Subject: Speech Processing

Topic: LPC and Parametric Speech Coding

Difficulty: Low

(i) LPC is used in the speech signal processing as a method that provides a signal source modelling.

(ii) It is widely used by researchers and mostly the linguists as a tool for extracting the formant efficiently.

(iii) LPC techniques is most suitable and dexterous for modelling signals that show periodicity. In speech it is the vowels that exhibit such properties hence it is a part for modelling vowels excepts some vowels that have morality.

(iv) Basics of LPC lies in the source and filter models of the speech signals.

(v) A model with a certain number of poles or formants produced as a result of observation of input-output sequence.

(vi) The coefficient that result describes the behaviour of a system which is still unknown.

(vii) These set of coefficients form the all pole model.

(viii) These all poles model can be viewed as a lesser complex version of the acoustics model of speech production mechanism.

(ix) The value of the signal are then estimated as a linear function of spectral samples.

(x) LPC model is a compressed from of spectral envelope representation is possible.

(xi) The LP model provides a powerful, accurate and a dependable technique for evaluating the linear system parameters that include vocal tract, glottal pulses and radiation characters of voice speech.

(xii) LP technique finds extensive demands in speech processing information theory.