hf-speech-to-speech-pipeline

What problem does it solve?

This Skill provides a structured approach to building a real-time speech-to-speech pipeline by coordinating VAD, STT, LLM, and TTS components using a queue-based, multi-handler architecture.

Core Features & Use Cases

• BaseHandler contract: enforce a consistent setup, process, and cleanup pattern across all pipeline stages.
• Queue wiring & lazy import: connect stages via queues with on-demand module loading to minimize startup cost.
• Session lifecycle & side-channel patterns: manage per-session state and optional side data without cross-session leakage.
• Progressive streaming: support real-time audio processing with progressive chunks for low latency.

Quick Start

Configure a single-session voice pipeline by wiring VAD, STT, LLM, and TTS handlers using the provided BaseHandler pattern.