About

My name is John Vinyard, and my goal is to develop a sparse, interpretable, and easy-to-manipulate representation of audio. Ideally, this is also a perceptually lossless and compressive representation, i.e. an audio codec.

You can read more and find relevant code in my matching-pursuit repository on GitHub.

Imagine a codec that you can view and edit in a way that “fits” your perception of audio in the world; a collection of sources or events in a shared space.

Or, when your internet connection cuts out briefly, your meeting partner’s voice can decay naturally in the space they occupy, rather than robotically repeating the last vowel they uttered until the connection returns!