Multimedia signal processing using AI

Audio signal recovery is a frequent problem in digital audio restoration field because of corrupted samples that must be restored. In this paper, we look at a subband multirate architecture with RBF nonlinear predictor for audio signal recovery. The subband approach allows for the reconstruction of...

Full description

Saved in:
Bibliographic Details
Main Authors: Seng, K.P., Hui, L.E., Ming, T.K.
Format: Conference Proceeding
Language:English
Subjects:
Artificial intelligence
Digital signal processing
Digital systems
Frequency domain analysis
Microelectronics
Neural networks
Signal analysis
Signal processing
Signal restoration
Streaming media
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Audio signal recovery is a frequent problem in digital audio restoration field because of corrupted samples that must be restored. In this paper, we look at a subband multirate architecture with RBF nonlinear predictor for audio signal recovery. The subband approach allows for the reconstruction of a long audio data sequence front forward-backward predicted samples. In order to improve prediction performances, RBF neural networks are used as narrow subband nonlinear forward-backward predictors. Previous neural networks approaches involved a long training process. In our case, the small networks needed for each subband are considered to the speed-up the convergence time and improve the generalization performances, the proposed signal recovery scheme works as a simple nonlinear adaptive filter in on-line mode. EKF (extended-Kalman-filter) is used to adjust the parameters of the RBF network. Simulation results show good results for the reconstruction of over 100 ms of audio signal with low audible effects in overall quality.
DOI:10.1109/APCC.2003.1274475