My master's thesis (2020) was about developing new ways to leverage internet-scale audio content for creative work.
With more music, sound effects, audio books, podcasts, etc. being created than ever before, there is an incredible wealth of audio to potentially sample, remix, and assemble. However, digging through all these recordings takes a lot of time and effort. The tools we have for editing audio are designed for working with small, organized collections - not vast libraries of unstructured content. This is partially a consequence of their origins; their design language is rooted in hardware paradigms like tape machines and mixing consoles.
My goal was to build creative tools that help people efficiently sift through large amounts of audio to find interesting bits, gather them into personalized libraries, and turn those samples into original compositions.
The thesis covers 3 main areas:
Curation - The Sifter tool lets you load long recordings and automatically break them into short segments based on changes in volume or frequency content. This makes it easy to scan through and pull out useful parts.
Collection - The Observatory allows you to explore segments extracted by the Sifter and group them into collections and subcollections, like building a palette of sounds.
Composition - The Sketchpad lets you create audio collages by arranging samples from your collections and sketching volume and filter changes over time.
Together these tools were developed to transform tasks like searching, browsing, and arranging sounds into open-ended creative exploration. They provide assistance without replacing human judgment, so you can focus on aesthetic choices rather than technical hurdles. The system was designed both for personal use (e.g. with someone's own audio library), and for collaborative projects using shared sound databases.