Digital audio is the representation of sound information in discrete level (digital) formats. The use of audio in digital allows for simple storage, processing, and transmission of audio signals.

Sampling Rate - Sampling rate is the rate at which signals in an individual channel are sampled for subsequent modulation, coding, quantization, or any combination of these functions. Sampling is the process of taking samples of an electronic signal at equal time intervals to typically convert the level into digital information. The sampling frequency is usually specified as the number of samples per unit time.

Sampling Levels (Bit Depth) - Bit depth is the number of bits that are used to represent the sample levels in an audio recording. The larger the number of bits, the more accurate the information can be represented providing for increased quality.
Quantization Noise - Quantization noise (or distortion) is the error that results from the conversion of a continuous analog signal into a finite number of digital samples that can not accurately reflect every possible analog signal level. Quantization noise is reduced by increasing the number of samples or the number of bits that represent each sample.

Digital Companding - Digital audio systems may use digital companding to reduce the amount of amplification (gain) of a digital audio signal for larger input signals (e.g., louder talker). Digital companding assigns weights to bits within a byte of information that is different than the binary system. This skewing of weighting values gives better dynamic range. The companding process increases the dynamic range of a digital signal that represents an analog signal; smaller bits are given larger values that than their binary equivalent.

This figure shows how the audio digitization process can be divided into sampling rate and quantization level. This diagram shows that an analog signal that is sampled at periodic time intervals and that each sample gets a quantization level. This diagram shows that the quantization level may not be exactly the same value of the actual sample level. The closest quantization level is used and this causes quantization distortion.