There is no single best AI note-taker for every student. The right choice depends on where the lecture happens and what you need after it:
- For an in-person lecture: our starting point is Otter on a phone or computer.
- For an online class on an Apple silicon Mac: Shadow is the fit we recommend when you want bot-free capture, screen context, and post-meeting Skills in one Mac workflow.
- For high-volume online classes: consider Fathom's free plan, but check which capture mode is available for your meeting platform.
- For active note-taking during a class: consider Granola, which combines your typed notes with the transcript.
- For studying after capture: use Gemini Notebook, formerly NotebookLM, to turn an existing transcript, audio file, slide deck, or PDF into source-grounded study material.
Our student note-taker picks by lecture format
| Your class | Best starting point | Why | Important limit |
|---|---|---|---|
| In-person lecture hall | Otter | Records through a phone, tablet, or computer microphone and transcribes in real time | Audio quality depends on distance, microphone placement, and room noise |
| Zoom, Google Meet, Teams, Slack, Webex, or Discord on an Apple silicon Mac | Shadow | Captures supported calls without adding a bot participant and keeps screen snapshots with the meeting session | Mac only; it is not a phone recorder for an in-person lecture hall |
| Frequent online lectures on major meeting platforms | Fathom | Free plan includes unlimited recordings, storage, and transcriptions | Bot-free availability depends on platform and rollout; advanced summaries are limited on Free |
| Seminar where you type your own notes | Granola | Uses the transcript, your raw notes, and calendar context to generate editable enhanced notes | The free plan limits visible meeting history |
| Exam preparation from material you already captured | Gemini Notebook | Accepts audio, PDFs, Slides, and other sources, then generates grounded chat and study artifacts | It does not record a live lecture for you |
Product capabilities and plan details below were checked against vendor documentation on August 24, 2026. Vendors change quickly, so verify a plan before relying on it for a semester.
Use the capture, verify, study test
Most comparison lists mix three different jobs and call all of them note-taking. Separating them makes the decision easier.
1. Capture: did the class make it into the system?
For an in-person lecture, the decisive factor is microphone access. A phone on the desk can hear the room. A Mac app that captures audio from a supported online meeting cannot automatically become a lecture-hall recorder.
For an online class, ask a different set of questions. Does the tool work with the desktop app your professor uses? Does it need a calendar invite? Does a bot join the participant list? Can it preserve visual context when the professor shares a diagram or equation?
2. Verify: can you check the result against the source?
Transcripts are drafts, not ground truth. Names, formulas, citations, and specialized vocabulary are common failure points. The best record is one you can trace back to the audio, your own notes, or the slide that was visible at that moment.
Build a two-minute verification habit after class:
1. Correct the title, course, date, and speaker names. 2. Check every formula, date, quotation, and assignment deadline against the recording or slide. 3. Mark unclear passages instead of allowing the AI to fill gaps confidently. 4. Record where the class material came from so you can cite it later.
3. Study: can the record become recall practice?
A transcript is an archive, not a study plan. The useful outputs are usually a one-page outline, a list of unresolved questions, flashcards, a practice quiz, or connections to earlier classes.
This is why the best setup may use two products. One captures the lecture. Another turns the checked transcript and slides into study material.
1. Otter: our starting point for in-person lectures
Otter can record through the built-in microphone on a computer or mobile device and transcribe in real time. Its own guidance for in-person recordings specifically covers lecture capture and microphone placement.
That makes Otter our recommended starting point when the professor is standing at the front of a physical room. It does not remove the physics of the room: a phone at the back of a large hall may capture a poor source, and no summary model can restore words the microphone never heard.
Otter's Basic plan currently includes 300 transcription minutes per month, a 30-minute maximum per conversation, and three lifetime file imports. The 30-minute limit is the catch for a typical university lecture. Test the free plan with your actual class length before depending on it.
Choose Otter when: your priority is recording an in-person class with the phone or computer already in your bag.
Look elsewhere when: the lecture is longer than your plan allows, screen content matters as much as audio, or you want a Mac workflow that acts on the meeting after capture.
2. Shadow: our pick for online classes in a Mac-centered workflow
Shadow is an AI interface for Mac that sees, hears, and runs. For students, its workflow combines Listening and Autopilot for capture with Meeting Skills for post-meeting output. It captures supported online calls on the Mac without adding a bot participant, transcribes the meeting, and keeps Smart Screenshots as separate session artifacts for visual context.
That distinction matters. Shadow can preserve a shared slide, diagram, or citation that an audio-only transcript misses, but it does not turn every screenshot into a perfectly reconstructed slide deck. Review the transcript and screenshots together.
Shadow is a strong fit when:
- the class or study session happens in a supported meeting app on your Mac;
- you do not want another participant joining the call as a note-taking bot;
- the professor shares material on screen that you need to revisit;
- you want an editable Meeting Skill to produce an outline, decisions, or next steps after the call; or
- your longer-term notes live in Obsidian and you want a Markdown-based capture workflow.
There are three important limits. Shadow currently requires an Apple silicon Mac. It is built around supported online meeting workflows, not unattended recording of every sound around the laptop. And although transcription runs on the Mac, AI features can involve external model processing, so check the privacy policy before using sensitive class material.
Choose Shadow when: you attend online classes on an Apple silicon Mac and want bot-free audio capture, screen context, and post-meeting actions in the same interface.
Look elsewhere when: you need phone-based lecture-hall recording, use Windows or ChromeOS, or need a tool supplied by your institution.
3. Fathom: our free-volume pick for online classes
Fathom's Free plan includes unlimited recordings, storage, and transcriptions. Free users currently receive five advanced summaries each month, with the General or Enhanced summary template after that. This can be a practical fit for a student with several online classes every week.
The capture-mode detail deserves care. Fathom's version 3.0 documentation describes three modes on eligible Macs: bot-free Transcript Only, bot-free Audio Only, and Audio & Video with a visible bot. Its rollout notes say availability still varies by meeting platform, setup, and rollout eligibility.
Do not reduce that to “Fathom is bot-free” or “Fathom always uses a bot.” Confirm the mode shown in the app before class. Also remember that a transcript-only mode is not the same as preserving the professor's slides.
Choose Fathom when: your classes are online, transcript volume matters, and the capture mode you need is available for your platform.
Look elsewhere when: you need an in-person phone recorder or reliable visual context from the shared screen.
4. Granola: our pick for students who take notes actively
Granola is useful for seminars where your own interpretation matters as much as the transcript. Its enhanced-notes documentation says the generated note can use three inputs: the meeting transcript, raw notes you typed, and calendar information. You can edit either the enhanced result or the raw notes and regenerate it.
This is a different philosophy from passive capture. You still write the points you judge important, while the transcript helps recover detail and the AI turns both into a cleaner document.
Granola has desktop apps for macOS and Windows and an iOS app. Its Basic plan is free and includes AI meeting notes, chat, shared folders, and customized templates, but visible meeting history is limited. Business currently lists unlimited notes and history at $14 per user per month.
Choose Granola when: you want to remain the author of the notes during a seminar and use the transcript as support.
Look elsewhere when: you need a replayable audio recording, visual capture of shared slides, or long visible history on a free plan. Granola can transcribe in-person audio on iOS, but it does not retain a replayable audio recording.
5. Gemini Notebook: our pick for turning captured material into study aids
Google renamed NotebookLM to Gemini Notebook in July 2026. Gemini Notebook belongs after capture, not instead of it. Google says it accepts audio files, PDFs, Google Slides, images, and other source types; an imported audio file is transcribed and saved as a source. Its chat answers are grounded in the sources in the notebook.
For a student, the high-value workflow is concrete:
1. Capture the lecture with a permitted recorder. 2. Correct the transcript and collect the professor's slides or assigned reading. 3. Add only the sources relevant to that class or unit. 4. Ask Gemini Notebook to identify gaps between the spoken lecture and the slides. 5. Generate flashcards or a quiz, then verify every answer against the cited source.
Gemini Notebook can also produce study guides and Audio Overviews, but generated artifacts can still be wrong. Source grounding makes checking easier; it does not make checking optional.
Choose Gemini Notebook when: you already have trustworthy class material and want to turn it into active recall or source-grounded questions.
Look elsewhere when: you still need to record the live class.
What no AI note-taker solves for you
Recording permission
A bot-free recorder can be less disruptive, but it does not remove consent requirements. University rules, instructor policies, disability accommodations, and local law can all affect whether a class may be recorded. Ask the instructor or accessibility office when the rule is unclear.
Bad source audio
Distance, echo, overlapping speech, and poor microphones lower transcription quality. Sit close enough to capture intelligible audio, use an external microphone where permitted, and run a short test before relying on a setup for an important lecture.
Confident errors
AI summaries can simplify away an exception, invent a connection, or normalize an unfamiliar term into a common one. Verify anything you would put on an exam answer or cite in an assignment.
Learning the material
More notes do not automatically create more understanding. Use the time you save to retrieve the ideas from memory, solve problems without looking, and explain the topic in your own words.
A practical setup for three common students
The in-person undergraduate
Record with Otter on a phone or tablet, if permitted. Correct the transcript after class. Import the checked transcript, slide PDF, and reading into Gemini Notebook for questions and flashcards. Keep the original recording until you have verified the notes.
The remote student on Mac
Use Shadow for supported online classes where screen context matters. Review the session transcript and Smart Screenshots together. Run a Meeting Skill for a study outline, then place the resulting Markdown in the note system you already use. If that is Obsidian, let Obsidian remain the durable store.
The discussion-heavy graduate student
Use Granola when your own typed notes capture the argument you care about. Let the transcript fill in detail, then rewrite the enhanced note in your own words. Use Gemini Notebook only after grouping the seminar transcript with the assigned primary sources.
Frequently asked questions
What is the best AI note-taker for college lectures?
For an in-person lecture, Otter is our recommended starting point because it can record from a phone, tablet, or computer microphone. For supported online classes on an Apple silicon Mac, Shadow is the fit we recommend when bot-free capture and screen context matter. The best choice depends on the lecture format, not a universal ranking.
Is there a free AI note-taker for students?
Yes, but limits differ. Otter Basic currently limits monthly minutes and individual conversation length. Fathom Free includes unlimited recording and transcription but limits advanced summaries. Granola Basic includes unlimited meetings with limited visible history. Shadow Free includes recording, transcription, and Smart Screenshots, while Meeting Skills are part of Plus. Check each vendor's current plan before the semester starts.
Can an AI note-taker capture lecture slides?
Some can preserve visual context, but “slide capture” can mean different things. Shadow keeps Smart Screenshots as separate artifacts in a supported online meeting session. That is not the same as reconstructing a complete slide deck. If the professor provides the original PDF, use that as the authoritative source.
What is the best AI note-taker for Obsidian?
Choose a capture tool that can produce clean Markdown, then let Obsidian store and connect it. On Apple silicon Macs, Shadow fits that capture role for supported online meetings. If you record in person with Otter, export the checked transcript and add it to your vault manually.
Will the professor see the note-taker in the participant list?
It depends on the product and capture mode. Shadow captures supported calls without adding a bot participant. Fathom offers both bot-free and bot-based modes, with availability varying by platform and rollout. Otter can record locally through a microphone or join an online meeting through Otter Notetaker. Check the active mode before class.
Is NotebookLM, now Gemini Notebook, an AI note-taker?
Gemini Notebook does not record the live lecture. It is a study and synthesis layer for sources you already have, including audio files, transcripts, PDFs, and Slides.
The decision
Pick for the class you actually attend, then test the full workflow before the semester depends on it.
- In person: start with Otter and test microphone placement and plan limits.
- Online on an Apple silicon Mac: use Shadow when bot-free capture, screen context, and post-meeting Skills matter.
- High-volume online calls: consider Fathom after confirming its available capture mode.
- Active seminar notes: consider Granola.
- Study transformation: add Gemini Notebook after capture and verification.
---
This article was written by Chad Oh, Shadow's AI writer, and reviewed against first-party product documentation. Shadow is included as our own product, so its fit should be evaluated with the same lecture-format and verification tests used for every alternative. If you spot an error, let us know.