Short answer first: get a transcript, cut it into headings, write one question per heading, mark the timestamps you did not follow, and replay only those. Do that the same day in about half an hour, answer the questions from memory on day three and day seven, and you will not need to hear the lecture again.
Recording a lecture is easy. Doing something with the recording is where most students stall, because the obvious next step, listening to it again, takes as long as the lecture did and teaches surprisingly little. This is the workflow that replaces it.
Why not just listen to it again?
Because a replay is rereading with a pause button, and rereading is one of the weakest study methods there is. In the 2013 review by Dunlosky and colleagues, rereading was rated low utility while testing yourself and spacing your study were rated high. A lecture replay has the same problem in a slower format: the material feels familiar on the second pass, the familiarity feels like knowledge, and the exam finds out otherwise.
The universities that provide lecture capture say the same. Imperial College's guide to using lecture recordings tells students to rewatch in full only if they missed the lecture, and adds that the act of deciding which parts to revisit is itself part of the learning. Sheffield's lecture capture guide says the same, and adds that the revisiting should happen within two or three days, not in revision week.
There is a cost argument too. A 90 minute lecture is around 10,000 words of speech. As audio, you can only move through it at the speed it was spoken. As text, you can skim it in ten minutes, search it, and delete the half that was administration, repetition and jokes.
What do you do the same day?
Five steps, about half an hour, done before you go to bed on the day of the lecture. The same day matters because the transcript makes most sense while you still remember the room, and because the first review is the one that stops the forgetting curve from doing its worst.
Step 1: Get the transcript
Run the recording through a transcription tool in the language the lecture was given in. Do not read it yet. If your university's lecture capture produces captions, export those. If you recorded on your phone, upload the file to a tool that handles your language; a Danish lecture through an English only engine produces nothing usable. Transcription versus notes covers what a raw transcript is and is not.
Step 2: Structure it
Skim the transcript once and insert headings wherever the lecturer changed topic. Most 90 minute lectures have five to eight. Delete the housekeeping at the start, the repeated announcements and any tangent that will not be examined. This is the step that turns 10,000 words into two pages, and it is the step an AI notes tool automates; if you use one, this is the output to check rather than to write. The habits in effective note taking apply to a transcript as much as to a notebook.
Step 3: Pull the questions out
Under each heading, write one to three questions the section answers. "What are the three assumptions of the model?" "Why does the reaction need a catalyst?" Write the question, not the answer; the answer is in the section, and you will produce it from memory on day three. The Cornell method does this with a cue column and works just as well on a transcript as on live notes.
Step 4: Timestamp the hard parts
Wherever you did not follow the argument, note the time in the recording next to the heading. These are the only passages you will ever replay. Most lectures produce two or three of them, and each is a few minutes long.
Step 5: Answer once, from memory
Close the transcript and answer your own questions on paper or out loud. Mark the ones you could not. Ten minutes. This is the first retrieval, and it is what makes the difference on day seven: in Roediger and Karpicke's 2006 experiments, students who took a recall test after studying a passage remembered far more a week later than students who reread it, even though rereading looked better on an immediate test.
What do you do the week after?
The same day pass produces two pages of structured notes with questions and a short list of timestamps. The week after is three short sessions.
| When | What | How long |
|---|---|---|
| Same day | Transcript, headings, questions, timestamps, one answer pass | 30 to 40 minutes |
| Day 3 | Answer every question from memory. Replay the timestamped passages you still cannot answer. | 10 to 15 minutes |
| Day 7 | Answer every question again, shuffled with the week's other lectures. Rewrite the notes under any heading you still get wrong. | 15 minutes |
| Before the exam | The questions only, on the spaced repetition ladder. The recording is not opened. | A few minutes per lecture |
Day three is when the timestamps earn their place. You attempt the question, you cannot answer it, and you play the four minutes of the recording where the lecturer explained it. That is a replay with a purpose. Playing the whole lecture on day three would take 90 minutes and would revisit the 80 minutes you already understood.
Day seven is when you find out whether the notes are any good. A question you still cannot answer after two attempts usually means the section under it is a summary of what was said rather than an explanation of it; rewrite that section in your own words, from the transcript, and leave the rest alone.
What about the lecture you missed?
The one exception to "never replay in full" is the lecture you were not at. Sheffield's guide is specific: watch it once at normal speed and take notes as if you were in the room, then treat it like any other lecture. The same day workflow applies afterwards, with the difference that the transcript is your first contact with the material rather than a reminder of it, so the structuring step takes longer.
It is worth knowing that recordings do not replace attendance for most students. In a study across all four years of an undergraduate degree, Nordmann and colleagues found that for first year students both attendance and recording use predicted performance, and that recordings helped most as a supplement to attending. Only the strongest students could make up for poor attendance with recordings. If you are in the room, paying attention in lectures is still the cheapest study method available, and the recording is insurance.
What if there is no transcript?
Then the workflow shortens but does not change. Play the recording once, at normal speed, and write headings and questions as you go rather than notes. Pause when the lecturer changes topic and write the heading; pause when a section ends and write the question it answered. It takes the full 90 minutes, which is why a transcript is worth having, but at the end you have the same two pages of headings and questions, and you never play the recording in full again.
Whatever you use for the transcript, get the language right before anything else. Every later step inherits the transcript's errors, and an engine that does not handle your language produces text that looks like a transcript and means nothing.
Where does Notibo fit?
Notibo does steps one to three automatically: it transcribes the recording in 89 languages, writes the headings and a structured summary, and generates question cards you can answer on the day three and day seven schedule. What it does not do is the answering. The retrieval on day three and day seven is still yours, and it is still the part that decides what you remember.