The Interactive Audio LabWelcome to the Interactive Audio Lab, headed by Bryan Pardo. We are in the Electrical Engineering and Computer Science Department of Northwestern University . We apply machine learning, signal processing, natural language processing, and database search techniques to make new auditory tools and interfaces.
A music search engine you can sing to
Estimate missing information in audio
Spatial Source Separation|
Amplify sounds coming from a particular direction
Separation by repetition, by repetition, by repetition, ...
Adaptive User Interfaces|
Don't learn the tool, let the tool learn you
Separate music sources in real time with the help of a musical score
Learn about the music
Make videos from music
Multi-pitch Estimation & Streaming
Track more than one pitch at once
- November 2014Mark Cartwright received the Best Technical Demo Award at ACM Multimedia 2014 for SynthAssist (related papers: , )
- September 2014We are seeking one new PhD student to work on source separation and user interfaces.
- March 2014The Midwest Music Information Retrieval Gathering (MMIRG 2014) was hosted by our lab on Saturday, June 14, 2014
Demos and Products
- Reverbalize is a natural-language reverberation tool.
- SynthAssist is a tool for programming audio synthesizers using vocal imitations.
- Mixploration rethinks audio mixing from the ground up.
- Tunebot finds an iTunes version of any song you sing to it.
- SocialEQ sets your equalizer by having you rate options.
- Toneboosters TB EZQ, is a VST version of our 2DEQ audio equilizer.
- Tunebot dataset: 10,000 sung contributions to Tunebot.
- SocialReverb dataset: crowdsourced definitions for adjectives describing reverberation. They maps between words and reverb settings.
- SocialEQ dataset: crowdsourced definitions for words to describe equalization. These map from words to the EQ settings that elicit these words
- Bach10 datasset: a versatile polyphonic music dataset for Multi-pitch Estimation and Tracking, Audio-score Alignment and Source Separation.
- Jazz Performance dataset: a database of jazz pieces performed by professional Chicago jazz pianists using lead sheets. Performances are alinged to scores.