StreamPlayer use TextToAudioStream queue instead of engine's one

[phamson02]

Illustrate a change that can resolve issue #230

[KoljaB]

Thanks for this promising PR! I can’t merge it as-is, though, because of the changes to the synthesize method’s signature:

def synthesize(self, text: str, audio_queue: queue.Queue) -> bool:

This would require changes to the base class:
https://github.com/KoljaB/RealtimeTTS/blob/master/RealtimeTTS/engines/base_engine.py#L56

Plus, every other engine would need its synthesize method updated too, since all engines need the same signature. Merging this now would break support for all engines except Coqui.

Also, could you add a test Python file in the tests folder to highlight the benefits of this PR? That would help me validate the changes and see them in action.

Thanks again for your effort, much appreciated!

[phamson02]

Thanks for your nice words. I know that it would break things doing it that, but wanted to open this PR to illustrate a kinda hacky way that resolve my problem.

Will come up with a small example for you to test this out later.

[phamson02]

Hey @KoljaB, sorry for the delay on the promised test for this PR. I didn't realise that I only need to make some changes to your FastAPI example. Pushed it here.

[KoljaB]

Thank you for the test. Might take me a while to review, maybe one or two weeks.