Fix: Interrupt Plugin Performance Issues

⚠ Whisper model too large for hardware

→ The large model (3GB) uses ~4GB RAM and significant CPU. Switch to small (466MB) on lower-resource machines. Set whisperModel: "small" in config.

⚠ TTS voice causing delay

→ Edge-tts downloads voice data on first use per voice. Subsequent uses are faster. Stick with one voice for consistent speed.

⚠ sox monitor using CPU

→ The mic monitor samples at 10Hz — negligible CPU. If sox uses high CPU, check for audio device conflicts.

⚠ Too many plugins

→ Each plugin adds overhead. Remove unused plugins from opencode.json to reduce load.

⚠ Memory pressure

→ Whisper transcription + LLM inference simultaneously can strain 8GB RAM systems. Consider upgrading RAM or using API-based models.