Sound Bytes: The ABCs of Sound and Digitization

This article aims at opening a new series that will explore the world of audio deep learning.
I want to dive into building a solid foundation of understanding, analyzing and exploiting audio data, its use cases within deep learning, and practical examples.

This is an overview of the pieces I want to write for the series (titles will eventually be replaced by actual links as I write this series):

1. The ABCs — What sound is and how is it recorded digitally. What everyday issues does deep learning for audio solve? What are spectrograms and the reasons behind their significance?
2. Mel Spectrograms — The why, how, and applications.
3. Audio augmentation — the essentials and impact of enhancing spectrograms features.
4. Audio classification — an end to end example of how you can exploit an architecture to classify sounds.

In this initial piece, given that many might not be acquainted with the subject, I’ll present an insight into the realm of deep learning for audio purposes. We’ll delve into the nature of audio and its digital representation. I’ll discuss the extensive influence of audio tools in our everyday experiences and delve into the structures and methods employed in their design.