TTS on Mac

mac_tts.py

import time
import objc
from Foundation import NSObject, NSRunLoop, NSDate
import AVFoundation
import Cocoa

# https://gitlab.com/gltd/say-cocoa/-/blob/master/test.py?ref_type=heads
def main1():

    class SpeechSynthesizerDelegate(NSObject):
        def init(self):
            self = objc.super(SpeechSynthesizerDelegate, self).init()
            if self is None: return None
            self.finished = False
            return self

        def speechSynthesizer_didFinishSpeechUtterance_(self, synthesizer, utterance):
            self.finished = True

    synth = AVFoundation.AVSpeechSynthesizer.alloc().init()
    delegate = SpeechSynthesizerDelegate.alloc().init()
    synth.setDelegate_(delegate)

    utterance = AVFoundation.AVSpeechUtterance.speechUtteranceWithString_("The quick brown fox jumped over the lazy dog.")

    # enable via Accessibility > Spoken Content > System voice
    voice = AVFoundation.AVSpeechSynthesisVoice.voiceWithIdentifier_("com.apple.voice.premium.en-US.Zoe")
    utterance.setVoice_(voice)

    synth.speakUtterance_(utterance)

    # Run the loop until the speech is finished
    while not delegate.finished:
        NSRunLoop.currentRunLoop().runMode_beforeDate_(Cocoa.NSDefaultRunLoopMode, NSDate.distantFuture())
        time.sleep(0.2)

# deprecated according to docs, but more friendly
# https://developer.apple.com/documentation/appkit/nsspeechsynthesizer
def main2():
    from AppKit import NSSpeechSynthesizer

    voice = None
    # enable via Accessibility > Spoken Content > System voice
    voice = "com.apple.voice.premium.en-US.Zoe"
    # leave as `None` to have a better fallback voice
    # Apple doesn't allow picking a Siri voice for privacy reasons (an app could impersonate siri and phish user info)
    
    # available_voices = NSSpeechSynthesizer.availableVoices()
    # print("\n".join([v for v in available_voices if "premium" in v]))

    text = "The quick brown fox jumped over the lazy dog."
    
    synthesizer = NSSpeechSynthesizer.alloc().init()
    if voice is not None:
        synthesizer.setVoice_(voice)
    synthesizer.startSpeakingString_(text)
    while synthesizer.isSpeaking():
        time.sleep(0.2)

if __name__ == "__main__":
    main1()
    main2()

Personal voice

List the voices and you'll see one like com.apple.speech.personalvoice.<UUID>. In order to use it, you need to also give the terminal permission to use Personal voice.

https://apple.stackexchange.com/questions/464984/how-do-i-access-macos-sonoma-personal-voice-from-command-line

Python-ified version of above is to just make a new script with

import AVFoundation
AVFoundation.AVSpeechSynthesizer.requestPersonalVoiceAuthorizationWithCompletionHandler_(None)

to trigger the permission popup.
For some reason, if I just do this in the script that does the TTS, then there will be a crash.