The Interactive Audio Lab

Welcome 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
Audio Imputation
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
Score Alignment
Learn about the music
Music Story
Make videos from music
Multi-pitch Estimation & Streaming
Track more than one pitch at once


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.