MojoPad reads to you. Any page, web clip, PDF, or imported book can be read out loud — on your Mac, with the sentence and word you're hearing highlighted as it goes, so you can follow along or just listen. It's the twin of Dictation: that turns your voice into text; this turns your text into voice.
Open any page and press ⌥⌘R, click the speaker in the format bar, or choose
Edit ▸ Speech ▸ Read Page Aloud. Reading begins from the top, and a small reading bar appears with the
controls. To start partway down instead, put your cursor where you want to begin and choose
Edit ▸ Speech ▸ Read Aloud from Here (⇧⌥⌘R) — or right-click and pick it. That's the quick way to
skip a book's front matter and start at chapter one.
It reads whichever pane you're in. With a page open beside your writing in Split View, select words over there and press the speaker, or right-click in that pane and choose Read Aloud from Here: the reading follows the pane you acted in, not the page you happen to be writing in. Listening to a source on one side while drafting on the other is the point of it.
While a read is running, the bar shows play / pause, a stop square, and your place (sentence 12 / 340). Change the speed on the fly — 0.8× to 2× — and it takes effect from the next sentence. Pick a different voice from the same bar and the read switches to it. Reading also stops on its own the moment you leave the page, so a read always belongs to the page it started on.
System voices — every voice built into your Mac (System Settings ▸ Accessibility ▸ Read & Speak has more to download). They're instant and fully offline. Where your Mac's own name for a voice says it is an Enhanced or Premium download, MojoPad groups those at the top of the picker.
MojoPad Voice — a small neural voice that runs entirely on your Mac for a warmer, more natural read (choose from Heart, Bella, Nicole, Michael, Fenrir, Emma, or George). The first time you pick one, MojoPad asks to download the voice model once (roughly 90–330 MB, depending on your hardware). After that it lives on your Mac and nothing you read aloud ever leaves the machine — the download is the only moment it touches the network.
Writing a paper in English that quotes Japanese sources? Or working on a Mac set to Japanese, in a document that keeps slipping into English? Read Aloud hands each part of a sentence to whichever of your voices can say it — your main voice reads the English, your second voice reads the Japanese, inside the one sentence, without you touching anything as it goes.
It matters because a voice can only read the writing it was made for, and neither kind of voice tells you when it has met writing it cannot read. Handed Japanese, one of your Mac's English voices does not mispronounce it: it says nothing at all. The passage passes by in silence, at the speed of silence, and the reading carries on as though it had been read. A MojoPad Voice does something else — it reads out a short English description of each character in turn, so 研究 comes back as two English phrases rather than a word. Both are a limit of the voice rather than of your writing, and it is why the sentence is split between two voices rather than simply handed over.
Set both voices in Settings ▸ Reading Aloud, before you start — or from the reading bar while a read is running, which takes effect from the next sentence. There are two:
It works in both directions, which is the part worth knowing if your Mac runs in Japanese. The voice you already hear everywhere else on that machine is a Japanese one, and handed English it reads the words with Japanese phonetics. You do not have to nominate an English voice, or swap your two voices around when you change language: the question asked is never “is this Japanese?” but “can the voice in hand say this?”, so the English is handed to an English voice on its own.
The MojoPad Voices speak English, and only English — so this isn't a setting waiting to be found. Choose one and Japanese in the same sentence is read by your second voice: the on-device voice and your Mac's own take turns inside the one sentence, and you hear a sentence, not a seam. If your Mac has no Japanese voice installed, there's nothing to take the turn — add one in System Settings ▸ Accessibility ▸ Read & Speak, and the diagnostic report will tell you whether you have one. Until you do, Japanese passes in silence rather than being read: MojoPad will not hand it to a voice that would announce it instead.
Your Mac has more voices than it has installed, and better ones. Both lists show every voice on the machine, and macOS ships a great many it doesn't install: other languages, and other accents of English — Australian, Irish, Indian, South African. Add them in System Settings ▸ Accessibility ▸ Read & Speak ▸ System Voice ▸ Manage Voices…, then reopen Settings and they are in both lists. A voice whose name carries a note in brackets is the better half of a pair, and it is a download rather than a purchase — the plain versions are what most people have heard, which is why Mac voices have the reputation they do. If you want a particular accent reading your pages, that is where it comes from rather than from the MojoPad Voices.
If you read Japanese, do this one. Your Mac installs the plainest Japanese voice and nothing else, and the same window offers a better version of it — and a second Japanese voice besides — for nothing. MojoPad cannot tell which one you have, so it cannot prompt you: if your Japanese sounds flatter than your English, that is the first place to look, and it takes a couple of minutes.
A word of one character counts. 本, 人, 国 and 私 are each a whole word, and each gets the Japanese voice, the same as a longer term does.
Everything else goes on counting in whole sentences — the place-marker, the progress, the page a PDF turns to — so nothing else about reading changes. With no Japanese voice installed the second picker does not appear at all, and neither does any of this.
Read a note and the spoken sentence and word light up in place — the page scrolls to keep them centered, like a teleprompter. The same follow-along works in web captures, in PDFs (the reader turns pages as it reads), in ePub books (it reads the chapter you're on), and on canvas cards — the speaking card wears a ring. Folded-away content stays silent; what you've collapsed isn't read.
The thirty pages you have to read before Thursday, on the walk to the station. Or the draft of your own chapter, heard back on the drive home, where the sentence that does not work is obvious in a way it never is on screen. Or a paper in two languages that no audiobook app can read, because it does not have your second voice. Edit ▸ Speech ▸ Save as Audiobook… takes whatever MojoPad would read aloud and keeps it: a file with chapter marks, in a player you already have.
It is the same reading. The same sentences, the same voices, the same turn-taking inside a sentence that mixes writing systems — the difference is only that it is kept instead of heard once. Set your voices first in Settings ▸ Reading Aloud; whatever they are, that is what the file will sound like.
Your voices are matched for loudness. Voices are not built to a common volume and they are a long way apart — on an ordinary Mac the quietest system voice sits eight decibels below an ordinary one, and the voice chosen automatically for Japanese happens to be the quietest of all. Reading aloud you never notice, because you set the volume once and listen. In a saved book you would: every change of language would be a change of volume, in a file you cannot adjust as it plays. So each voice is measured before the reading starts and brought to the same level, and the join between them is a change of voice and nothing else.
Chapter marks are the whole difference between an audiobook and a long slab you scrub through, so MojoPad takes them from whatever the source already knows rather than guessing:
A short breath is left between chapters, and the mark sits on the first word rather than on the pause, so jumping to a chapter starts at the chapter.
The edited collection where three of the eleven chapters are in your field. Or the monograph you want without its forty pages of references, because a reference list read aloud is twenty minutes of surnames. Or the report you only need the findings from. Choose Save as Audiobook… and MojoPad shows you what it found, and asks. A document that divides into only one part goes straight to saving, as it always did — a note with no headings in it, with nothing selected, has nothing to choose between.
Everything is chosen to begin with, so pressing the button straight through gives you the whole document exactly as it always did. Uncheck what you do not want. Choose all at the top clears everything or brings it all back, which is the quick way to take three chapters out of thirty: clear it, then check the three.
Each part shows more than its name, because a name is often not enough — in a carelessly made book "Chapter 11" and "References" are equally uninformative. So every row also carries how long that part will take to read and the words it opens with, which between them settle what a part actually is at a glance. A part with nothing readable in it is not in the list at all — a page that is all picture holds no words to read, so it is left out before you are asked, which is why the page numbers in a scan's list can skip.
The running total under the list is the whole point. It is the length of what you have chosen, not of the book, and it changes as you check. Everything downstream follows the same choice: the room MojoPad asks for before it starts, the warning about a long reading, the check that there is space where you are saving it, and the progress on the card in the corner are all worked out from the parts you kept. Drop nine tenths of a book and MojoPad asks for a tenth of the room — which matters most on the machine where the whole book would not have fitted.
The chapter marks in the finished file are the parts you chose, in the order the document has them, named as the document names them. There is no gap where a dropped chapter used to be and no mark pointing at a chapter that is not in the file.
Quit part way and your choice comes back with the reading. What you chose is settled before any of the work starts, so a reading continued days later carries on through exactly the parts you kept — there is nothing to choose again and no way for a dropped chapter to reappear. Stop is different: it throws the reading away, so starting again is starting again, and you choose from the whole document once more.
The four pages of a long paper that are actually about your question. Select the words in the document first, then choose Save as Audiobook…: the list has two buttons at the top, and Just what you selected makes a recording of that passage and nothing else. It works wherever you can select — in a note, in a PDF, in an ebook.
The whole document stays chosen unless you say otherwise. Selecting text is something you do all day without meaning anything by it, and a recording is hours of work and a file, so MojoPad never quietly narrows one — the button is there, and it is yours to press. A passage is saved as a single part under the document's own name; if you want it split up, choose the parts from the list instead. Checking is remembered on both buttons, so you can see what the selection would give you and come back to the parts you had already chosen.
A book is hours, not seconds — it is a real reading, at reading speed, and there is no way around that. So it does not hold the app hostage while it happens. Choose where the file goes and MojoPad gets on with it in the background while you carry on working; a small card in the corner shows which book, how far in, and how much sound exists so far, with Pause and Stop on it.
Quit part way through and nothing is lost. The reading made so far is on disk, and the card comes back when you next open MojoPad with a Continue on it. That is the difference between a feature you can start on a Friday and one you have to sit and watch.
The diagnostic report has a Saved readings section: which readings are in progress, how far each has come, and — the part you cannot hear from the outside — which voices actually read it. If a book came out in one voice when you expected two, that is where it says so. It also names how many parts that reading has and how many it has reached, which is the quickest way to confirm that a reading you chose three chapters for is a reading of three chapters.
This is for listening to your own material: your notes, your drafts, the papers and the books you are working through. A recording of somebody else's book is still somebody else's book, and making one to hand around is not something the feature is for.