I removed bundled binaries and declared accessible-output2 as a dependency in pyproject.toml instead. I also cloned my fork in an empty directory to test it. It should be good now.

Update: I was added to the contributors file as Garrett Johnson. This name is not my real name. I changed this to my GitHub alias, but I can rebase with my real name if this is necessary. I also updated the PR to get rid of merge conflicts.

1 and 2 are probably better split into separate PRs, but please wait for dae's feedback first. Speech synthesis will also need to be tested on different platforms and confirmed working properly with different card types. Right now it's not really helpful for cloze cards for example. Anki already has built-in support for text-to-speech that works well. We should ideally use the same implementation for consistency and better cross platform support.
The reviewer is undergoing a refactor in #4289 and non-trivial changes are unlikely to be accepted at the moment to avoid wasted effort.

I understand, and am happy to re-implement these changes once the refactor is completed. I would also be happy to submit individual PRs for this.

Regarding text-to-speech implementation: I appreciate that you highlighted this, as I was unaware of TTS features existing in Anki already. I would, however, like to point out that a lot of blind people prefer content being announced with their screen reader wherever possible. This is because we like to have the same voice read things out to us consistently.

Screen readers have the ability to announce things differently, and they give you the chance of modifying those rules at any time for the individual user. I myself have a screen reader addon that switches languages on the fly when reading content - something Anki's TTS system wouldn't be able to do effectively and is vastly important for language decks. Accessible Output 2 also sends information to braille displays, which the built-in TTS feature wouldn't be able to manage. Braille displays are important to quite a number of people, including those who have both an auditory and a visual impairment. I do not have an auditory impairment, but I do use a braille display sometimes, so being able to receive information through both speech and braille carries a certain level of importance to me.

In this case, Accessible Output 2 is the best option for outputting information to a screen reader. However, I would be willing to implement a fallback system which uses the built-in TTS, should Accessible Output 2 not be available on the users system.

I can confirm that Accessible Output 2 works on Windows, Mac, and Linux. However, I do not believe it works on other platforms, which means this implementation would probably not work on Android. I'm not sure if this code base is used for the Android version, but if it is, this is where having a fallback system would be useful. That being said, I would not want to be rid of AO2 entirely, given it provides features that Anki's TTS system cannot.

I am happy to discuss accessibility topics more, either here or on other official communication channels, if that's what you would prefer. In the meantime, I understand if you wish to close this pull request.

Screen readers have the ability to announce things differently, and they give you the chance of modifying those rules at any time for the individual user. I myself have a screen reader addon that switches languages on the fly when reading content - something Anki's TTS system wouldn't be able to do effectively and is vastly important for language decks.

That's a good point. Looks like accessible_output2 is not much code, so it might be easy to port it to the Rust backend for better integration in the future. Qt also has some support for accessibility that we can check if it works.

I am happy to discuss accessibility topics more, either here or on other official communication channels, if that's what you would prefer. In the meantime, I understand if you wish to close this pull request.

Yes, I think it's a topic that requires some discussion and feedback from target users, which are unlikely to be following GitHub updates. I recommend making a post on the forums and maybe linking to it in Anki subreddit for more visibility.