DigiNews

Tech Watch by Johan Denoyer

← Back to articles

Rebuilding my local voice assistant pipeline

Quality: 8/10 Relevance: 9/10

Summary

A detailed look at rebuilding a local voice assistant pipeline, swapping wake word detection to microWakeWord, and integrating OpenAI for speech-to-text and response generation. The article discusses backend choices (Silero VAD, onnxruntime), latency considerations, and a configurable pipeline with cloud fallbacks.

🚀 Service construit par Johan Denoyer