Unified Driver for platform.

[willwade]

Here's a suggestion. There is a need to have a AVSpeechSynth driver for MacOS as Nss is depreciated but also it would allow this tts wrapper to support ssml across the board.

And on windows we also need uwp and sapi4 driver. To make things easier I'm suggesting the default would be a unified driver for each platform. Linux would be still just espeak (unless we can nicely integrate pico / piper or Sherpa onnx*)

This is thrown together code but you can see what I'm thinking https://gist.github.com/willwade/86ef5c0655badc90c2f3fec811c06cde

• I'm not keen on any of these as they all a) are dependency heavy b) rely on a download step. And piper is the only one with ssml on its roadmap so it would then be feature matched across engine. Let's have a think.

[willwade]

121 #287 #104

[cclauss]

SSML: https://cloud.google.com/text-to-speech/docs/ssml

[willwade]

SSML: https://cloud.google.com/text-to-speech/docs/ssml

Sapi. AVSynyh & espeak all support ssml. It's on my radar to do a pr for supporting this as soon as espeak and avsynth done

[willwade]

So I’m actually wondering if windows should have a second driver using the more modern .net api.

Here’s a quick thrown together attempt https://gist.github.com/willwade/cc268ef0b0b0ec106e644eec6835b228

it Would remove problems like #263

[willwade]

That sound very promising!

Please name the file system_speech.py instead of SystemSpeech.py to align with Python naming.

• https://docs.astral.sh/ruff/rules/#pep8-naming-n

Yeah. I’m not sure I like the name of that engine. It’s technically the .NET interface. Dotnetspeech synth” I don’t know 🤷

[cclauss]

DotNetSpeech, DotNetSpeechEngine, DotNetSpeechDriver