feat: add silence detection callbacks for microphone control

[Jah-yee]

Good day,

Thank you for working on this project. I saw issue #85 about detecting when the model is about to talk to mute the microphone, and I wanted to help implement a solution.

This PR adds optional on_silence_start and on_speech_start callback parameters to the LMGen class that can be used to detect when the model starts producing silence vs speech, enabling external microphone muting.

Usage

gen = LMGen(
    model,
    on_silence_start=lambda: print("Silence started"),
    on_speech_start=lambda: print("Speech started"),
)

These callbacks fire at the beginning and end of audio silence periods respectively, allowing external systems to know when it's safe to mute the microphone.

Thank you for your work on this project. I hope this small fix is helpful. Please let me know if there's anything to adjust.

Warmly, RoomWithOutRoof