Speech Coding

The voice compression coding technique is widely used in the modern digital communication systems. In this technique, a voice coder is used to set up a model to simulate the voice and noise produced by human vocal organs. The parameters to form the model will be transmitted through the TCH channels.

The voice coder is based on the residual excited linear prediction (REIP) coder. Moreover, the long term predictor (LTP) is used to enhance the compression effect. LTP can make the coding of residual data more advantageous by removing the vowels from the voice. With 20ms as the unit, the voice coder outputs 260bits after compressed coding. Therefore, the code rate is 13kbps. According to the different classes of the importance of the information, the output bits can be classified into three categories: 50 very important bits,132 important bits and 78 ordinary bits.

Comparing with the traditional PCM line on which the voice is coded directly and transmitted (64kbps), the 13kbps voice rate of the GSM system is much lower. The more advanced voice coder in the future can further reduce the rate to 6.5kbps (half-rate voice coding).