Amplitude Compression

In acoustic signal processing, it is rare that the phase of a detected sinusoid is meaningful. Therefore, we can compress the $2k$-dimensional amplitude vector ${\bf b}$ into a $k$-dimensional vector of squared magnitudes ${\bf d}=[d_1,d_2 \ldots d_k],$ where

$$d_i = \alpha_i^2 + \beta_i^2.$$

The module function software/module_c2r.m implements this compression.