5
The number of these bits generated each second is a function of sampling rate. At a relatively low
sampling rate of 8 kHz (suitable for voice) far fewer code bits are produced each second than, for exam-
ple, at the 44.1 kHz sampling rate used for commercial compact disks. For a two-channel stereo signal at
a 44.1 kHz sampling rate, some 1.4 million bits are generated each second. That's about five billion bits
per hour,which is why you'll need at least an 800 Megabyte hard disk to record an hour of compact disk
quality music.
To visualize the analog-to-digital conversion process, refer to Figure 1. At the top is one cycle of an
analog input signal wave. We've used a simple sine wave to make visualization easier. In this example,
the signal has a peak-to-peak amplitude of 20 units, measured by the scale on the left. The sampling
frequency is many times higher than the signal being sampled and is shown along the bottom of Figure
1. Once for each cycle of the sampling frequency, the sample-and-hold circuit "slices" the input signal,
allowing the quantizing circuit to generate a (digital) number equal to the closest (of the 65,536 possible
discrete values) quantization value of the input signal at the time the sample is taken. This repeats for
each successive cycle of the sampling frequency and the quantizer generates a continuous "bit stream"
which represents the quantized signal. The continuous stream of digital audio information is converted
into a digitally modulated signal using a technique known as linear pulse code modulation.
Digital-to-analog conversion (used in playback) is the exact opposite of the analog-to digital conversion
process and is illustrated in Figure 2.
In digital-to-analog conversion, the PCM bitstream is converted at the sampling frequency to a
continuously changing series of quantization levels which are individual "steps" of discrete voltage equal
to the quantization levels in the analog-to-digital process. The shape of this continuously changing
stream of quantization levels approximates the shape of the original wave. This is shown in the top half
Figure 1: Analog-to-Digital Conversion