Speech Central Next-gen testing

The only new button is 'Random', previously this was managed as part of the list (at least on iOS, on the Mac it was closer to this solution).

I am not sure if the last button is titled differently in the current version. In the beta it is titled to reflect the visual appearance which I find as correct, but it could be that this may be confusing from VoiceOver perspective and that title "More..." may be more appropriate.

Thanks, this is a very good reasoning!

I have just tried the iOS Files app that has a similar menu and it appears to do the same as my current beta, but I agree that that's not the ideal experience.

The good thing is that this part is already handled by my custom component, so "teaching" it to do this trick won't be too hard.

It should be in the next beta version which is likely to be published in the next week.

There is the new version today that addresses all the feedback, including few crashes and improved accessibility (the issues mentioned here and few more that I have noted - now VoiceOver indicates the 'on' state of few buttons).

Currently there is not beta testing on the Mac. To make a spoiler, it is submitted to Apple but it is in the approval process and that process if frozen for the holidays...

I will publish in the Mac forum when it is ready

If Apple warns you that it may happen, I can't say that it is impossible as they have more data on that than me.

As far as I can tell, and I think that people here can confirm, this isn't something that one should expect to happen and I would guess that it requires some accident to happen.

Certainly you can avoid the risk by having either iCloud backup of the device or the backup of the device on your Mac and restoring it if something goes wrong.

Speech Central isn't the app that provides voices. You can use either Apple's voices (that you can use with VoiceOver) or Microsoft's voices (that you can't use with VoiceOver).

At the same time that you wrote the comment regarding the Bookshare I made a post that explains the current state: https://www.applevis.com/forum/ios-ipados/bookshare-speech-central

This button is generated by the Apple's framework.

The app is only declaring toolbar items and those that are declared as secondary can be shown in the overflow menu that is generated by Apple when buttons can't fit the screen. It could be that they made the wrong title to it, you can report that to them.

I might consider in the future to have a custom solution as I don't like that there is no divider in this menu and there are enough buttons that divider might be welcome, which would make it possible to create my own button that would have a proper title.

Thanks for confirming that improvement has been made.

The number after the voices indicates the number of selected voices. In general if you select just one voice it will be shown by its name. There is one case when it is still a number, I don't remember why at the moment as I made that feature many months ago (but I think that it is if you have multiselect on).

Regarding the setup of voices, there is information in the app's help: https://speechcentral.net/ios-help/#azure

Unfortunately I can't provide the link at the feature page as Apple would consider that I am trying to boost potential sales of a digital product and would require that I sell that from the app so that Apple can have its cut from that...

This will be fixed for the folder, the icon was made to be different on whether the button is in Books & Articles (where it is in the Add menu so Folder is appropriate) or in the Archive, but the title wasn't.

As this is the post about the beta testing there was no link to the app description thus far and you can find all information there: https://apps.apple.com/us/app/speech-central-ai-voice-reader/id1127349155

This was most notable problem on the beta thus far - that on the iOS 16 only one tab was available and thus the app couldn’t be used.

If anyone here was affected, it is fixed now.

I am not sure if you have followed the link above. If you haven’t then it is likely to resolve your problem.

I assume that is the case and then it can’t be provided in more details in the app as it would cause Apple to want the cut from your eventual payment for the service which would mean that I would have to resell the service and that would be quite bad for users, most notably no free tier as I get I would get only one free tier for all users.

If you’ve read the link please specify in which step you had the problem.

You have two links in the help article. The other one brings you exactly to the form to create the resource, here is the link: https://portal.azure.com/#create/Microsoft.CognitiveServicesSpeechServices

Otherwise I do agree that it is fairly complicated if you try to create the resource from their portal directly.

If the play button is disabled it is one of regular states with Microsoft voices. In the new version I have also add the loading indicator inside that button but it could be that I've missed to make it accessible.

To conclude, Microsoft voices are loading speech from the internet and the loading state is a regular thing as such. If for some reason it is not possible and you have disabled "Fallback to Apple voices" it will keep trying to load and it may last indefinitely in that specific combination.

Technically it could be possible that disable state comes for some other reason, but I cannot make any better guess based on this description.

The help page is made for the regular version. While more than 90% can be applied, some things can't. The .eml files are not supported at the moment, I can't say whether that will be available in the future as this is again one area of extremely low usage (less than 1% of 1% of Speech Central usage and I would guess even those mostly out of curiosity) and it is hard to justify the effort.

Thanks for your honest and constructive feedback.

Regarding the email specifically, it was "advertised" as it was available in the "Add" menu among other item types and if someone clicked it would lead him to the description. But even this hasn't resulted in some big uptake...

If you talk about the beta testing, the requirement since the beginning of it was and still is that macOS Ventura is the minimum version.

As such every M1 Mac can be capable to run the app if its operating system was updated, but if it wasn't then it is possible that it can't.

As TestFlight is Apple’s app, I don’t know how it works.

But basically it should check the metadata of the software and compare to your system.

I have checked and metadata is still the same, Ventura is still minimal version and both architectures are supported (and even if they were not M1 can still emulate old architecture). It doesn’t require any particular device or hardware characteristics.

So it is likely that something has changed on your system other than the update of Speech Central.

I have tried it now, and it works, it says "Works on this Mac" for my M2 Pro with Sonoma.

One thing to note is that if you click on the link of the iOS app beta (in this thread) you will receive the information that it is "Incompatible on this Mac", which is correct. Maybe you did that?

Got that bug for Microsoft voices, you were right. The idea was to use the iPhone built-in sound processing to simulate different pitch levels, but it turns out that this produces artifacts so this will be removed (this isn't possible in the current official version either). However I do have the plan to have a switch that will allow the pitch to work, just it will require sending of meta data which basically means that each request would consume twice more characters than before.

While the app is overall fairly stable now, some issues are still being encountered. Those issues require either some unlikely racing situation (when the match of two things can cause the crash, but such match is unlikely to happen) or some specific not too frequent configuration of the reading content. As such it is still essential to keep testing.

I have switched to less advanced but more reliable technologies to play Microsoft voices, so in the new version you should notice a much better quality.

Some trade-offs are that now you can't set the pitch for them.

Key is stored in the device dedicated secure storage (where passwords should be stored) called KeyChain.

I agree that Apple should delete app's entries from KeyChain as part of uninstall process as it is technically possible and more than reasonable thing to do, but it doesn't.