An awesome voice assistant with a keyword-detection-neural-network running on an Arduino and a configurable AI-Pipeline!
What you need:
ei-clap-arduino-1.0.8.zipincluded as ZIP-Library in Arduino IDE- A working Arduino Nano 33 BLE Sense Rev2
- A USB-to-Micro-USB cable
- A working computer with USB ports
demo.mov
Transition: Starts automatically at first and restarts after it receives 2 or '2' in serial
LED: 
Serial: Empty, when SHOW_RESULTS is false
Transition: When keyword was detected starts automatically
LED: 
Serial: All the info
Transition: Starts when it receives 1 or '1' in serial
LED: 
Serial: Empty
Tested on Python v3.12.3 Download-Link
pip install pyserial numpy pydub openai-whisper pyttsx3 python-dotenv openai
pip install pyserial numpy pydub whisper-mps pyttsx3 python-dotenv openai
Linux: sudo apt-get update && sudo apt-get install ffmpeg
Windows: winget install "FFmpeg (Essentials Build)"
MacOS: brew install ffmpeg
Add that key into a .env-file within CLAP like OPENAI_API_KEY=<secret-key-we-dont-share-on-github>
Simply run pipeline.py in CLAP/pipeline
useOpenAI in the first lines of pipeline.py can be set to True, if you want to have ChatGPT as the LLM, it is using Ollama in case it is set to False.