Understanding Vocoding and Vocoders

This page explains the basics of vocoding, covering what a vocoder is and its fundamental function. In the early days, voice transmission occurred without any compression methods applied to the voice samples, leading to a significant bandwidth requirement on the channel. This was common in first-generation analog systems.

In current generation wireless standards like 2G, 3G, and 4G, voice is transformed into digital 1s and 0s. This digital data is then modulated onto a transmitted carrier using digital modulation techniques such as ASK, PSK, and FSK.

Nowadays, BPSK, QPSK, 16QAM, and 64QAM modulations are widely used. To employ digital modulation, the analog voice signal is converted into digital form using an Analog-to-Digital (A/D) converter. The ADC samples the level of the voice signal and transmits a bit stream of ones and zeros to indicate either relative (differential step up or down) or absolute level. Two key parameters for A/D conversion are the A/D sampling rate and the number of bits needed to represent the steps.

Based on these techniques, a single voice channel’s data rate is approximately 64 Kbps in the PCM (Pulse Code Modulation) standard. This consumes a considerable amount of bandwidth in wireless channel transmission. Bandwidth is a scarce resource in wireless communication systems.

Techniques like ADPCM (Adaptive Differential Pulse Code Modulation) represent samples differentially, reducing the data rate from 64 kbps to 32 kbps. However, this is still relatively high. To further reduce the data rate, complex vocoders have been developed.

The function performed by a vocoder is referred to as vocoding. Vocoders represent voice signals in the form of bits using a technique known as a codebook. Using vocoders, data rates below 13 kbps can be achieved. The following illustrates the vocoding process.

Before vocoding, the analog signal is converted into a digital signal. This process compresses the number of bits required to represent the analog voice via prediction.

Examples of Vocoders:

• CDMA-One uses QCELP (Qualcomm Code Excited Linear Prediction)
• NA-TDMA uses VSELP (Vector Sum Excitation Linear Prediction) and ACELP (Algebraic Code Excited Linear Prediction)