Make WASAPI togglable for SAPI 5 - and off by default

[seanbudd]

Is your feature request related to a problem? Please describe.

There are several bugs with SAPI 5 voices and WASAPI.
Users should be able to toggle off SAPI 5 for WASAPI

Describe the solution you'd like

Like SAPI 4, there should be a toggle for disabling WASAPI output for SAPI 5.
This should ideally be in speech settings not advanced settings, with a user friendly name like "legacy audio output vs modern audio output"

Describe alternatives you've considered

None

Additional context

• SAPI5 voices cause NVDA to freeze in versions 2025.1-> #18298
• SAPI 5 Eloquence does not speak the first character in German #18003
• Sapi 5 performance issues in nvda 2025 beta #17967
• nvda switches to windows one core voices after restart of program and or reboot. #18301

[seanbudd]

@gexgd0419 - you mentioned here #17967 (comment) you might do this, do you still intend to?

[seanbudd]

Suggestion from @Qchristensen - make this and #18310 options specific to the synthesizer config rather than in speech

[gexgd0419]

Does that mean that it should be added as SomeSetting like RateBoostSetting, then added in the supportedSettings list, so that it can show up when the SAPI 5 synthesizer is selected? I'm not sure whether it is as general as something like RateBoostSetting, though.

I never thought that the WASAPI implementation would bring so many problems, and it's very difficult for me to reproduce them. I should admit that I'm sighted, not a screen reader user, and maybe not an expert in programming. As neither SAPI4 nor SAPI5 is open-sourced, the voices for them are often commercial products, and all kinds of other things such as add-ons and external software, it is really not an easy task for me to identify the cause of those bugs.

As for what should be the default value, I am not sure. If it should be defaulted to "off", do we still need to make the option more easy to find, or just leave the options in the advanced settings as an "experimental" feature? Should we explain to the user why they may want to enable this, to encourage more users to use it?

[zstanecic]

Hi to all,
let me add my few pennies to this discussion:
RHVoice works correctly with the wasapi implementation, as well as sapi5 version, and the the synthesizer packaged as the NVDA addon. There's no crash, and it boosts productivity. Even in the enviromnments where few milliseconds are crucial.

[cary-rowen]

Can the default value be enabled?

[seanbudd]

@gexgd0419 yes exactly like a RateBoostSetting

[seanbudd]

As for what should be the default value, I am not sure. If it should be defaulted to "off", do we still need to make the option more easy to find, or just leave the options in the advanced settings as an "experimental" feature? Should we explain to the user why they may want to enable this, to encourage more users to use it?

Using WASAPI for SAPI 5 still is a big benefit to many users. Being able to turn it off for specific synthesizers allows users to get the best of both worlds.

[seanbudd]

It would be even better if it could be toggled on a voice level. That would mean we could keep it on by default