I'm looking for a free way to do AI-powered "text to speech", so I can copy-paste an article or document and have it read to me. I've tried say on the command line, but the lack of AI means it is much harder to comprehend. All of the non-AI TTS-libraries that I've found are also much harder to comprehend than the new generation of AI-powered TTS. Specifically, the new AI-powered TTS services seem to understand some of the meaning and structure of the sentence that is being read, and so they provide better natural cadence. This alleviates stress on the listener to parse the meaning of what is being read out. All of the apps I've come across so far have tried to get me to sign up for services that cost well in excess of $100 per year. My best alternative at the moment is to write my own code to use an API such as the Google Cloud text-to-speech API, but even that will cost me around $0.12 to speak a typical 1300-word article. Perhaps there is some kind of open-source software that works on Mac Silicon, that I can install to generate better speech from text?

In macOS Big Sur and onward there is a built-in accessibility tool for text-to-speech, a floating window with play/pause/speed buttons that can read aloud almost any selectable text on screen. [Have your Mac speak text that’s on the screen](https://support.apple.com/guide/mac-help/have-your-mac-speak-text-thats-on-the-screen-mh27448/11.0/mac/11.0) refers to this box as the 'TTS controller'. However, in earlier OSes, TTS exists but there's no equivalent GUI popup. Does anyone know of a way to make it (or something like it) appear on earlier OSes?

For some reason, my text to speech isn't working. I don't know why it stopped working. It works when I manually click it, but when using the default hotkey alt+esc, it doesn't do anything. I have no idea what happened. How can I fix this? I tried changing hotkeys, but the problem persists.

When this line is read by Siri (Voice 4), found under **Spoken Content > System voice**, it pronounces "comma" after "Africa" instead of just pausing:

bash
say "Illegal mining is common in South Africa, once the world's largest gold producer, now home to thousands of abandoned mines."

Yet, for some reason, it is only that first comma, not the second. I thought maybe it had something to do with the fact that "Africa" is a pronoun. I tried lower-casing it, though, and the result was the same. For now I am just planning on preprocessing all the text read by say to replace , with [[slnc 300]]. This is a possible solution for me since I am calling it from a Go program. But is there a more universal way to eliminate this behavior? In the age of AI, I would expect a more reliable real-time macOS native reading program to exist but I haven't found one. APIs using Google's free service inherit its character limit of 200 characters, which is an unacceptable trade-off for me, and other APIs I've found are even lower quality than say.

I know that Apple Watch comes with Siri, which is capable of understanding spoken phrases and respond with a short answer. But can it also read an entire website or web page, similar to Android apps like Culinary Pal or Wiki Pal?

I am trying to use Live Speech to easily dictate things repeatedly instead of having to say them I see a signet icon at the right of the Live Speech window, but I cannot understand how to use it. When I click it, it does nothing and the sentences disappear after having been spoken by the computer. Can one bookmark their Live Speech items for future reuse? How does it work? ---

The default voices in Monterey are stored in /System/Library/SpeechBase/Voices/. I used **System Preferences > Accessibility > System Voice > Customize...** to install a few more US, UK, and novelty voices, but I can't find where they are stored. Prior to Monterey, the voices resided in /System/Library/Speech/Voices/, which still exists but is empty. The /System/Library/SpeechBase/ directory is apparently new with Monterey, but it does not contain the additional voices I installed. Where are these voices being stored?

In many (if not, most) Mac apps that display or edit text, you can select text and have it spoken out loud, via a right-click or the Edit menu: **Note:** I am aware of the Accessibility setting that assigns a keyboard shortcut to speak text… but it doesn't always behave as you'd expect and sometimes reads text from the page that is not highlighted. Some apps that you would expect to have this ability, don't… which is a pretty big accessibility issue for lots of folks (myself included). Is there any way to forcibly get this functionality into uncooperative apps?

Years ago, I turned on "announce time on the hour," but when it turned out to be stupid (for US locale, instead of "Ten o'clock" it says "it's ten hours") I turned it off. Three days ago, when I wasn't even near the laptop, it started announcing time again. I knew I hadn't changed it, but I looked and it was still Unchecked. There have been no recent updates to the system to explain the change in behavior. I tried turning it on and back off, but it continues to announce. Before I turned it off again, I also changed "on the hour" to "every quarter hour" but it is still only doing it on the hour (when it shouldn't do it at all). Why? and How to fix? I moved Library/Preferences/com.apple.menuextra.clock.plist to /tmp but that changed nothing (and changing the setting did not recreate it). I opened that file in Xcode, and it contains only two of the eight time preferences.

Thanks to this answer , I was able to grant permission to the Terminal and iTerm2 apps to use a Personal Voice with the 'say' command; e.g. say -v MyVoice "Hello, world!" The say command has a number of options, including -o output.aiff which writes the spoken text to an AIFF file. say -v Alex -o test.aiff "Hello world" afplay test.aiff However, I can't get it to work for my personal voice: say -v MyVoice -o test.aiff "Hello world" afplay test.aiff produces nothing but a moment of silence. It seems most likely, judging from the number of loose ends I've hit playing with the Personal Voice feature, that this is a bug, but perhaps I've overlooked something.

I've finished training MacOS 14.0 on my Personal Voice ("Myself"), and Live Speech correctly speaks phrases in my voice using the Live Speech interface. In **Terminal**, the command say -v '?' correctly lists my voice, but the command say -v Myself "Hello, World"; echo $? produces no audio output and indicates normal completion (echo $? => 0). I've tried rebooting. Is this a bug? Is there an additional step needed?

Mac OS X 13.2.1 Lately, I have been having trouble with some voices on my Mac mispronouncing words depending on context. An example is the word *headlines*, which is pronounced correctly by itself, but is mispronounced by an assortment of voices (e.g. Kate) in this sentence: **Just last week, a Federalist Society event at Stanford Law School made national headlines after protesters heckled U.S. Circuit Judge Stuart Kyle Duncan, a Trump appointee to the Fifth Circuit, causing him to cut his remarks short.** I've been fixing these issues as they crop up by giving the offending words alternate or phonetic spellings which somehow evade whatever the bug is. But now I've stumbled over a sentence that one particular voice won't read at all. This command produces no output, audio or otherwise. **say -v 'Tessa' 'Just last week, a Federalist Society event at Stanford Law School made international headlines after protesters heckled U.S. Circuit Judge Stuart Kyle Duncan, a Trump appointee to the Fifth Circuit, causing him to cut his remarks short.'** That is, replacing the word *national* with *international* in the problem paragraph produces something completely unreadable for Tessa. Using the 'Tessa (Enhanced)' voice produces the same result. Curiously, if I change *international* to *enternational*, the voice can read the sentence but it skips the next two words after *enternational*. My question is, are these voice modules complete black boxes? A mispronounced word is easily detectable in playback, a dropped sentence or paragraph isn't. I need an error code or an exception thrown to let me know that something has gone badly wrong so that I can choose another voice in those cases. Update: The bugs are still present in Sonoma 14.1.

I'm on macOS 11.2 using a 16 inch 2019 MacBook Pro. I noticed when I use the text to speech functionality (system preferences > Accessibility > spoken content), the Siri voice doesn't adjust it's speech rate when using the slider (or the button on the popup controller). One thing I've noticed is that any other non-siri voice works. Furthermore, the voice was working till I reimaged my computer. Lastly, I have an older mid-2015 mac and it works just fine (even after I completely wiped it). I tried to copy over some plist files from my old mac to see if I could get that to work, but that failed. Any suggestions on how to fix the broken functionality?

I am using the Chrome browser and I know OS X has built-in voice synthesis that can speak to assist visually impaired users. Can I use this feature to have the computer read contents of websites for me? Eg., I'd like to select a couple of paragraphs and have them read. I know Safari has that built-in, but for one, there I don't know how to change the voice (e.g. I'd prefer a male voice), and secondly, I'd prefer to have it with Chrome. Is it possible?

In Ventura, under System Settings > Accessibility > Vision > Spoken Content > System Voice, I have the option to set a single preferred voice. However, I would like to maintain a preferred voice for language 1 and also a preferred voice for language 2, so that toggling languages lets me easily toggle between my preferred voices. I would like this also for switching between different dialects of the same language, E.g. UK English and US English. Where are the settings to do this in Ventura?

*Applescripts for Ventura's new System Settings* My old Applescript stopped working on Ventura. Any ideas on how to fix this script?

tell application "System Preferences"
		reveal anchor "TextToSpeech" of pane id "com.apple.preference.universalaccess"
	end tell
	
	tell application "System Events" to tell process "System Preferences"
		set theCheckbox to checkbox "Speak items under the pointer" of group 1 of window "Accessibility"
		tell theCheckbox
			set checked to true
			click theCheckbox
		end tell
	end tell

A little background: I'm helping a friend who had a stroke a few years ago and who has global aphasia; as a result it's very difficult for his brain to consistently derive meaning from written words and letters. (You can almost think of it as an extreme form of dyslexia.) He has no trouble with understanding pictures or anything spoken. MacOS Ventura has a built-in accessibility tool for text-to-speech, a floating box with play/pause/speed buttons that can read aloud almost any selectable text on screen. Apple docs refer to this box as the 'controller' . This is a lifesaver for my friend! But the option to make the window appear is buried in a system settings menu tree, to the point where it's impossible for him to open it again on his own if he accidentally closes it. There is a keyboard command to re-open it that can be customized, but because of my friend's written language processing disorder and some motor issues, **it's not viable for him to press the correct keys on the keyboard. I need something on the screen that he can double-click.** My best thought for the simplest solution was to use Automator to set up a desktop shortcut that would automatically re-open it if it gets closed. Has anyone had experience with this? I'm completely open to other solutions to make the TTS controller as easy as possible to summon, without using the keyboard. The options box within System Settings / Accessibility that the TTS controller is triggered through. "Show Controller: Always" is currently selected, but this does nothing to prevent the controller from being accidentally closed. The TTS controller. I have very little experience with Automator, but from my explorations so far, I don't see a way to access very specific parts of the system settings. I've tried recording the cursor, but if the system settings window is in an unexpected place, it doesn't work. Ventura has Shortcuts, but it seems like the same issues there -- no pre-baked way to access system settings.

In the voice list output by say -v ?, some names can speak multiple languages. (The answers on [this question](https://apple.stackexchange.com/questions/3454/say-in-different-language) only deal with single-language voices.) A subset of what I see includes:

Sandy (German (Germany)) de_DE    # Hallo! Ich heiße Sandy.
Sandy (English (UK)) en_GB    # Hello! My name is Sandy.
Sandy (English (US)) en_US    # Hello! My name is Sandy.
Sandy (Spanish (Spain)) es_ES    # ¡Hola! Me llamo Sandy.
Sandy (Spanish (Mexico)) es_MX    # ¡Hola! Me llamo Sandy.
Sandy (Finnish (Finland)) fi_FI    # Hei! Nimeni on Sandy.
Sandy (French (Canada)) fr_CA    # Bonjour! Je m’appelle Sandy.
Sandy (French (France)) fr_FR    # Bonjour, je m’appelle Sandy.
Sandy (Italian (Italy)) it_IT    # Ciao! Mi chiamo Sandy.
Sandy (Portuguese (Brazil)) pt_BR    # Olá, meu nome é Sandy.
Sara                da_DK    # Hej, jeg hedder Sara. Jeg er en dansk stemme.
Satu                fi_FI    # Hei, minun nimeni on Satu. Olen suomalainen ääni.
Shelley (German (Germany)) de_DE    # Hallo! Ich heiße Shelley.
Shelley (English (UK)) en_GB    # Hello! My name is Shelley.
Shelley (English (US)) en_US    # Hello! My name is Shelley.
Shelley (Spanish (Spain)) es_ES    # ¡Hola! Me llamo Shelley.
Shelley (Spanish (Mexico)) es_MX    # ¡Hola! Me llamo Shelley.
Shelley (Finnish (Finland)) fi_FI    # Hei! Nimeni on Shelley.
Shelley (French (Canada)) fr_CA    # Bonjour! Je m’appelle Shelley.
Shelley (French (France)) fr_FR    # Bonjour, je m’appelle Shelley.
Shelley (Italian (Italy)) it_IT    # Ciao! Mi chiamo Shelley.
Shelley (Portuguese (Brazil)) pt_BR    # Olá, meu nome é Shelley.

But say -v Shelley es_ES "Hola amigos" doesn't work, and actually neither does say -v Shelley "Hi there"; both fail with a repeated error like:

2023-04-14 16:15:34.968 say[96121:9557619] Could not retrieve voice [AVSpeechSynthesisProviderVoice 0x60000092cf00] Name: Shelley, Identifier: com.apple.eloquence.en-GB.Shelley, Supported Languages (
    "en-GB"
), Age: 0, Gender: 0, Size: 0, Version: (null)

How do I use this voice and specify which language I want?

I notice that on Sierra, Siri's voice is much superior to the selection of system voices you can select in System Preferences → Accessibility → Speech. In fact, I suspect they use a different system entirely as the command say -v ? doesn't seem to list any voices which are used with Siri. I have a key command set up to speak the select text. Is there no way I can get it to use Siri's voice to do it? **Edit:** To be specific, I'm after the Siri voice: 'British (Male)'

In **System Preferences** > **Accessibility** > **Spoken Content**, I can set a shortcut so that the computer can speak highlighted text. Further, I can change the speed to how fast I want by opening the following and changing it there

open /Users/**/Library/Preferences/com.apple.speech.voice.prefs.plist

I set the voice rate to 800 which is about 700 words per minute. The controller gives the option to speed it down by 0.75 or speed it up by 1.25 I'm wondering if there a way to change those to 0.85 and 1.15 for example? I'm talking about the turtle buttons [in the black controller](https://support.apple.com/en-gb/guide/mac-help/mh27448/mac) .