Integrate Text-to-Speech with LLMs

[IgorSwat]

Problem description

Current implementation of Text-to-Speech module allows a one-side streaming - we give it a full text, and it calculates the resulting audio segment after segment. We can enhance this streaming ability to also support an incrementing input text - by waiting until the next sentence appears.
This would significantly enhance the experience when using Text-to-Speech with LLM's, since LLM output is incrementally updated.

What should be done

1. Add a separate API method or adjust the current streaming method of TTS module to support incrementally updated input text.

Benefits to React Native Executorch

• Enhanced integration between Text-to-Speech and LLM modules.
• Significant marketing potential with new demos