Bayesian Auditory Scene Analysis

 Back to main
Click on gammatonegrams in order to hear the corresponding audio!

Natural sounds

 Unlike most other models of human auditory scene analysis, ours is a generative model on the acoustic waveform. This means that we can apply our model to arbitrary sounds, including recorded audio. Here we show preliminary model inferences on simple recorded audio.

Two birds near waves

 Nature sound: two birds near waves Nature sound: two birds near waves, mixture Nature sound: two birds near waves, source 1 Nature sound: two birds near waves, source 2 Nature sound: two birds near waves, source 3 Nature sound: two birds near waves, source 4

Xylophone with closed hi-hat

 Music sound: hi-hat and xylophone Percussion sound: short hi-hat hits with xylophone notes, mixture Percussion sound: short hi-hat hits with xylophone notes, source 1 Percussion sound: short hi-hat hits with xylophone notes, source 2

Two birds at a stream

 Nature sound: rhythmic bird and owl at stream Nature sound: two birds by a stream, mixture Nature sound: two birds by a stream, source 1 Nature sound: two birds by a stream, source 2 Nature sound: two birds by a stream, source 3 Nature sound: two birds by a stream, source 4

Vibraphone with open hi-hat

 Music sound: long hi-hat hits with single vibraphone note Percussion sound: long hi-hat hits with single vibraphone note, mixture Percussion sound: long hi-hat hits with single vibraphone note, source 1 Percussion sound: long hi-hat hits with single vibraphone note, source 2

Dog panting and bird chirping

 Nature sound: dog panting and bird chirping Nature sound: dog panting and bird chirping, mixture Nature sound: dog panting and bird chirping, source 1 Nature sound: dog panting and bird chirping, source 2 Nature sound: dog panting and bird chirping, source 3
Back to top
Back to main