In the previous chapter, the principle of parametric stereo coding alone was presented. In this chapter, a more precise study of this type of coding is made. The article by Breebaart [BRE 05] is an excellent introduction, but reading this article is not always particularly easy since certain important results are often presented without the luxury of details and since these results are sometimes only very good approximations for an expert, but far from clear to a beginner.
The two stereo channels are denoted as x1(n) and x2(n). To be able to link these results with those of the standard signal processing, it is useful to be able to interpret these two signals as though they were realizations of zero mean stationary ergodic random processes X1(n) and X2(n) with powers and and normalized intercovariance function:
We can note that:
the intercovariance matrix by setting .
The three inter-channel indices ...