conda
environments and package dependency scripts/files are available in the environment directory.
The files provide dependency setup for the magic_packet module as well as working with the mlops or
notebooks directories.
An example of setting up a Mac device with an M1 processor:
% brew bundle --no-lock --file environment/m1/Brewfile
% conda env create -f environment/m1/environment.yml
magic_packet module development may also be supported using the .devcontainer, requirements files, or setup.py
.
The magic_packet directory contains the CLI and the magic_packet package. The data directory is a work in progress for processing the Mozilla Common Voice dataset into aligned audio files. The deploy directory contains a React client, FastAPI server, and Elastic Beanstalk configuration demonstrating the use of embedding models for custom wake word detection. The docker directory contains the Dockerfiles to support magic_packet and deploy development. The mlops directory contains notebooks and terraform code to deploy EC2 instances on AWS. The notebooks directory contains notebooks demonstrating a few models for wake word detection.