Become a sponsor to John Vinyard
My interest is in building open-source tools and machine learning models that decompose musical audio into sparse, interpretable and easy-to-manipulate representations. These models open a huge array of possibilities for musicians and non-musicians alike, including:
- the ability to transition freely to and from time-domain and symbolic, MIDI-like representations
- the ability to "edit" musical audio in the time domain
- a more efficient and meaningful representation (as opposed to frame-based approaches) for LLMs and other sequence learning models
You can see a summary of my recent work here!
Featured work
-
JohnVinyard/matching-pursuit
This repository contains research and experiments aimed at producing sparse, interpretable representations of audio.
HTML 1 -
JohnVinyard/zounds
Zounds is a dataflow library for building directed acyclic graphs that transform audio. It uses the featureflow library to define the processing pipelines.
Python 23 -
JohnVinyard/annotate-api
Cochlea is a REST API that allows users to annotate audio files on the internet. Segments or time intervals of audio can be annotated with text tags or arbitrary structured data hosted on another s…
Python 1 -
JohnVinyard/featureflow
A python library for building feature extraction pipelines in a declarative way, and storing the results for later use.
Python 6