Subtitles · Hearing but not following

You can hear it.
You just cannot follow it.

The hearing test came back fine, and speech still falls apart the moment there is background noise, or two people talking, or someone speaking quickly down a bad connection. The sound arrives; the sentence does not. Captions help with that, and they help for a reason that has nothing to do with volume.

A hearing test measures the wrong thing for this

An audiogram asks whether you detect a tone in a quiet booth. That is a different question from whether you can pull one voice out of a room, or keep up when speech is fast, clipped or overlapping. Plenty of people pass the first and struggle daily with the second, and clinicians have a name for that gap: auditory processing difficulty.

This page is not a diagnosis and this app is not a treatment. If speech in noise is a persistent problem, an audiologist can test for it specifically, and what they find may point at accommodations far more useful than any software. What follows is only what captions do for the day in between.

Why reading is easier than listening harder

Following degraded speech is reconstruction. You catch most of a sentence, and the rest gets rebuilt from context, fast, while the next sentence is already arriving. It works, and it costs something, and the cost is why you are tired by the afternoon in a way that does not show up on anyone else's calendar.

  • Noise stops mattering. The captions come from the audio the app is playing, not from the air in your room. A café behind you, a fan, someone typing: none of it reaches the transcription, because none of it ever entered the Mac.
  • Speed becomes your speed. A caption holds for about a second and a half plus a tenth per word, so a fast talker is no longer setting the pace. Hold ⌥ and the last few come back to scroll through, for the sentence that went past while you were still on the one before.
  • Overlapping voices get separated. Turning on a new box per speaker means a meeting stops being one undifferentiated stream. Knowing who said what is most of the work when a conversation is being read rather than heard.
  • Bad microphones stop being your problem. Somebody calling in from a car, a laptop mic in a hard room, a compressed line: these are exactly the conditions where processing difficulty bites, and exactly what you can read instead of decode.

What it does not do

The limits matter more here than usual, because the thing most people with this want is help in a room full of people, and that is precisely what this cannot give.

  • It never captions the room. The microphone is not opened, ever. A meeting you are physically in, a conversation at a table, a lecture hall: none of it produces anything on screen. macOS Live Captions can take microphone audio and Live Listen on an iPhone is built for exactly that, and for the room those are the right tools.
  • It is speech recognition, so it fails where you do. Crosstalk, heavy accents and fast colloquial speech are hard for the model too. It will usually be right where you were unsure, and occasionally wrong in the same place you were.
  • It writes speech, not sound. No speaker names, no marking of the doorbell or the laughter. A voice detector deliberately drops everything that is not speech, which is what stops music becoming invented lyrics.
  • It keeps nothing. Captions are drawn and gone. There is no transcript afterwards to check what you thought you heard.

The part that is quietly the point

Nobody has to switch anything on for you. No host enabling captions, no admin approving a policy, no explaining to a room why you are asking. Nothing joins the call and nothing appears in a participant list, so whether you tell people is a decision you keep making on your own terms rather than one the software makes at the moment you need help.

It also runs entirely on your Mac. A cloud captioning service would receive a transcript of every meeting, appointment and call you needed help following, which is a record of your day held by a company you never chose to tell. Nothing here is uploaded, recorded or written to disk.

Questions

My hearing test was normal. Is this even for me?
If you follow speech fine in quiet and lose it in noise, at speed, or over a poor connection, then yes, this is aimed squarely at you. Captions do not compensate for volume; they remove the reconstruction. That is useful whatever your audiogram says.
Is auditory processing disorder what I have?
That is a question for an audiologist, and worth asking if this is a daily problem: assessment goes well beyond a standard hearing test, and what it finds can unlock workplace and study accommodations that software cannot. Nothing on this page is diagnostic, and buying an app is not a substitute for being tested.
Will it help in a meeting room?
No, and this is the honest limit. It reads what your Mac plays and never what it hears, so an in-person meeting produces nothing. It covers the calls, videos, recordings and lectures that come through the machine, which for most people is most of the week but not all of it.
Does it work on phone calls?
On anything routed through an app on the Mac: FaceTime, a WhatsApp or Signal desktop call, a number dialled from the Mac, a voicemail played back. Not a call on the phone in your hand. You read the other person and never yourself, because your microphone is never opened.
Does reading captions get tiring too?
Less than reconstructing sentences does, in most people's experience, but it is not free. It is worth setting the box size and line count to something you can read at a glance rather than squint at, and worth putting the overlay just under the speaker's face so reading and lip-reading stop competing for your eyes.

Read it instead of rebuilding it.

14-day refund, no questions asked

One-time $9, every future update included · macOS 14.2 or later · Apple Silicon
Unlimited live captions, for as long as you own the Mac. No minutes to buy, no account, no subscription.