The ‘Global System for Mobile communications’ (GSM) is a digital mobile radio system which is extensively used throughout Europe, and also in many other parts of the world. The GSM codec provides good quality speech, although not as good as the slightly higher rate G728 codec.
The GSM full rate speech codec operates at 13 kbits/s and uses a regular pulse excited (RPE) codec. This codec has short-term and long-term predictors to estimate how the voice pattern will change over time.
The input speech is split up into 20 ms long sections, called frames, and for each frame a set of eight filter coefficients are calculated for the short-term predictor. Each frame is then divided into four 5 ms sub-frames, and for each sub-frame the encoder calculates the delay and gain needed for the codec’s long term predictor. Finally, after both short and long term prediction has been applied, the residual signal is quantized for each sub-frame. The filter coefficients for the short term predictor, the delay and gain of the long-term predictor and the coded residual signal are all transmitted. The receiver uses all these values to reconstruct the signal.