Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Scribd
Upload
29 views

Lecture LPC

This document discusses speech codecs and linear predictive coding (LPC). LPC models digital speech as the output of a digital filter excited by either an impulse train or white noise. It can represent 160 speech samples with just 13 filter coefficients. By analyzing the excitation signal, LPC can determine the voiced/unvoiced status, gain, and pitch period of speech in each 20ms frame with low bandwidth.

Uploaded by

Sajid Ali
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PDF, TXT or read online on Scribd
29 views

Lecture LPC

This document discusses speech codecs and linear predictive coding (LPC). LPC models digital speech as the output of a digital filter excited by either an impulse train or white noise. It can represent 160 speech samples with just 13 filter coefficients. By analyzing the excitation signal, LPC can determine the voiced/unvoiced status, gain, and pitch period of speech in each 20ms frame with low bandwidth.

Uploaded by

Sajid Ali
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 7

Speech CODEC

 Let a speech signal is sampled at a rate of 8000 samples/sec.


 Typically, each sample is represented by 8 bits.
 This corresponds to an uncompressed rate of 64 kbps.
 So at least 64kbps connection is required to establish a
communication link in real time.
 Using a simplest compression technique it is possible to reduce
the rate to 8 kbps with almost no perceptible loss in quality.
 This way we can adjust 8 users at same link of 64kbps.
 Further compression is possible at a cost of lower quality.
 All of the current low-rate speech coders are based on the
principle of linear predictive coding (LPC)
 When you speak: Air is pushed from your lung through your
vocal tract and speech is produced.
 For certain voiced sound, your vocal cords vibrate. The rate at
which the vocal cords vibrate determines the pitch of your voice.
 For certain fricatives and plosive (or unvoiced) sound, your vocal cords
do not vibrate but remain constantly opened.
 The shape of the vocal tract changes relatively slowly (on the
scale of 10 msec to 100 msec).
 The amount of air coming from your lung determines the
loudness of your voice.
 The given model is often called the LPC Model.

 The model says that the digital speech signal is the output of a digital
filter (called the LPC filter) whose input is either a train of impulses or
a white noise sequence.
 Vocal Tract ======H(z)(LPC Filter)
 Air ====== u(n)(Innovations)
 Vocal Cord Vibration ====== V(voiced)
 Vocal Cord Vibration Period ====== T(pitch period)
 Fricatives and Plosives ====== UV(unvoiced)
 Air Volume ====== G(gain)
 The LPC filter is given by:

 which is equivalent to saying that the input-output relationship of the filter is


given by the linear difference equation:

 At a sampling rate of 8000 samples/sec, 20 msec is equivalent to 160 samples.


The digital speech signal is divided into frames of size 20 msec. There are 50
frames/second.
 The model says that

 is equivalent to

 Thus the 160 values of is compactly represented by the 13 values of .


 There's almost no perceptual difference in if:
 For Voiced Sounds (V): the impulse train is used
 For Unvoiced Sounds (UV):} a different white noise sequence is used.
 LPC Synthesis: Given A, Generate S
 LPC Analysis: Given S, find the best A(this is described in the next section).
 Consider one frame of speech signal

 Compute autocorrelation

 Generate

 The above matrix equation could be solved using The Levinson-


Durbin recursion.
 To get the other three parameters U/UV,G,T we solve for the
innovation:

 Then calculate the autocorrelation of u(n).

 Then make a decision based on the autocorrelation:

You might also like