Sage Audio

Projects, experiments, and examples involving audio processing, analysis, visualization, and classification using machine learning.

AudioBasics

Simple notebooks and scripts that demonstrate various basic tasks in audio processing and visualization. 01/2021 - 02/2021.

• Note_Generator.ipynb - A simple notebook which uses interactive widgets to build and visualize a sound wave. 01/2021
• Webcam_Recorder.ipynb - Records a short audio clip using a webcam microphone, writing it to a wave file and visualizing a waveform and a spectogram. 02/2021
• Audio_Visualizer.ipynb - Pulls a random song from the FMA dataset, plays it, and lists basic metadata info of it. 06/2021
• usbaudio.py - Script built for Raspberry Pi that records audio from a usb microphone and optionally plots various visualizations. 02/2021

MusicGenre

Explorations in machine learning and music information retrieval focused on the task of genre classification. Based on work by Michaël Defferrard, et al., and Nagesh Singh Chauhan. 03/2021 - 05/2021.

• Genres.ipynb - Builds a genre classification model and makes genre predictions on user-submitted audio data. 04/2021
• Features.ipynb - Extracts audio features from a subset of the FMA dataset. 04/2021

Summer2021

Educational notebooks written for summer 2021 camp on basic ML/AI concepts and their applications in scientific research. Notebooks are written to be as clean and accessible as possible for high schools students who may not have any programming experience, heavily utilizing the ipywidgets library. Various notebooks written during spring 2021 for AudioBasics and MusicGenre have been adapted here. 06/2021 - 09/2021.

• Spotify.ipynb - Searches a database of Spotify songs and plots various high level features tracked by Spotify for each song. In cleaning stage. 06/2021
• Audio_Visualizer.ipynb - Plots visualizations of input audio files such as waveform plot, spectrum diagram, and spectrogram. In cleaning stage. 06/2021
• Data_Explorer.ipynb - Pulls a random song from the FMA dataset, plays it, and lists basic metadata info of it. In cleaning stage. 06/2021
• Datasets.ipynb - Builds subsets of FMA dataset for students to train models on. In prototyping stage. 06/2021

EdgeAudio

Notebooks written to explore audio problems in edge computing environments. 08/2021 - Current project.

• EventRecorder.ipynb - Edge audio recorder using PyWaggle that detects sound. Prototype. 12/2021

Students: Emily Brown
Mentors: Michael Papka, Nicola Ferrier (Argonne)

This material is based upon work supported by the National Science Foundation under Grant No. OAC 1935984.