Reference: sndGetSamples

sndGetSamples(snd) returns a n×c real matrix with the n sampes of the c-channel sound snd. A given column thus represents an individual channel, and if the elements are f(t), t ∈ [1, n], then f(t) is the speaker membrane's displacement at time t. The scaling is chosen so that f(t) ∈ [-2^31, 2^31]. Consequently, when superposing sounds, it is important to make sure that f(t) of the resulting sound never exceeds 2^31 in absolute value.