How to transcribe lecture recordings: a seven-step student workflow
A seven-step student workflow to transcribe lecture recordings free in your browser, turn the text into notes and search a whole term's lectures at once.
To transcribe lecture recordings, record on a phone near the front, copy the file to a laptop, and run it through a transcription tool that works on your own computer, so a 90-minute lecture goes through in one piece with no upload and no length cap. Then turn the text into notes using the lecturer’s own signposting, file it with the rest of the term, and search across all of it before exams. This guide walks through those seven steps, using FreeTranscribe for the transcription step because it’s free and the recording never leaves your machine.
1. Ask before you record
Recording policies vary. Some lecturers are happy for you to record, some want to be asked, and some institutions have a written policy that covers it. Some lectures are already recorded and posted by the institution, which saves you the trouble. Check the module handbook or the student portal first, and if nothing is written down, ask at the end of the first session.
Be clear about what you’re asking for: a recording for your own study. Sharing it is a separate question, so don’t post recordings or pass them around, even in a group chat.
One more thing worth saying plainly. Students with hearing or processing differences often rely on transcripts rather than treat them as a bonus, and many institutions provide official captions or transcripts on request through their accessibility or disability service. If that applies to you, ask. An official transcript is usually corrected by a person, and it’s the one to keep.
2. Record well, from the front
The speech model’s weak spots are noise, accents and technical terms, and a phone at the back of a lecture theatre gives it all three at once. The recording is the biggest thing you control.
Sit near the front. Put the phone on the desk, screen up, so nothing covers the microphone. Switch on airplane mode so a call doesn’t interrupt the recording halfway through. Use the phone’s built-in voice recorder and check what it saves; M4A and MP3 both work with the tool.
Do one test recording on the first day: record 30 seconds, then play it back with headphones. If you can’t make out the lecturer clearly, the model won’t either, so move closer next time.
Name the file the moment the lecture ends: module code, date, topic, like ECON101 2026-10-03 supply curves.m4a. That habit is what makes step 6 work.
3. Get the file onto a laptop
The transcription runs on desktop Chrome or Edge, not on a phone, so the file has to move. Use whichever route your phone already uses: a USB cable, the cloud folder it syncs to, its nearby-sharing feature, or an email to yourself.
Make one folder per module on the laptop and drop the recording in. Moving it through your own cloud drive is your choice; the transcription step itself uploads nothing.
4. Transcribe lecture audio locally, with no cap
Open the tool in desktop Chrome or Edge and pick the file. The first run downloads OpenAI’s open-source Whisper model, about 200 MB, which the browser caches for next time. The recognition then runs on your graphics card through WebGPU, and the recording never leaves your laptop, which matters when the lecturer asked you not to share it.
On speed: in our test a desktop with a graphics card ran at about 1.5 times real time, so a 90-minute lecture takes roughly an hour. A thin laptop takes about as long as the lecture itself. Leave the tab open and do something else. There’s no 30-minute cap and no per-minute charge, so a double lecture goes through the same way as a short one. Why free transcription sites cap you at 30 minutes explains why that matters.
The limits, stated plainly: desktop Chrome or Edge only, English only for now, and it’s the base model, so a chemistry lecture will come back with some reagent names spelt as they sound. Phones, Firefox and Safari aren’t supported yet.
When it finishes, export TXT for the notes in the next step. If you also want to jump back to a moment in the audio, export SRT too; every cue carries a start time. Read the transcript through before relying on it.
5. Turn the transcript into notes
A raw transcript is a wall of text. The notes come from three searches through it.
Headings from the lecturer’s signposting. Lecturers announce structure out loud: “there are three reasons for this”, “the second point”, “to summarise”, “the key thing to remember”, “this will be on the exam”. Search the TXT for those phrases. Each hit is a heading. Under each heading, cut the transcript down to its claim, in your own words where you can.
A definitions list. Search for “is defined as”, “we call this”, “which means”, “in other words” and “that is”. Each hit is a term and its definition, in the lecturer’s words. Put them in a list at the bottom of the notes. Check the spelling of each term against the slides or the reading list, and use find and replace for a term the model got wrong more than once.
Questions to ask. Anything you still don’t understand after reading it twice goes into a short list for office hours or the seminar, with the SRT timestamp if you exported one.
Budget 20 to 30 minutes for a 90-minute lecture. That’s a planning figure, not a measurement; a lecture built on equations or diagrams takes longer because the transcript can’t see the board.
6. Build a searchable archive for the term
The point of transcribing every lecture rather than just the hard ones is search, and search only works if the files are tidy.
Keep one folder per module, and inside it the recording, the transcript and the notes for each lecture, all with the same date-and-topic name. Keep the transcripts as plain TXT. Nearly everything can search plain text, and a TXT file will still open in ten years.
Then add one index file per module, index.txt, with one line per lecture: date, topic, three or four keywords. It takes a minute after each lecture and gives you a table of contents for the term.
To search across the folder, use anything that searches file contents: most text editors have a find-in-files command, and your operating system’s file search may index plain text too. Try it once in week two, so you know it works before you need it.
7. Use the archive for revision
This is where the earlier steps pay back. Search the whole term’s folder for one phrase. “Marginal cost” across twenty lectures gives you every place it was said, in the lecturer’s own words, in the order it was taught. You can see how a concept was introduced in week 2 and how it was used in week 9.
Build the module glossary by merging the definitions lists from each lecture’s notes. Pull every “this will be on the exam” hit into a single list. Listen to the audio only for the passages the transcript flags as unclear, jumping to them with the SRT timestamp rather than scrubbing.
Reading is a lot faster than listening, and searching is faster than either. That’s the case for starting in week one rather than the week before exams.
Here’s the workflow at a glance. Times are planning figures for a 90-minute lecture, not measurements.
| Step | Time it takes | What you end up with |
|---|---|---|
| 1. Ask before you record | A few minutes, once per module | Permission, or a link to the official recording |
| 2. Record well | The lecture, plus a 30-second test on day one | A clear M4A or MP3, named by module, date and topic |
| 3. Get the file to a laptop | A couple of minutes | The recording in the module folder |
| 4. Transcribe locally | About an hour on a desktop with a graphics card, about 90 minutes on a thin laptop | A TXT transcript, plus SRT if you want timestamps |
| 5. Turn it into notes | 20 to 30 minutes | Headed notes, a definitions list, questions to ask |
| 6. Archive it | A minute per lecture | A searchable folder and an index file per module |
| 7. Revise from it | As long as you like | Search across the term, a glossary, an exam-hints list |
To start with this week’s lecture, transcribe it in your browser. It’s free, there’s no account and no length limit, and the recording stays on your laptop.
Frequently asked questions
Can I transcribe a 90-minute lecture in one go? Yes. There’s no length cap, so the whole recording goes through as one file: about an hour on a desktop with a graphics card in our test, roughly 90 minutes on a thin laptop.
Can I transcribe on my phone straight after the lecture? Not yet. The tool needs desktop Chrome or Edge with WebGPU, so the file has to reach a laptop first. Phones, Firefox and Safari aren’t supported for now.
Will it get the technical terms right? Not all of them. It’s the base model and technical vocabulary is one of its known weak spots, along with names and heavy accents. Check terms against the slides or the reading list and use find and replace for repeated mistakes.
What if the lecture isn’t in English? The tool is English only for now, so this workflow doesn’t cover other languages yet.
Sources, checked 15 September 2026
- This post makes no claims about a specific institution, law or product, so no third-party sources were needed. The speed and accuracy figures are from our own test of the tool.