Sound Search by Vocal Imitation
Searching audio by imitating it
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

  • I-SED is an interactive sound event detector.
  • Audealize is a crowdsourced interface for reverb and EQ.
  • SynthAssist is a tool for programming audio synthesizers using vocal imitations.
  • Mixploration rethinks audio mixing from the ground up.
  • SocialEQ sets your equalizer by having you rate options.
  • Toneboosters TB EZQ, is a VST version of our 2DEQ audio equilizer.


  • SocialFX dataset: crowdsourced labels for reverberation, compression, and equalization. Consists of 4297 words from 1233 users. Combines SocialReverb and SocialEQ, and adds data for compression.
  • VocalSketch dataset: 10,000+ vocal imitations and identifications of a large set of diverse sounds.
  • Tunebot dataset: 10,000 sung contributions to Tunebot.
  • SocialReverb dataset: crowdsourced definitions for adjectives describing reverberation. These map 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 dataset: 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.