Postdoctoral position (Imperial College London):

My current postdoctoral research involves in the development of algorithms that will enhance the speech from a poor quality audio recording or of a talker operating in a noisy environment.

• Funded by QinetiQ Ltd
• Co-workers: Mike Brooks and Dr. Patrick Naylor

PhD position (Imperial College London):

My PhD research focused on the development of adaptive filtering algorithms that are robust to changes in the sparseness of the impulse response, for supervised and blind system identification.

                - Thesis : pdf (3470 KB)

Supervised system identification

We proposed a class of sparseness-controlled algorithms which achieves improved convergence compared to classical NLMS and typical sparse adaptive filtering algorithms. We incorporated the sparseness measure into the traditional algorithms to achieve fast convergence that is robust to the level of sparseness encountered in the impulse response of the echo path.   

                - MPhil to PhD transfer examination report: pdf (727 KB)

We also proposed a frequency-domain adaptive algorithm for acoustic echo cancellation, which dynamically adjusts its step-size according to the sparseness variation in acoustic impulse responses that arise in a mobile environment. For this section, we finally presented an analysis on the tracking performance of IPNLMS for time-varying echo paths.

Blind system identification

We developed time-domain SOS based blind SIMO identification algorithms by exploiting the cross relation technique with proportionate step-size control.

Some useful links for MATLAB functions:

• Room Impulse Response Generator (Dr Emanuel Habets)
• Signal Generator (Dr Emanuel Habets)
• Speech Processing Toolbox (Mike Brookes)