How to record a lecture and turn it into a transcript

The short version

If you read one section, read this one. Everything below is detail on why each step matters.

1. 1Check your university's rules on recording teaching before the first session, not during it.
2. 2Sit in the front third of the room, as central as you can get. Distance beats every setting on your phone.
3. 3Put the phone flat on the desk, screen up, bottom edge pointing at the lecturer, nothing on top of it.
4. 4Turn on Do Not Disturb. An incoming phone call cuts a recording short on both iPhone and Android.
5. 5Start recording a minute early. The first minute is where deadlines and the reading list get mentioned.
6. 6Lock the screen and take your own notes anyway. Recording carries on while you write or photograph a slide.
7. 7Stop at the break, start again after. Two clean files beat one file with twenty minutes of corridor noise in it.
8. 8Import the audio the same day, read the summary, and fix the technical terms that came out wrong.
9. 9File the note into a folder for that module, so a whole term stays searchable in one place.

Ask first: what universities actually require

Settle this before the first lecture. Most universities publish a policy on recording teaching in the student handbook. Some allow it for personal study by default, some want you to ask the lecturer, and almost all forbid sharing or uploading the file. Personal study and redistribution are treated very differently, and it is redistribution that gets people in trouble.

If you have a documented accessibility need, there is normally a standing arrangement that already permits recording. Use it rather than asking each lecturer separately. Seminars are a different case, because you are recording other students rather than someone paid to stand at the front, so say out loud at the start that you are recording for your own notes. The wider legal picture is in do you have to tell people you're recording.

Sort the policy question once, at the start of term, rather than lecture by lecture. How permission, university policy and a documented accommodation play out across a whole term is covered in how students use an AI note taker across a term.

Where to sit, where to put the phone, and how to beat the room

A phone microphone is good, but not at distance. Every metre between the lecturer and your phone brings in more room and less voice, and in a hall with hard surfaces it brings in echo too.

The biggest improvement available to you is free: sit closer. Row three sounds far better than row twelve, and no app setting closes that gap. If the hall is amplified, sitting near a speaker is a decent second option, because you are recording the amplified voice instead of the distant one. Placement on the desk matters more than people expect.

• Flat on the desk, screen up. Face down muffles the microphone against the table. In a bag or a pocket you lose most of the room.
• Bottom edge toward the lecturer. The main microphone on most phones sits next to the charging port.
• Nothing on top. A notebook or a jacket over the phone works like a blanket.
• Away from your own laptop. Fan noise and typing thirty centimetres away are louder in the file than a lecturer ten metres away.
• Away from the aircon vent. A steady hiss is the hardest noise to deal with afterwards, because it never stops.
• In a big echoey hall, get closer or get amplified. Echo is the one problem software cannot fix. A sound that arrives twice, half a second apart, confuses word boundaries.

How to record a lecture on iPhone

You have two routes. Voice Memos is already on your phone and free. It records the lecture, and on iOS 18 and later it also writes out the words: tap the finished recording, then tap the transcript button next to the playback controls, or find it in the three-dot More menu. Nothing to install, no Apple Intelligence needed.

That free transcript is plain text and nothing else. For a ninety-minute lecture that is the limit: no speaker labels for the question-and-answer part, no summary, no timestamps you can tap to jump back to a moment, and no PDF or Word file to revise from or hand to anyone. That is the honest reason students import the audio into a second app afterwards, not because Apple charges for the words. The longer walkthrough of both routes is in how to transcribe a voice memo on iPhone. An app that records and transcribes together removes the import step entirely. The steps below are the Voice Memos route, because it works even if you install nothing.

Either way, locking the screen does not stop the recording on iOS. That matters, because it means you do not need to leave the display burning for ninety minutes.

1. 1Open Settings > Voice Memos > Audio Quality and pick Compressed unless you have plenty of free space. Compressed is fine for speech.
2. 2Swipe down from the top right and turn on Focus > Do Not Disturb.
3. 3Open Voice Memos and tap the red Record button.
4. 4Put the phone flat on the desk, bottom edge toward the front of the room.
5. 5Press the side button to lock the screen. The red indicator stays in the status bar and recording continues.
6. 6At the break, wake the phone, open Voice Memos and tap Stop. Start a fresh recording afterwards.
7. 7Tap the new recording, tap its name at the top of the player, and type the module and date.
8. 8For the free transcript, tap the recording and tap the transcript button next to the playback controls. On iOS 18 and later the text appears under the waveform, and tapping a line jumps the audio to that point.
9. 9For speaker labels, a summary, action items and an export file, tap the three-dot menu, choose Share, and pick your transcription app from the share sheet.
10. 10If the file lives in the Files app or iCloud Drive instead, open your transcription app and use its Import Audio option.

How to record a lecture on Android

Android has one extra step that iPhone does not, and it is the one people skip. Android suspends background apps to save battery, so a recorder that has not been excused from battery optimisation can be paused or killed part-way through a long session. Set it once, per app, and it stops being a problem. The built-in recorder varies by phone: Pixel has Recorder, Samsung has Voice Recorder, and both are fine for a lecture.

1. 1Open Settings > Apps, find your recording app, tap Battery, and set it to Unrestricted. On Samsung, check Settings > Battery > Background usage limits and make sure the app is not in the sleeping apps list.
2. 2In Settings > Apps > your recorder > Permissions, confirm Microphone is allowed.
3. 3Swipe down and turn on Do Not Disturb.
4. 4Open the recorder and tap the Record button.
5. 5Place the phone flat, charging port toward the lecturer.
6. 6Lock the screen, then check that a recording notification stays in the notification shade. That notification is what keeps the recording alive.
7. 7Tap Stop at the break, then start a new recording for the second half.
8. 8Tap Share on the finished file and send it to your transcription app, or open that app and import the file from Files.

Ninety minutes without the phone sleeping, dying, or filling up

A ninety-minute recording is not demanding in itself. What fails is everything around it. Run this check in the two minutes before the lecture starts and you will not lose a session. For planning, a compressed voice recording runs around 30 MB an hour; lossless can be ten times that.

Recording a lecture that switches between two languages

This is where most transcription tools quietly fail, and it affects a lot of students. Two situations are common.

The lecture is in one language and you revise in another. A course taught in German, studied by someone who thinks in English. You need transcription in the source language, then a translation of the summary rather than of the nine-thousand-word transcript, because nobody revises from a raw transcript.

The lecture mixes two languages. Normal in medicine, engineering and business teaching across much of the world: the explanation runs in Arabic, Hindi or Spanish while every technical term stays in English. Most apps make you pick one language before you press record, and whichever you pick, the other one is destroyed. The tool transliterates the second language phonetically, so "differential diagnosis" comes back as nonsense spelled in the first language's alphabet, or it is dropped.

The fix is a model that tags language word by word instead of locking the file to one language. Noter AI works that way, so a bilingual lecture comes back with both languages written properly in one transcript, and there is no language to pick before you start. It transcribes in 60+ languages with detection always on and translates a finished note into 18. How the main apps compare is in the best app for transcribing multilingual meetings, and coverage is in what languages AI transcription can handle.

A strong accent is a smaller problem than people assume. Speech models handle accented speech far better than distance and echo. If a lecturer with a heavy accent transcribes badly, the room is usually the culprit. Sit closer before you blame the model.

After the lecture: import the file and get the transcript

If you recorded inside an app that also transcribes, there is nothing to import and the transcript starts the moment you stop. If you used Voice Memos, a Pixel Recorder, a dictaphone, or audio your department posted, you are importing a file. M4A, MP3, WAV, AAC and the other audio formats iOS recognises all work. There is a longer walkthrough in how to transcribe audio on iPhone.

1. 1Open the app holding the recording, for example Voice Memos or Files.
2. 2Tap Share, or the three-dot menu, on the recording you want.
3. 3Choose your transcription app from the share sheet. The file imports and transcription starts.
4. 4Or open the transcription app first, tap + or New Note, and choose Import Audio to browse Files and iCloud Drive.
5. 5Expect a few minutes for a 90-minute lecture, not an hour.
6. 6Read the summary first, then skim the transcript for the parts the summary flags.
7. 7Correct the specialist terms that came out wrong, then re-run the summary so the fix carries through into the summary and the action items.
8. 8Rename the note to module code, week number, topic and move it into that module's folder.

• One folder per module, one note per session. "BIO201 W04 Enzyme kinetics" beats "Recording 27". By week six, unnamed notes are unusable.
• Search across everything, not one note at a time. The question you have in April is "where did she explain the second derivation", and you have no idea which week that was. Asking across every note at once is compared in asking questions across all your notes.
• Use timestamps and synced playback for the dense parts. Reading a hard passage twice rarely helps. Tapping the line to hear it again, with the lecturer's emphasis and pauses, usually does.
• Photograph what the microphone cannot hear. Slides, board work, diagrams. Audio captures none of it, and "this term cancels here" means nothing on its own.
• Revise from the summary, not the transcript. Export it to Word or PDF and write your own notes on top. Can AI transcribe and summarize a lecture recording walks through that output.
• Then it stops being about one lecture. Naming, foldering and searching only pay off once there are forty recordings behind them. The term-long version of that — which classes are worth recording, how to revise from text instead of re-listening, and what it costs on a student budget — is in how students use an AI note taker across a term.

Troubleshooting: what goes wrong and how to fix it

Six failures cover nearly everything students run into.

• The recording stopped when a call came in. A phone call takes over the microphone on both platforms, so the file ends there. Everything before the call is saved, so check it before assuming it is lost. Prevention is Do Not Disturb, or Airplane Mode if you truly cannot be interrupted.
• Storage filled up mid-lecture. Recording stops and you keep whatever was already written to disk. Check free space before class and switch from Lossless to Compressed, which cuts file size by roughly ten times with no meaningful loss for speech.
• You cannot hear the lecturer in the playback. Almost always distance, not volume. Raising the gain afterwards raises the room noise with the voice, so it will not rescue a recording made from row fifteen. Next session, sit closer, put the phone on the desk rather than your lap, and check nothing was covering it.
• Everything echoes and the transcript is full of half-words. Hard floors, glass and high ceilings reflect sound, so the microphone hears each word twice. Reverb cannot be removed after the fact. The fixes are physical: closer to the source, phone on a soft surface, out of the middle of a wide empty room.
• The transcript spells every technical term wrong. Expected and fixable. Drug names, theorem names and citations are where errors concentrate, because the model is guessing at a word it heard once. Edit the term and re-run the summary so the fix carries through, as in editing an AI transcript and rerunning the summary.
• The Android recorder died halfway through. Battery optimisation suspended it. Set the app to Unrestricted under Settings > Apps > Battery. If the phone also got hot, take the case off and stop charging during the session.

If you would rather do the whole thing in one app

Everything above works with the recorder already on your phone. If you would rather skip the export-and-import step twice a day, an app that records and transcribes in one pass removes it.

Noter AI records the room from an iPhone, iPad or Android phone, keeps going with the screen locked, and starts a recording in one tap from a Control Center widget. When you stop, you get a transcript with timestamps, speaker labels for the question-and-answer section, synced playback so tapping a line replays that moment, a summary, and the readings and deadlines pulled out as action items.

The parts that matter for study come after the transcript. It is editable, so you fix a mis-heard term once and re-run the summary. Output styles reshape the same recording, so one lecture can come back as key bullet points for revision or a formal report for a write-up. AI chat answers questions across every note you own. Folders and note search keep modules apart, and export goes to PDF, Word or plain text.

It costs $9.99 a month or $49.99 a year, flat, with a free trial you can test on one real lecture. There is no team plan and no minimum number of seats to buy into: one student, one phone, one price. Install it for iPhone and iPad or for Android.

The honest limits. Transcription happens in the cloud, so processing needs a connection, though the recording itself is captured offline and uploads later. There is no live captioning during the lecture; the note is produced after you stop. And nothing on a slide or a whiteboard is captured, because it is an audio tool.