Cochleotopic/AMtopic (CAM) and Cochleotopic/Spectrotopic (CSM) map based sound sourcce separation using relaxatio oscillatory neurons

Author

Pichevar, Ramin ; Rouat, Jean

Author_Institution

Dept. of Comput. Eng., Sherbrooke Univ., Que., Canada

fYear

2003

fDate

17-19 Sept. 2003

Firstpage

657

Lastpage

666

Abstract

We use a two-layered unsupervised bio-inspired neural network to segregate sound sources, e.g. double-vowels or vowels intruded by nonstationary noise sources. The network consists of spiking neurons. The spiking neurons in both layers are modeled by relaxation oscillators. The first layer of the network is locally connected, while the second layer is a fully connected network. We show that in order to correctly segregate sound sources, we should either use Cochleotopic/AMtopic map (CAM) or Cochleotopic/Spectrotopic map (CSM) depending on the nature of the intruding sound source.

Keywords

audio signal processing; auditory evoked potentials; neural nets; physiological models; relaxation oscillators; source separation; bioinspired neural network; cochleotopic-amtopic map; cochleotopic-spectrotopic map; nonstationary noise sources; relaxation oscillatory neurons; sound source intrusion; sound source separation; spiking neurons; Acoustic noise; Auditory system; CADCAM; Computer aided manufacturing; Hair; Image segmentation; Neural networks; Neurons; Oscillators; Source separation;

fLanguage

English

Publisher

ieee

Conference_Titel

Neural Networks for Signal Processing, 2003. NNSP'03. 2003 IEEE 13th Workshop on

ISSN

1089-3555

Print_ISBN

0-7803-8177-7

Type

conf

DOI

10.1109/NNSP.2003.1318065

Filename

1318065