generation, but the control is imprecise. If a topic needs to dominate the conversation, the best approach is to upload only the materials related to that topic.
Specialized terminology can trip up the hosts. If your course covers a niche subject — energy healing modalities, advanced dog training techniques, specific therapy frameworks — the AI may mispronounce terms, use them in slightly wrong contexts, or flatten distinctions that matter to practitioners. Always listen through the full audio before sharing it with students who will notice these errors.
The hosts also occasionally get facts slightly wrong. They are synthesizing your uploaded material, not quoting it verbatim, and the synthesis step can introduce small inaccuracies. A statistic might get rounded differently than you presented it. A process you described in five steps might become four. These are usually minor, but in a course context where accuracy matters, they are worth catching before you publish.
Finally, you have no control over pacing or structure. You cannot tell the hosts to spend more time on step three or to skip the introduction. The conversation unfolds however the model decides, and if it does not match your teaching priorities, your only option is to regenerate with different source material or a different prompt.
Frequently asked questions
Is NotebookLM Audio Overview free?
Yes. Audio Overview is included in the free version of NotebookLM, which requires only a Google account. Google also offers NotebookLM Plus with faster generation and higher usage limits, but the free tier is sufficient for creating audio overviews from your course materials.
Can I customize what the AI hosts talk about?
Partially. You can add a prompt before generating the audio to steer the conversation toward specific topics or adjust the audience level. But you cannot script the dialogue, choose which points get emphasis, or control the pacing. The AI hosts decide how to structure the discussion based on what they find in your sources.
How long are the generated audio overviews?
Most audio overviews run between 5 and 15 minutes, depending on the amount of source material. You cannot specify a target length. Uploading a single lesson's notes tends to produce a shorter, more focused conversation than uploading an entire module's worth of material. You can add audio files to Ruzuku lessons as supplementary resources, so students find the overview right alongside the written content.
Give the audio a home inside your course
You have a generated podcast episode that introduces your material in a format students can listen to anywhere. The natural next step is hosting it where students actually take your course. On Ruzuku, you upload the audio file directly into a lesson alongside your written content, video, and exercises. Built-in media hosting means no separate podcast platform to manage — students find the audio overview right where they are already learning.
Pair it with the written version of the same lesson, and you are giving students two paths into the material — which is exactly what the research on multimedia learning recommends.
Related guides
- How to Research Your Course Topic Using NotebookLM — use NotebookLM's core research features to build a grounded course outline
- How to Create Course Voiceovers Using ElevenLabs — generate narration for slides and lessons with AI voice cloning
- How to Record and Edit Course Videos with Descript — edit video by editing text, with AI-powered audio cleanup
- How to Turn Course Content Into a Podcast Using ElevenLabs GenFM — higher voice quality and more control than NotebookLM's audio summaries
- How to Create Your First Online Course — complete guide from idea to launch
- Ruzuku Course Builder — host audio, video, and course materials together in one platform