Digital Audio

What is digital audio?

In order for a computer to be able to process an analog audio signal, this signal must first be converted into a stream of discrete numbers, representing the original form of the signal.

This is achieved during recording with an analog to digital converter (ADC), a device which takes an analog signal from an input device and converts the signal to a series if digital numbers.

On playback, this process is reversed. The sequence of numbers, that were previously recorder are transmitted into a digital to analog converter, which converts the numbers back to sound.

Sampling and Quantization

To enable an analog signal to be converted into a digital signal it must first be sampled, that is at certain points in time the input value of the signal is taken. The fixed intervals in which the samples are taken are called sampling intervals. The number of sampling intervals per second is called the sampling rate.

Once a sample is taken, its value is then translated into an equivalent number, representing the amplitude of the sample. This process is called quantization. As a computer only has a discrete number of bits to represent the audio signal, ultimately the input value of the signal (which, being analog can take on an indefinite number of possible values) has to be rounded. The accuracy of this number is ultimately determined by the word size.

Word Size

Through the process of quantization, the amplitude of the signal has to be rounded to a discrete number that represents the level of the signal at a specific time. The word size is the number of bits used to represent an audio signal. By increasing the word size the accuracy of the recorded signal is improved, and the distortion caused by quantization reduced.

The word size used by a standard audio CD is 16 bits though for most modern AD/DA – converters support up to 24 bits.

Sampling rate

The sampling rate is an even more important consideration than the word size, for this rate directly determines the highest frequency of an audio signal that it can represent. Their relationship is stated in the Shannon – Nyquist theorem, which says that the sampling frequency must be twice as high as the audio signals highest frequency.

An Example: The human ear can hear sounds above 20Hz and below 20000Hz. Therefore according to the  Shannon – Nyquist theorem, the sampling rate must be at least 40000Hz for the digital signal to contain all audible frequencies. A standard audio CD uses a sampling rate of 441000 Hz and therefore covers the entire audible spectrum.