31-07-2012, 02:37 PM
ADAPTIVE MULTIRATE – SPEECH CODING
AMR.ppt (Size: 199.5 KB / Downloads: 24)
SOURCE CODING
Compresses the amount of information over the given channel.
Improves the spectral efficiency of the radio interface.
Related to the network capacity.
Allows the system to introduce more powerful encoding techniques to counter the propagation & interference effects.
CHANNEL CODING
Conditions the output of the source encoder for transmission over the channel.
Transcoded speech is error protected by passing it through the channel encoder.
Related to the network quality.
Includes :
Coding for forward error correction & detection
Bit interleaving
Modulation
REGULAR PULSE EXCITATION – LONG TERM PREDICTION
Used in order to reduce the data sent b/w MS and the BTS.
When a voltage level of a particular speech sample is quantified, the Mobile Station's internal logic predicts the voltage level for the next sample.
When the next sample is quantified, the packet sent by the MS to the BTS contains only the error.
CODE EXCITED LINEAR PREDICTOR (CELP)
Coder & decoder have predetermined book of stochastic excitation signals.
For each speech signal, the transmitter searches its code book of stochastic signals and the index of one that gives the best match is transmitted.
The receiver uses this index of code book to pick the correct excitation signal for its synthesizer filter.
CELP are extremely complex, but can achieve bit rates of as low as 4.8kbps.