Artificial Stereo Extension of Speech Based on Inter- Channel Coherence

In this paper, an artificial stereo extension method of speech is proposed to control the width of auditory image. The proposed stereo extension method employs a decorrelator to make stereo-channel signals without any additional information from mono speech signal. In other words, uncorrelated signal is first generated from mono speech signal passing though the decorrelator, and it is added to or subtracted from the mono signal to obtain the left or right channel signal, respectively. Here, the degree of inter-channel coherence is controlled to vary the width of auditory image. The performance of the proposed stereo extension method is evaluated using a MUSHRA test and compared with that of a parametric stereo method. It is shown from the test that the mean opinion score of the signals extended by the proposed stereo extension method is 92, which is around 5% higher than that of the parametric stereo method.