Overview

MicroLLM is the world’s first real-time voice assistant that runs entirely on a Raspberry Pi Pico W! By leveraging Groq’s blazing-fast LLaMA API and ElevenLabs for natural-sounding text-to-speech, it delivers contextual AI responses through a DAC via I2S audio output—all on a $6 microcontroller.

✨ Features

Hardware Requirements

Wiring

Pico Pin MAX98357A Pin
GP1 BCLK
GP0 LRCK
GP9 DIN
GND GND
3V3 VDD

Setup

  1. Flash CircuitPython - Use the provided .uf2 firmware
  2. Copy Files - Transfer code.py and lib/ folder to Pico
  3. Configure API Keys - Edit settings.toml:
CIRCUITPY_WIFI_SSID = "YourWiFiNetwork"
CIRCUITPY_WIFI_PASSWORD = "YourPassword"
ELEVENLABS_API_KEY = "your-elevenlabs-api-key"
GROQ_API_KEY = "your-groq-api-key"
  1. Power On - The Pico connects to WiFi and starts listening!

How It Works

┌─────────────┐    ┌─────────────┐    ┌─────────────┐
│   Pico W    │───▶│  Groq API   │───▶│ LLaMA 3 LLM │
│   (WiFi)    │    │  (Cloud)    │    │             │
└─────────────┘    └─────────────┘    └──────┬──────┘
                                             │
                          Text Response      │
                                             ▼
┌─────────────┐    ┌─────────────┐    ┌─────────────┐
│   Speaker   │◀───│  I2S DAC    │◀───│ ElevenLabs  │
│             │    │ (MAX98357A) │    │ TTS Stream  │
└─────────────┘    └─────────────┘    └─────────────┘

Example: Groq Query

import adafruit_requests

response = requests.post(
    "https://api.groq.com/openai/v1/chat/completions",
    json={"model": "llama3-8b-8192", "messages": messages},
    headers={"Authorization": f"Bearer {groq_api_key}"}
)

answer = response.json()["choices"][0]["message"]["content"]
print("Response:", answer)