Most recent post:
13 May 2021
Linear Predictive Coding (LPC) is a method for estimating the coefficients of a Source-Filter model (post) from a given data.
The input consists of a time-series representing amplitudes of speech collected at fixed intervals over a period of time.
The output is a matrix of coefficients corresponding to the source and filter model and is much more compact, so this method can be used for compressing audio.
In this post we’ll study the encoding of a audio signal using LPC, which can achieve 15x compression. We’ll then decode it into a very noisy but intelligible speech.