Wrong System Voice Speak Text and Alerts - Catalina

[SaguaroSeven]

I just set up a new Mac mini that came with Catalina and I cannot get System Voices to work correctly. Have downloaded enhanced Ava, Siri Female, Siri Male, and Tom. Whenever I set the System Voice to either Siri voice, Speak Text (from right click) and System Announcements use a crappy voice (Vicki I think). Although using the Speak Text shortcut (Opt+ESC) gets the correct voice! If I set Tom as Sys Voice, it works as expected. (Announce Time also produces unexpected results.)

I've tried: reinstall macOS from Recovery, remove unwanted voices, remove Speech preferences, Safe Boot, even repair repairHomePermissions. None of these corrected this.

Can anyone help?

 

[casperes1996]

In my opinion, the Alex voice is still the best for fast reading. But anyway that’s beside the point. I never actually use anything other than the shortcuts when possible, but if I recall everything works properly for me, but I’m not on my Catalina install right now.

Siri’s voice at least I can say is changed from a different menu to regular Text to Speech and it can be set to a different voice than your regular speech synth setting; From the Siri preference you can change Siri, but the others are under Accessibility. Try to validate that they match up if you want them to have the same Voice for everything

 

[kriebe]

I've also noticed this same problem and it's really annoying. If I make a keybind to have Siri speak either the entire document or selection, it then works. But if I right click and go to the speech sub-menu and choose "Star Speaking" I just get the Alex voice. If I select any of the non-Siri voices, it works fine. So I deleted the male voices so that they weren't options anymore. Now I get some super-low quality female voice that isn't even one of the options that I still have available. I even tried deleting all voices except Siri and still get the terrible quality female voice. The main reason why I want this to work is that I have some apps that just automatically read the audio for you that the keybind won't work in, so I don't have that option.

I'm assuming that the main problem is that when using the Siri voice, it still sends the text somewhere on the internet and then it comes back. The reason I assume this is happening is that depending on the length of the selected text, it can take a long time to start actually reading if it's a lot of text.

If anything, it would be nice if I could at least set the fallback voice instead of it automatically choosing one of the voices I don't like.

 

[getyoTTSout]

Talked to Apple on the phone today for over an hour and shared my screen. Was not able to solve this issue and the agent has the same issue on the Mac she worked on. Apparently it is a bug known to Apple and she said that she sees a big focus on accessibility updates and expects it to be solved rather sooner than later. Given that Catalina is about to be replaced this is not really credible – the bug is nearly a year old now – at least.

I have had this issue also on an older iPad Mini 2 or so, still on iOS 13 and in the native Books app and there has not been any improvement here in a long time.

It is absurd that TTS fails in Apple's Books app, where you can not get the keybinds to work and need to use the context menu, which then triggers the old voice. I tried a few key combos, not thoroughly testing though, assuming either alt or shift or cmd might be what causes issues in Books. No luck. Can you try?

In /System/Library/Speech/Voices/ you can delete everything "compact" since those are only the sample files. You can also probably delete everything that does not include Siri in its name since those are lower quality voices.
@SaguaroSeven after killing Fred it uses Victoria. Kill her too and it stays quiet.

You probably want Nora, which is the most natural US female Siri.

Can somebody try to rename the files? This might be a workaround that is faster than waiting for Apple to solve this.
Killing the actual Victoria, making a clone of Nora and renaming her Victoria where relevant. That way allowing "Speak Text" to still use its beloved Victoria while secretly giving it Nora. Should even seamlessly work after the fix.

Why does all this remind me of Westworld?

I found someone on Reddit commanding killall SpeechSynthesisServer in context of a similar issue but it does not even cause a reaction Terminal to me. https://www.reddit.com/r/MacOS/comments/cy4tg8 [automerge]1593369662[/automerge]
Seriously, if someone could Dr. Frankenstein the s*** out of Victoria and Nora that might be our best try. I have no idea how to approach this and no idea how the plist files etc. function.
[automerge]1593370202[/automerge]
Can someone confirm this bug persists in Big Sur?
I just tried to simply rename a Nora clone folder to Victoria and that is not enough. You really need to start messing around with the files inside.

 

[getyoTTSout]

I would like to add this:

Make the `say` terminal utility and NSSpeechSynthesizer work with Siri voices

The say command-line utility seems to be unaware of Siri voices as of macOS 11 (Big Sur): say -v '?' doesn't list Siri voices. Targeting a Siri voice explicitly doesn't work: say -v NoraSiri hi!...

stackoverflow.com

 

[casperes1996]

I would like to add this:

Make the `say` terminal utility and NSSpeechSynthesizer work with Siri voices

The say command-line utility seems to be unaware of Siri voices as of macOS 11 (Big Sur): say -v '?' doesn't list Siri voices. Targeting a Siri voice explicitly doesn't work: say -v NoraSiri hi!...

stackoverflow.com

Apple officially says that Siri voices are not "yet" officially supported for NSSpeechSynthesizer. It was even said at this years WWDC though I forget the specific talk. It was something like "Intelligently use speech in your app". It said it is expected behaviour that a fallback voice for the same language will be used

 

[SaguaroSeven]

@getyoTTSout: Thanks for the stackoverflow link. I would comment there but not enough rep. I tried messing with voices in many ways. Deleting all but desired Siri voices, changing permissions, etc. None of it helped. Even removing Aron, etc, I still got a crap voice -- must be hidden somewhere else. Eventually reinstalled Catalina to undo fixes. I'm through messing with Catalina.

@casperes1996: Thanks for this tip and comment. The talk is called "Create a seamless speech experience in your apps."

At a minimum this situation is a UI foul. If a voice can't be used, don't present it to the user. #badUX
I suspect the root cause is moving to new tech, while keeping older tech. Of course this all works fine in Mojave, so...

So the "answer" is: maybe this will be fixed in macOS 11.

Here's hoping. And that quality of this year's sw looks more like iOS 12 than 13 and Catalina. Reports so far are promising.

On the plus side, thanks to stackoverflow, I discovered this, which is pretty hysterical. (try with Siri Nora set as system language)
say -v Nora "Doesn't she look exactly like Laura Palmer?"

 

[casperes1996]

@@casperes1996: Thanks for this tip and comment. The talk is called "Create a seamless speech experience in your apps."

Yup THat's the one; Cheers.

At a minimum this situation is a UI foul. If a voice can't be used, don't present it to the user. #badUX
I suspect the root cause is moving to new tech, while keeping older tech. Of course this all works fine in Mojave, so...

Yeah I think you're right there. Though with respect to not presenting the choice to the user; I think the issue is that it works for some things but not for others. And creating two separate menus one for one type of voice and another for another type of voice with different options also felt like a less than ideal option so they just chose what they felt was the lesser of two evils I guess. Hopefully temporarily.

On the plus side, thanks to stackoverflow, I discovered this, which is pretty hysterical. (try with Siri Nora set as system language)
say -v Nora "Doesn't she look exactly like Laura Palmer?"

I'm on macOS 11; What does this do for you? To me it just sounds kinda Swedish/Norwegian.
Speak Selection is a bit buggy on macOS 11 right now, but they've definitely tried to improved it. I've filed 3 bug reports about different issues with it, but I suspect it will be quite nice once its released. It includes automatic language detection and switching which is super nice if you use several languages. Though it currently doesn't seem like there's an override if it gets it wrong which is a bit sad. It also highlights the spoken sentence and words like you can enable on iOS.
Needs some fixing but it could be good. - The issue from this thread is still there in 11 though and I doubt it will be fixed since they talked about it relating to iOS in that Tech Talk you linked. - Maybe macOS 12

 

[SaguaroSeven]

I'm on macOS 11; What does this do for you? To me it just sounds kinda Swedish/Norwegian.

Sorry to take so long to get back to you... the phrase is kinda a joke, perhaps an obscure one.
Ref:

I hope this (and so many other things) are sorted in Big Sur.

 

[Takuro]

Apparently it is a bug known to Apple and she said that she sees a big focus on accessibility updates and expects it to be solved rather sooner than later. Given that Catalina is about to be replaced this is not really credible – the bug is nearly a year old now – at least.

Thanks for providing all that background. This has been driving me nuts and just sounds like something they've never fixed.

I decided just to map a hotkey in System Preferences > Accessibility > Speech, and that method causes select text to be read in "Siri" voices correctly (male or female, any language). If I right click selected text and choose Speech > Start Speaking, it does default to Alex if a Siri voice is selected as the System Voice.

I mean, it's gotta be a bug at that point. The context menu and keyboard shortcut have no reason to behave differently, and they respect the System Voice setting with total consistency so long as it's any non-Siri voice.

 

[Le0M]

On Big Sur 11.1 here. The issue is still present.

 

[SaguaroSeven]

On Big Sur 11.1 here. The issue is still present.

Thanks for this update. I can't say I'm surprised, but still disappointed.

This isn't the only Catalina issue that goes back to fall 2019. I've been working with Apple Support for 6 months because Catalina often does not respect DND settings while locked/sleeping. This can allow a notification to unlock Mac when Apple Watch is in the next room. The issue was first reported in Oct 2019 on Apple's Discussions, after Catalina was first released. That is also going nowhere....

 

[lsatterfield]

Just jumping in to say that I have the same issue. I chose Siri Male (Erin), and I get Alex.
If I setup "Speak selection" through Settings > Accessibility, I can use the keyboard shortcut (Option + Esc), and it will be Siri Male/Erin. ?‍♂️

 

[sirozha]

There's a major issue in Big Sur 11.0, 11.0.1, and 11.1 (which is the latest at the time of this writing) in that the voice selected as System Voice is not honored when a word is selected and Voiceover is invoked (Option + Esc). For example, if the System Voice selected is one of the French voices, but the primary language in System Preferences > Language & Region > General is English (US), the words spelled the same in English and French are pronounced with an English US voice even though it should be pronounced with the currently selected French system voice.

For example, the word "journal" is now always pronounced by a male English US voice even though I have a female French voice (Chantal) selected in System Preferences > Accessibility > Spoken Content (formerly "Speech"). I've opened a case with the Apple Accessibility department, they sent it to engineering, and it came back as "expected behavior". This is totally bonkers. It worked properly in Catalina and before, and it's a complete mess in Big Sur.

I use the Voiceover/Speech (now called "Spoken Content") feature for learning pronunciation when studying foreign languages. It was a great tool before, and now it's totally messed up. In the example above, I uninstalled all English system voices, but a male English US voice still reads the word "journal" in English even though it should be read by system voice Chantal in French. It's now anyone's guess which words are pronounced with the selected System Voice and which are with a hidden default language system voice. Some obviously non-English-looking words are still pronounced by the male English US voice and are obviously completely butchered. If the word is clearly French, then System Voice Chantal is used. However, it's impossible to predict which voice will read a selected word. The great feature that macOS has had for years and years is now completely destroyed.

If anyone cares, please call Apple Accessibility support at demand that they squash this bug.

 

[LJ143]

There's a major issue in Big Sur 11.0, 11.0.1, and 11.1 (which is the latest at the time of this writing) in that the voice selected as System Voice is not honored when a word is selected and Voiceover is invoked (Option + Esc). For example, if the System Voice selected is one of the French voices, but the primary language in System Preferences > Language & Region > General is English (US), the words spelled the same in English and French are pronounced with an English US voice even though it should be pronounced with the currently selected French system voice.

For example, the word "journal" is now always pronounced by a male English US voice even though I have a female French voice (Chantal) selected in System Preferences > Accessibility > Spoken Content (formerly "Speech"). I've opened a case with the Apple Accessibility department, they sent it to engineering, and it came back as "expected behavior". This is totally bonkers. It worked properly in Catalina and before, and it's a complete mess in Big Sur.

I use the Voiceover/Speech (now called "Spoken Content") feature for learning pronunciation when studying foreign languages. It was a great tool before, and now it's totally messed up. In the example above, I uninstalled all English system voices, but a male English US voice still reads the word "journal" in English even though it should be read by system voice Chantal in French. It's now anyone's guess which words are pronounced with the selected System Voice and which are with a hidden default language system voice. Some obviously non-English-looking words are still pronounced by the male English US voice and are obviously completely butchered. If the word is clearly French, then System Voice Chantal is used. However, it's impossible to predict which voice will read a selected word. The great feature that macOS has had for years and years is now completely destroyed.

If anyone cares, please call Apple Accessibility support at demand that they squash this bug.

I’m having issues with spoken content (English only) in the new M1 MBA 2020. It sounds crackly and static like at times as if playing on the radio. Thanks for pointing out it used to be called voiceover/speech. It’s been many years since I’ve bought a new mac and sadly many phone Apple support techs know little if anything about the M1’s. Sometimes I feel like we are speaking two different languages. Pun not intended. Best.

 

[SaguaroSeven]

On the one hand, I'm really happy with my Apple devices and getting ready to buy an M1 Mac. On the other hand, it's amazing how something as basic as the System Voice (in whatever form you use it) can be broken since early in Catalina. Speaking text is a feature my Mac LC could do in 1992. <sigh>

I'm thankful not to require accessibly features, but this must be a major problem for those who do.

 

[Takuro]

I'm thankful not to require accessibly features, but this must be a major problem for those who do.

Apple pioneered the very idea of accessibility features on computers. But yeah, sometimes I feel a lot of this stuff is "best effort" and despite them putting it in the lime lite to spin up positive PR, they obviously do not put as much effort to fix bugs in accessibility features as they do for core features.

 

[SaguaroSeven]

Apple pioneered the very idea of accessibility features on computers. But yeah, sometimes I feel a lot of this stuff is "best effort" and despite them putting it in the lime lite to spin up positive PR, they obviously do not put as much effort to fix bugs in accessibility features as they do for core features.

On a number of podcasts I've heard it said that Apple, despite it's size, only has so much attention, so some things just don't get worked on. Clearly for the past year has been consumed the the transition to Apple Silicon (which largely seems to be a big success, yet you can read about many problems in early versions of Big Sur). But the problems with software quality have been building (again) since iOS 12, when they had to spend time and PR effort on quality vs features. Or so they said. I guess that's over for a while...

 

[macintologist]

I'm on Monterrey and I still can't change the language or voice when I do speak selection.

 

[JV8]

My new M1 MacBook Air running Monterey macOS 12.01 occasionally get uses my Siri female #2 voice. But I'd expect it to either work or not. WTF?! I don't get it. It's REALLY frustrating, especially since I teach people how to use  devices. If nothing else, Apple should consider the potential Public Relations hit. Time to tweet about this. That seems to get much faster responses.

 

[SaguaroSeven]

My new M1 MacBook Air running Monterey macOS 12.01 occasionally get uses my Siri female #2 voice. But I'd expect it to either work or not. WTF?! I don't get it. It's REALLY frustrating, especially since I teach people how to use  devices. If nothing else, Apple should consider the potential Public Relations hit. Time to tweet about this. That seems to get much faster responses.

That is strange. As for frustrating, how do you think people feel who have been here since Catalina...

Tweet all you want. Not much has changed in >2 years, so I wouldn't expect much. Good luck tho...

 

[JV8]

I just set up a new Mac mini that came with Catalina and I cannot get System Voices to work correctly. Have downloaded enhanced Ava, Siri Female, Siri Male, and Tom. Whenever I set the System Voice to either Siri voice, Speak Text (from right click) and System Announcements use a crappy voice (Vicki I think). Although using the Speak Text shortcut (Opt+ESC) gets the correct voice! If I set Tom as Sys Voice, it works as expected. (Announce Time also produces unexpected results.)

I've tried: reinstall macOS from Recovery, remove unwanted voices, remove Speech preferences, Safe Boot, even repair repairHomePermissions. None of these corrected this.

Can anyone help?

I found a workaround. One caveat: I'm on an M1 MacBook Air running macOS 12.1. When I tried using Siri voice 2 it failed. But this works: Go to System Preferences>Accessibility>Spoken Content>Speak Selection>System Voice (I chose Siri Voice 2). Click Speak Selection>Options then Show Controller: Always. Now it works perfectly. Every time.

 

[SaguaroSeven]

I found a workaround. One caveat: I'm on an M1 MacBook Air running macOS 12.1. When I tried using Siri voice 2 it failed. But this works: Go to System Preferences>Accessibility>Spoken Content>Speak Selection>System Voice (I chose Siri Voice 2). Click Speak Selection>Options then Show Controller: Always. Now it works perfectly. Every time.

Ha! I’m still on 11.6.2, and voices are worse. When I use any Siri voice to speak content, it starts speaking at default rate and is broken and choppy fo the first 30 sec or so.

When I can’t get a voice to work/be selected, you can unselect the check mark and it should remover the high quality download. Then close the system preferences, then reopen and try selecting/downloading again.

 

[JV8]

I found a workaround. One caveat: I'm on an M1 MacBook Air running macOS 12.1. When I tried using Siri voice 2 it failed. But this works: Go to System Preferences>Accessibility>Spoken Content>Speak Selection>System Voice (I chose Siri Voice 2). Click Speak Selection>Options then Show Controller: Always. Now it works perfectly. Every time.

Update on that workaround. After setting that, it completely tanked my Sound Preferences. It became impossible to change input and output. I could click them, but the audio in/out was only available in my LG 4K display. No bluetooth either. As soon as I turned off Spoken Content, my audio worked again but I was back to the ****** old artificial Mac voice.

 

[aviko]

Apparently it's a lost cause huh?