Fully Customizable Voice AI with multi-modal open source LLMs and esp32 (clone your own voice too with simple tools)

A step‑by‑step guide to building a fully local voice AI using fully open source EchoKit, swapping ASR/TTS models, integrating open‑source LLMs, and deploying on ESP‑32. Walkthrough of how I run the open source EchoKit (https://github.com/second-state/echokit_server) voice AI stack to connect any LLM with speech-to-text, text-to-speech, and custom prompts—all running locally or in the cloud. I’ll start with a minimal working setup, then show how to swap in different ASR/TTS models (Whisper, VITS), integrate LLMs (GPT5, Kimi K2, DeepSeek), and add domain-specific knowledge bases. We’ll explore the full data flow, from audio capture to streamed responses, plus edge deployment with the open source runtime WasmEdge for low-latency inference.