@rommix0 @MutedTrampet @TheQuinbox @BTyson @samtupy @Tamasg @AlexSpectrum93 @datajake1999 I tried to vibe code a bestSpeech tts multilingual say terminal and speak window with latest chat-gpt thinking mode. but it is not working. maybe I installed the wrong python version? I can't tab to controls, but NVDA AI vision assistant shows the GUI is there. figured I would give it a try and if it works great, if not, oh well I tried.
https://drive.google.com/file/d/1SDqEXb44df1XhSjJAmK1lcD7OZefuH_4/view?usp=sharing
BeSTspeech_SayTerminalSpeakWindow.zip

Google Docs
@joshknnd1982 @MutedTrampet @TheQuinbox @BTyson @samtupy @Tamasg @AlexSpectrum93 @datajake1999 okay. I'll look over what you got in a moment, and I'll update you on what I see.
BeSTspeech_SayTerminalSpeakWindow.zip

Google Docs
@rommix0 @MutedTrampet @TheQuinbox @BTyson @samtupy @Tamasg @AlexSpectrum93 @datajake1999 Did you use the speak window with a screen reader loaded, navigating with the keyboard-only?
@joshknnd1982 @MutedTrampet @TheQuinbox @BTyson @samtupy @Tamasg @AlexSpectrum93 @datajake1999 I don't have a screen reader. I just opened the speak window alone.
@rommix0 @MutedTrampet @TheQuinbox @BTyson @samtupy @Tamasg @AlexSpectrum93 @datajake1999 I wonder if the accessibility is a limitation of the python 3.7.0 GUI? if this is the case, could you convert my speak window to c++, the language that your original speak window was written in?
@rommix0 @MutedTrampet @TheQuinbox @BTyson @samtupy @Tamasg @AlexSpectrum93 @datajake1999 gain is volume so it seems. So I set the gain to 16 and it is much louder. gain of 20 is too high, makes voice clip and pop.
@joshknnd1982 @rommix0 @MutedTrampet @TheQuinbox @BTyson @samtupy @AlexSpectrum93 @datajake1999 sounds like good experimenting here. I wouldn't have the time to refactor the existing B32 wrapper to work with the newer DLLs, this would need to be done after you get a solid speak window. Everything is open-source, anyone's welcome to pick up Sam's work from his repo and open a pull request against it if you feel the pace isn't quick enough. These newer DLLS don't just neatly swap into the wrapper because calling conventions are probably different, so putting an AI to compare what you just built in your speak window and Sam's B32_wrapper, then modernizing on existing NVDA 2025.3 is a good start. Maybe AI can rewrite the function signatures n the wrapper for you.
But nobody should feel obligated to an open source project if they don't have time or energy to refactor it, this is where the rest of the community can come together if there's interest on improving it.
@rommix0 @Tamasg @MutedTrampet @TheQuinbox @BTyson @samtupy @AlexSpectrum93 @datajake1999 ok everyone, thanks to me, we now have stand-alone execuatbles that you can run without needing a python development environment to be installed. compiled binaries now in their own folders in this zip file along with the source and everything else. you may get an error when closing the speak window, hit enter it goes away, does not seem to affect its operation though. https://drive.google.com/file/d/1SDqEXb44df1XhSjJAmK1lcD7OZefuH_4/view?usp=sharing
BeSTspeech_SayTerminalSpeakWindow.zip

Google Docs