Linear prediction

[Ida S]

I find Ellis’s discussion of linear prediction in section 7 quite vague:
“In a discrete-time implementation, a resonant filter involves a few delays applied to the output signal, then feeding back these delayed outputs (with particular scaling constants) to the input. In effect, in the absence of inputs, the output at a particular time is a linear combination of a few recent output values, and the process of fitting a resonant filter to a particular signal consists of choosing the scaling constants that do the best job of matching (or predicting) each output sample from its immediate predecessors.”
I’m not even sure what ‘input’ and ‘output’ refer to here. It would be great if we could go over the reasoning behind linear prediction during the lecture.

[Simon]

We’ll look at this in the lecture.

[Author]

Posts