In general, the process of digitizing the signal creates an adequate digital model for a non-digital process. The challenge is trying to represent a real-life scenario as an algorithm. In some cases, the task appears to be much easier because of the existence of repeated samples. Human voice transfer requires a 4-kHz frequency band. Based on the Nyquist sampling theorem, digital sampling must take place at least twice the highest frequency to reconstruct the analog signal accurately. As a result, the sampling theorem requires that, given that fa equals a 4-kHz frequency band , fs is greater or equal to 4000 x 2 = 8000 Hz. Digitalization of voice as an analog signal includes several steps:
A unit called a channel bank can perform sampling, quantization, and coding. A coder -decoder (codec) device performs the first two functions of pulse code modulation (PCM), and the coding is performed by a data service unit/channel service unit (DSU/CSU). The process is effective and is much less vulnerable to noise, but it is more expensive. The digitalized voice is called PCM voice. |