Applying speech enhancement to audio surveillance

The authors describe ways to cancel interference where an available reference signal is not synchronized with the surveillance recording, i.e., the reference is obtained later from a phonograph record on air check recording from a broadcast source. As a specific example, they discuss their experiences processing a wiretap recording used in an actual court case. They transformed the reference signal to reflect room and transmission effects and then subtracted the resulting secondary signal from the primary intercept signal, thus enhancing speech from the desired talkers by removing interfering sounds. The intercept signal was subject to time-scale modifications due to variable phonograph and tape recorder speeds, which adversely affected the ability to cancel interference automatically. The two methods that were most successful for enhancement were least-mean-squares adaptive cancellation and spectral subtraction