Unfortunately, I can't figure out how to pass the utterance ID to these signals because I don't see that the GDNative Rust bindings support objects of one type creating objects of another (I.e. nodes creating references.) If I'm wrong, PRs or suggestions welcome.
This resolves the previously-noticed bindgen issue, where gdnative and speech-dispatcher require conflicting versions.
The early iOS build failures appear due to attempting to link AppKit on iOS, where it doesn't exist. This seemed to work on Rust 1.45, but now fails on 1.46.
Hopefully, this plus tts-rs changes get this building under iOS again.
I'd initially done this because 0.9 uses a newer bindgen and, for some reason, I can't build this with a local tts-rs under Windows. This, despite bindgen only being a Linux dependency. Seems to work fine with the dependency from crates.io, though.
0.9 also appears to be broken on MacOS/iOS targets.
Apologies for the impending force-push to update the release tag.
This reverts commit 6bbddcc6dd.
* Replace ints with floats when non-zero. Not sure if this is explicitly necessary, but at least it makes the type more obvious.
* Set initial JavaScript rate to 1.0, not 50. 50 was a remnant of the rate being a percentage, rather than a value between platform-specific minimums and maximums.
* Add ability to detect when speaking. Doesn't yet work under Linux or WinRT.
* Add signal to indicate when speaking is done.
* Refine screen reader detection under Android.