The GSM Codec
GSM codec are very complex and use more power.
Block diagram of GSM speech encoder:
- The encoder is comprised of four major processing blocks.as shown in figure 15.6
- The speech sequence is first pre-emphasized, ordered into segments of 20 ms duration, and then hammingwindowed.
- This is followed by short-term prediction (STP) filtering analysis where the logarithmic area ratios (L&R's) of the reflection coefficients rn (k)are computed.
- The 8 LAR parameters have different dynamic ranges and probability distribution functions, and hence all of them are not encoded with the same number of bits for transmission.
- The LAR parameters are also decoded by the LPC inverse filter so as to minimize the error e n.
- LTP analysis which involves finding the pitch period pn and gain factor g nis then carried out such that the LTP residual rn is minimized.
- To minimize r npitch extraction is done by the LTP by determining that value of delay, D, which maximizes the crosscorrelation between the current STP error sample, en and a previous error sample en- D.
- The extracted pitch and gain are transmitted and encoded at a rate of 3.6 kbps.
- The energies of these sequences are identified, and the one with the highest energy is selected to represent the LTP residual.
- The pulses in the excitation sequence are normalized to the highest amplitude, quantized, and transmitted at a rate of 9.6 kbps.
lock Diagram of GSM Decoder:
- It consists of four blocks which perform operations complementary to those of theencoder.
- The received excitation parameters are RPE decoded arid passed to the LTP synthesis filter which uses the pitch and gain parameter to synthesize the long-term signal.
- Short-term synthesis is carried out using the received reflection coefficients to recreate the original speech signal are important bits called lb bits, and the last 78 bits are called type II bits.
- Since type Ia bits are the ones which effect speech quality the most, they have error detection CRC bits added.
- Both Ia and lb. bits are convolutionally encoded for forward error correction.
- The least significant type II bits have no error correction or detection.