Make the input vector for ML classifiers using natural_vs_media.py. A five second wavefile, recorded at 16kHz on a Kinect One microphone should be used.
The two classes for classifiers are 0 (media) and 1 (natural).
Trained StandardScaler, SVC, LR, DT, QDA, GNB, and KNN are available.
Nicholas C. Georgiou, Rebecca Ramnauth, Emmanuel Adeniran, Michael Lee, Lila Selin, and Brian Scassellati. 2023. Is Someone There or Is That the TV? Detecting Social Presence Using Sound. J. Hum.-Robot Interact. 12, 4, Article 47 (December 2023), 33 pages. https://doi.org/10.1145/3611658