Integrated Piper.tts (was ffmpeg audio until this point, piper was setup at a basic level but needed to actually be integrated)

[hamizfaraz]

Switched the assistant’s speech output to use Piper TTS in the live BrainCharge pipeline, replacing the older basic audio/TTS path that was ffmpeg while keeping the voice output tied into the existing response flow.