Audio Recorder
Record a talk and mark the moments worth answering, or rehearse your own comment against the clock — then hear it back. Free, no account, nothing uploaded.
~5 minutes to learnTwo jobs, one recorder
Most voice recorders give you one long file and leave you to find your way back into it. This one is built around the two things people actually do with a recording when they are preparing to speak.
Listening to someone else. A talk, a discussion, a class. You are not trying to keep every word — you are trying to catch the three or four moments worth answering. Tap Mark when one goes by, and the recorder remembers where it was.
Rehearsing yourself. Paste what you plan to say, pick the setting you are saying it in, and say it out loud against that clock. Then hear it back. Reading a comment silently tells you nothing about how it lands; hearing your own voice tells you almost everything.
Getting to it
- Go straight to comment.coach/record.
- From the Dashboard, the recording tile.
- On a phone, the + in the bar at the bottom. It opens the dashboard's list of what you can start — recording is one of them — rather than assuming you meant to type.
- From the main menu, beside the Meeting Timer.
Choose what you are doing
The opening screen offers three ways in, plus your library:
- Record audio — starts immediately. For anything you are listening to.
- Rehearse my comment — asks for your text and your setting first, so it can hold you to a time.
- Record with live transcription — hands you to Live Capture, which turns speech into text as it happens. That one needs a connection and an account; the other two need neither.
There is also Add a recording at the bottom, and again at the top of your library — a Comment Coach card someone sent you, or any ordinary audio file.
Set your target, if you are rehearsing
Paste or type your comment, then pick the setting. Each setting carries its own sensible length, and the screen shows two numbers before you start: estimated from your words, and your target.
The estimate is arithmetic on your word count at an ordinary speaking pace. It is there to tell you whether you have written a comment or an essay — the recording is what tells you the truth.
Watch the level, not the timer
While you record, a bar meter shows what the microphone is actually hearing, with a plain verdict under it: Level is good, Very quiet — move closer, or Nothing is being heard.
This exists because the worst thing a recorder can do is let you find out afterwards. A phone face-down on a table, a laptop across the room, a hand over the microphone — automatic gain cannot rescue any of those, and the meter tells you while you can still move.
Mark the moments
Above the Mark button are four one-tap labels — 👍 Good point, ❓ Ask about, 🔁 Reuse, ✋ Not sure. One tap and the moment is marked and named. Mark itself leaves a bare flag, and Note gives you the same mark with a line of your own attached — “this is the point I want to answer”. Any of them and you keep listening; you are never taken out of the recording.
Marks land where the thought began, not where your thumb did. You tap because of something, and by then it is already several seconds behind you. So the recorder walks back through the level history to the last real pause in the speaking and anchors the mark there. When it has moved one, the mark says from the pause. Without that, every mark points at whatever the speaker said next — the wrong sentence, every time.
Stop, name it, keep it
Done ends the recording and shows you the length. Give it a name — one is suggested from your first note, or from the opening words of what you rehearsed — and press Save. Or Discard, and it is gone.
If the recording came out very faint, this screen says so rather than letting you discover it later. If saving fails for any reason, the audio is still in front of you with a Download button beside the error, so a full disk never costs you the take.
Listen back
The player draws the recording as a waveform you can drag — the shape of your own voice, with your marks standing on it as amber ticks. You can see the pauses before you hear them.
- Drag the waveform to move anywhere in the recording.
- ⏮ and ⏭ jump to the previous or next mark.
- ↺10 / 10↻ step back and forward ten seconds.
- 0.75× to 2× changes the speed without changing the pitch.
- ↻ loops it, so a passage you are working on comes round again without you reaching for play.
- Tap any mark in the list to jump straight to it. The one you are inside is highlighted as it plays.
The same four labels are here too. Marking while listening is how you critique a recording — your own on the second pass, or one somebody sent you — and a mark made here lands exactly where you are, with no anchoring, because listening you already know where the moment is. The ✎ beside any mark writes on it.
Three-quarter speed is the setting most people never think to use and should. Your own filler words are almost impossible to hear at full speed and impossible to miss at 0.75×.
Pinning a memo to where you made it
Some voice notes are about a place — where you parked, a sign you want to read again, the shop you meant to come back to. After you stop, under the name field, 📍 Add where I am attaches the coordinate to the recording.
It is off unless you tap it. Nothing about your location is collected while you record, or at any other time.
- Give it a name you will understand. The coordinate is stored; the words are yours. “Level 3, blue section” will mean more next week than a street address would.
- Tap the coordinate to open it in whatever map app you already use.
- Remove takes it off again, any time before you save.
It is rounded to about ten metres on the way in. That is enough to find a car park and not enough to point at a front door, and the exact reading your phone gave is discarded rather than stored — so it cannot be shared later, by you or by anyone you send it to.
Sending a pinned memo is a choice you make each time. If a recording has a location, the send panel offers to include it, ticked, with a reminder that anyone you send it to — and anyone they forward it to — will see where it was made. Untick it and the card goes out exactly like any other. If you do include it, whoever opens the card gets the pin properly, not a line of text to copy out.
Your recordings
My recordings lists everything on this device with how much space they take. Tap one to listen. Under each row:
- Share — on a phone, hands the audio to WhatsApp, Messages, email, anything that accepts a file.
- Coaching card — see below.
- Transcribe — turns it into a subject you can comment on.
- Download — the plain audio file.
- Delete — asks once, then removes it.
Past four recordings a search box and an order appear. Search matches names and the notes inside recordings, because what you remember is usually the thing you wrote down rather than what you called the file. The ☆ keeps a recording at the top of every order but one.
Arrange turns on the arrows, so you can put recordings in the sequence you mean to say them. That order is the one exception to the star: in a running order the sequence is the content, and floating a favourite to the top would rearrange the talk.
Select turns on the checkboxes, and then you can act on several at once. Play in order runs them back to back — with Loop if you want the sequence to come round again — and Join into one welds them into a single recording.
Sending a recording to someone
There are two ways out, and which one you want depends on the machine you are sitting at.
On a phone, use Share. The recording goes into the share sheet like any other file and lands in whatever app you pick. This is the good path, and it is what phones are built for.
On a computer, make a coaching card. A card is a single web page file with the audio inside it and your notes laid out beneath, each with its timestamp — plus a line from you at the top if you want one. You send that one file however you normally send files. Opened on a computer, it plays. Opened anywhere at all, it reads: the notes and their times are the document, and the player is a bonus on top.
Being sent one
If someone sends you a card, you can just open it — on a computer it plays straight away, with no app and no account. If you want to keep it, open comment.coach/record, choose Add a shared recording, and pick the file. It joins your own library with their notes and timestamps intact, and from there it behaves exactly like something you recorded yourself: scrub it, jump between their marks, transcribe it, send it on.
You can drop the file anywhere on the recorder page instead.
A recording that did not come from here
Any ordinary audio file works: an .m4a, .mp3, .wav, whatever someone sent you of a talk they gave. There are
three ways in, and they all do the same thing: Add a
recording at the top of your library, the same on the opening
screen, or drop the file anywhere on the page.
It is measured as it arrives — its true length, and the same waveform the recorder would have drawn — so it behaves exactly like one made here: drag to scrub, change the speed, loop a passage. And you can mark it. Play it, tap ❓ Ask about when something needs a question, write a note on it, and you have a critique with timestamps that you can send back as a coaching card.
Rehearsing a whole talk, not one comment
Once a subject has points and you have chosen which comment says each one, its outline offers Rehearse N points. That walks the whole thing in the order you will actually say it — down through each main point and its sub-points, not all the headings and then all the detail — and records each one as its own take.
Only points you have decided are included. A point still holding three drafts you have not picked between is left out rather than guessed at: the count on the button tells you how much of the talk is actually ready, which is usually the more useful number.
- As you go, a bar shows where you are: point 3 of 7, what this point was allotted, and what the whole talk was allotted.
- When you finish, the takes are already in order in your library, and Play in order plays the talk back end to end. Nothing to arrange by hand.
- Stopping early is fine. Every take you have made is kept.
The useful part is the arithmetic. A stopwatch tells you one comment ran long. The outline tells you the talk is four minutes over and shows you which point is spending it.
Turning a recording into something you can work on
Transcribe takes a recording and makes a subject out of it: the words, cleaned up and split into sections you can comment on one by one, with the audio kept as the reference. That is the bridge from “I recorded the talk” to “I have a comment prepared on paragraph nine”.
That step uses AI, so it needs an account and a connection. Everything before it does not.
Keeping them safe
Recordings live in this browser and nowhere else. That is the privacy promise and it is also the risk: a browser can clear its own storage when space runs short, and clearing site data takes them with it.
- Install the app. Adding Comment Coach to your Home Screen or desktop is what lets the browser mark your storage as protected. The recorder tells you which state you are in.
- Download all as a zip. At the foot of your library. One file with every recording in it, properly named and dated. This is the backup.
Tips
- Keep the app in front while recording. On a phone, opening another app that plays sound will take the microphone and end the recording. Everything up to that point is kept, and the recorder says so plainly rather than pretending.
- Nothing playing on an iPhone? Check the ring/silent switch. It mutes this kind of playback, and it is the single most common cause.
- Mark generously. A mark costs one tap and no attention. Five marks make a forty-minute recording navigable; none leave you with forty minutes.
- Rehearse twice. Once to find out how long it really is, once to say it the way you meant it.
- Name it while you remember. The suggested name is fine for today and useless in a month.
FAQ
Does any of this need an account?
No. Recording, marking, listening back, the library, sharing, cards, importing and the zip all work signed out. Live transcription and Transcribe need an account, because they use AI on a server.
Is my audio uploaded anywhere?
Not unless you ask for it. Recording, playback and storage happen entirely on your device. The only thing that sends audio anywhere is Transcribe, and that is a button you press.
Does it work offline?
Yes. Once the page has loaded once, recording and playback need no connection at all.
How long can a recording be?
Long enough for a meeting. The level history kept for the waveform costs about 144 KB for four hours, and the audio itself is recorded at a speech bitrate, so length is limited by your device’s free space rather than by the recorder.
What file format do I get?
An .m4a where the browser supports it, which is everywhere that matters — it plays on phones, computers and in messaging apps without anyone having to think about it.
Can I mark up a recording someone else made?
Yes — that is what importing is for. Add their audio file, play it, and mark and annotate it exactly as you would your own. Nothing about a recording depends on this app having made it.
Does pinning a memo tell anyone where I am?
No. Your browser asks your permission, the coordinate is written to this device beside the audio, and nothing is sent anywhere. No address lookup, no map tile — both of those would mean a request to somebody else's server, and that is the one thing the recorder does not do.
Why is the location not exact?
On purpose. It is rounded to about ten metres before it is stored, which is enough to find the place again and not enough to identify a doorway. Precision that is kept is precision that can end up shared, so the exact reading is thrown away rather than written down.
Can I record a phone call?
No. Phones do not let a browser near call audio, and the laws about recording calls differ from place to place. This records what the microphone hears in the room.