Wildlife Audio Stream Sampler is a proof of concept(PoC) for a power over ethernet (PoE) streaming microphone. While this is intended for acoustic monitoring of wildlife, the server code can be deployed on any devices with a microphone.
The PoC was developed with:
- Raspberry PI 3 Model B+
- Adafruit I2S MEMS Micophone
- PoE Splitter with MicroUSB, also available from Adafruit
There are detailed setup instruction on the Adafruit site for connecting the microphone to a Raspberry PI
Before setting up a Python virtual environment a couple of extra dependencies are needed on the PI.
sudo apt-get install libportaudio0 libportaudio2 libportaudiocpp0 portaudio19-dev
# Note: If you just setting up the server to test on a regular Linux machine libportaudio0 should not be needed.python3 -m venv audio
source audio/bin/activate
python -m pip install pip --upgrade
python -m pip install pyaudio
python server.pyTo set up a virtual environment on the client,
python3 -m venv audio
source audio/bin/activate
python -m pip install pip --upgrade
python -m pip install librosa
python -m pip install matplotlib
# Update sampling parameters in the client.py file as needed
python client.py [PI IP]The default setting will request 60 seconds of audio every 5 minutes. The audio is broken into 4 seconds samples with 2 seconds of overlap between samples.
