@jaybird110127 When the 1-minute trial period expired, it sounded like a rather low quality telephone line, with static and a pretty authentic phone filter thing going on. I've never heard any TTS trial do that before. This synth is very impressive. I hope more people purchase these voices.

I several minutes ago purchased the two new voices for the Blastbay TTS engine, and I'm very impressed with it. It is a hybrid engine (combination of neural and formant), has two American English voices, a male named Richard, and a female named Libby. Libby is my favorite. This synth is both 32-bit and 64-bit, is SAPI 5 compliant, and has a sample rate of 48 kHz. It's very responsive with NVDA, and it can go very fast, even without using rate boost (I never use high speech rates, I just keep it at default rate values for all of my speech synthesizers I have, for consistency reasons). Each voice costs $19.95, and it uses the classic email and serial number authorization method. Note: when you try entering the serial number during the installation, the is a password field, so you will hear ***... in NVDA.

I was in a conversation with @joshknnd1982, and he can confirm that, Randy Carlstrom is in fact the voice provider behind DoubleTalk! The reason why I have proof of this is because he actually spoke to him over the phone.

@cachondo @datajake1999 @rommix0 Monologue 97 is actually based on a formant-based synth called Primovox, which is also SAPI 4 compliant. There's also Monologue 95, which is based on a later version of a diphone based synth called ProVoice, which has multiple male voices and a female voice. It also supports the other typical TTS first languages (Spanish, German, French, and Italian). However, if you want the actual version that essentially uses Randy Carlstrom's voice (he's the voice provider behind the diphone based DoubleTalk synth), you have to get Monologue 16, which has that earlier version of ProVoice on it. There is a version on Jake's site that has the 22 kHz version of that voice.

#audiomo day 12: I've been on a bit of a suno fixation, and I decided to turn a wesley Willis song into Bluegrass. So, enjoy the chicken cow, bluegrass style!

Also yesterday, I contacted @samtupy via email requesting he make the CyberTalk NVDA addon (giving him the core files in the process) and he said he'd look into it. I Haven't received any further word on the project, other than that he has other things to work on, so I won't interfere with his other ambitions, nor will I get my hopes up/set high expectations for such an addon. Moral of the story, always set lower expectations so that if either a project doesn't come to fruition or if it does come out but it doesn't meet your standards, don't complain, just try and accept it, and move on to other things you enjoy, so as to minimize negative emotions. TL;DR, if things don't work out, remain neutral and keep going.

So, as of right now, I have both my 2023 Mac Mini M2 and my iPhone SE 3 upgraded to developer beta 1 of iOS 26 and macOS 26 (Tahoe). Here's what I've observed, in the world of new TTS voices. We have some new languages for people who live in India, Gujarati and Punjabi, and for these two new languages, as well as for the existing languages Indian English, Hindi, and Marathi, we have two new neural voices, a male named Aman, and a female named Tara. Tara is my favorite of the new voices.

Here's the reason why I am wanting human NVDA addon developers to please make synth addons for FlexVoice and CyberTalk. About one or two months ago, I tried as an experiment having ChatGPT write NVDA addon code for CyberTalk, putting the DLL files through Dependency Walker and other things that are way too technical for me to explain, and it was a total failure. I got error after error after error while trying to have it write the code and other prompts I gave ChatGPT. That's the reason. And I know, based on the feedback on the samples I've posted on this account, other people like these synths as well, so I'm glad I'[m not the only one wanting them, though they are not as desperate as I am. Please boost for reach, reply with your thoughts, and send me a private mention if anyone wants the core files for CyberTalk, and I'll happily give them to you. Thanks in advance!

@masonasons I believe it was, I was using version one specifically. I really desperately am wanting these two speech synths for NVDA, but I believe that there are no human programmers interested in such synthesizers. Best speech might have to be the only one we’ve successfully ported to NVDA, and even though I love that one, I refuse to use it for a rather weird reason, that is because of the unfairness of not having others ported to NVDA as well.

OK, I'm about to give up on CyberTalk and FlexVoice. CyberTalk, according to ChatGPT and me putting the DLL files through a program called Dependency Walker (which exposes various dependencies that the DLL files rely on) says that it is possible to make an NVDA addon for it, but even after running an Auto Hotkey script to try and interface with one of the dll files directly with embedded sample text to at least try and make it speak, it said that it couldn't load the DLL file. I don't know if that's an error on ChatGPT's part, or the DLL itself. It's just, these two synths are some of the most unique ones I've ever heard, and I don't want to stay stuck switching to, and using, the same synths over and over again. At least we can give @rommix0@mastodon.social credit for finding them.

Interesting how the version number strategy is changing for Apple operating systems, to keep things more consistent, and to have the OS versions go by the last two digits of the calendar year that the majority of that version's release falls into. The only thing I wonder is if Apple is going to keep naming specifically macOS versions after California locations. Now this is coming from someone who is not an Apple hater, but doesn't use Apple devices as primary, because Windows and Android just have so many TTS engines to play with!

@rommix0 I bet you'll find it safe here, considering this very instance is largely consisting of blind/visually impaired individuals, the main administrator is blind himself. Your findings in ultra-rare TTS engines are what made me start posting on Mastodon in the first place, since it is the thing I am extremely passionate about, collecting unique retro speech synths instead of using the same ones over and over again, which is what I'm looking for. Mastodon.social sounds like the "official" Mastodon instance, and I bet that seemed a little bit too mainstream for your tastes.

Question. When I make my very initial code for others to improve on for either CyberTalk or FlexVoice NVDA addon, what method would be the best to provide the requirements (source code, required DLL files, etc.) without using Github?

@mckensie Here's a direct link to the .apk file:

https://github.com/dectalk/microdectalk/releases/download/v2.0.2/app-release.apk

@mckensie Actually, new builds of Micro DECtalk did come out, you can find them on the releases page of the Github repository. When you install it on your Android, it will be labeled "DECtalk". It sounds pretty much like the one we all know and love! So there's your Eloquence replacement!

It's kind of ironic that the UAC dialog on Windows 11 still says "Hard drive on this computer", even though I'm using an internal 256 GB SSD on my HP Pavilion TP01-0066 tower desktop. Also ironic is that even though it's Windows "11", the OS "Version" is 10.0.26100 Build 26100, at least on my system. So shouldn't it be really called "Windows 10, 11 Edition"?

@danestange No, because the installer still says on the welcome page:

WARNING: This program is protected by copyright law and international treaties

I've read countless license agreements, and they all pretty much say the same things. Even the longer FOS ones say some of the same things.

I'm practically one of the only people who actually reads license agreements, to test extremely formal text with various TTS engines (sometimes I have a certain TTS engine read its own EULA). The weirdest one comes from the original Learnout and Hauspie (L&H) RealSpeak SAPI 4 unit selection 11 kHz Engine from 2001. This is the text:

To add your own license text to this dialog, specify your license agreement file in the Dialog editor.

Navigate to the User Interface view.
Select the LicenseAgreement dialog.
Choose to edit the dialog layout.
Once in the Dialog editor, select the Memo ScrollableText control.
Set FileName to the name of your license agreement RTF file.

After you build your release, your license text will be displayed in the License Agreement dialog.

@cachondo I’m not sure, because I’m more of an “install and use“ person, I don’t do programming in any way. That question is better reserve for someone with a lot more development experience than me.

For those who are interested in taking a look at the files for these two synths for NVDA addon development/trying to get working with SAPI:
FlexVoice 3 SDK:

https://datajake.braillescreen.net/tts…/old_software_synths/Mindmaker%27s%20FlexVoice%20SDK.exe

CyberTalk core files:

https://drive.google.com/file/d/1BIBqX1Fwk2CC2QcbSHWP9uLdpenJXFSh/view?usp=sharing