Voice assistant on Raspberry Pi

• Audio Processing: pyaudio
• Voice AI: vapi.ai

• Docker installed on your system
• Make (optional, for using Makefile commands)

1. Build the Docker image

   make build
   # or without make:
   # docker build -t assistant.py .

2. Run the application

   make run
   # or without make:
   # docker run --rm assistant.py

3. Run the application locally

   sudo apt install portaudio19-dev
   python3 -m venv .venv --system-site-packages
   source .venv/bin/activate
   pip install -r requirements.txt
   python3 main.py

4. Control the volume

   amixer get Master
   # amixer -c 0 sset Master 100%
   amixer set Master 10%+ # increase 10% volume
   amixer set Master 10%- # decrease 10% volume
   amixer set Master unmute

5. Control the volume

Create a .env file in the project root (copy from .env.example):

VAPI_API_KEY=your_vapi_api_key_here
VAPI_ASSISTANT_ID=your_vapi_assistant_id_here

Use the development mode to mount your local code into the container:

make dev

This allows you to edit code locally and run it immediately in the container without rebuilding.

assistant.py/
├── Dockerfile          # Docker configuration
├── Makefile           # Build and run commands
├── requirements.txt   # Python dependencies
├── main.py           # Main application script
├── .env              # Environment variables (create from .env.example)
└── README.md         # This file

1. Add the package to requirements.txt
2. Rebuild the Docker image: make build

1. Create your Python script in the project directory

2. Run it using:

   docker run --rm -v $(PWD):/app assistant.py python3 your_script.py

import sounddevice as sd
import numpy as np

# Record 5 seconds of audio
duration = 5  # seconds
fs = 44100    # Sample rate
recording = sd.rec(int(duration * fs), samplerate=fs, channels=2)
sd.wait()     # Wait until recording is finished

import pyaudio
import numpy as np

p = pyaudio.PyAudio()
stream = p.open(format=pyaudio.paFloat32,
                channels=1,
                rate=44100,
                output=True)
# Generate and play a tone
t = np.linspace(0, 1, 44100)
tone = np.sin(2 * np.pi * 440 * t)  # 440 Hz
stream.write(tone.astype(np.float32).tobytes())
stream.close()
p.terminate()

If you encounter audio device issues on Raspberry Pi:

1. Ensure audio devices are properly connected

2. Check device permissions

3. You might need to run Docker with additional privileges:

   docker run --rm --privileged assistant.py

• Ensure .env file exists in the project root
• Use make run-env instead of make run
• Check file permissions on .env

Feel free to submit issues and enhancement requests!