Speech recognition

4 boards

KiCad boards from 3 GitHub repositories tagged “speech-recognition”, the most starred first; open any of them in PCBKit.

2 layers31 × 43 mm
NEO Lite
techmiguel/neo-robot
Open-source ESP32-S3 voice assistant: offline wake word with TinyML (TensorFlow Lite Micro + ESP-NN), Whisper speech-to-text, local LLM (Ollama, Gemma, Qwen) or Groq, streaming TTS.
35 parts1MIT
4 layers120 × 94 mm
Alfred
mkol-ME/A1-4RD
Local voice assistant with a character: Whisper and a 26B LLM on a reflashed datacenter GPU, a custom voice, a 3D-printed body and a KiCad PCB
125 partsMIT
2 layers70 × 50 mm
RPI4B Dual SPH0645 RevA
notgate/sph0645f_asr
Dual-SPH0645 Raspberry Pi prototype with audio capture, FFT signal processing, PCB design, and offline LLM inference.
20 parts
4 layers120 × 94 mm
Alfred
mkol-ME/A1-4RD
Local voice assistant with a character: Whisper and a 26B LLM on a reflashed datacenter GPU, a custom voice, a 3D-printed body and a KiCad PCB
123 partsMIT

NEO Lite