Audio bit depth

The sampling depth, also called bit depth specifies the number of bits in the digitizing an analog signal per sample ( = sample ) can be used. So you determined in how many shades of the analog amplitude can be represented digitally. It is in addition to the sampling rate of the second parameter by a digitizing operation is described.

Basics

The sampling depth determines the dynamic range of the signal. Approximate the momentum increases with each bit by 6 dB. The higher the sampling depth, the higher the dynamics can be and the more space it can occupy the digitized signal.

Typical sampling depth is 8, 12, 16, 24 or 32 bits. The exact resolution of the analog - to-digital conversion can be lower than these nominal values ​​. Only when the entire digital value range is used, it is called real 8/16 / ... bit. It reflects the quality of the conversion.

Example: Audio

Typically, the spectrum is divided linearly, but also non- linear or adaptive gradations are possible.

Audio applications use internally mostly a higher accuracy than the audio that they work. Since mixing operations ( multiplication, addition of samples) often generate intermediate results within the given sample values ​​, the margin is increased in calculations, so as to avoid cutting off or clipping. For linear subdivisions of the value 0 represents the center position of the speaker, go where the minimum and maximum value of the data type corresponding to the maximum deflection of the membrane to the front or rear.

8, 16 and 24 bit samples are often signed integer types, 32 bit samples come in studio environments as IEEE floating point numbers before. An over-or undershooting of the corresponding ranges of values ​​leads to over-or underflow. A slight volume increase to the upper end of the possible level thus does not result in even louder sounds, it is ready cyclically, the lower end of the sound scale. Way, the speaker of the maximum deflection must flash swing to the maximum position opposite to what is perceived for the listener as overdrive (clipping).

Arithmetic example of 16 bit linear PCM

The audio CD uses 16 bit linear PCM:

• Range of values: ...
• C type: int16 (signed )
• Ada: type cd_sample is range -2 ** 15 ... (2 ** 15 ) - 1;

For maximum positive value (0111111111111111, decimal 32767), 1 is added and the maximum negative deflection is obtained ( minimum value) ( 1000000000000000, decimal -32768):

This audible as clipping overflow can be avoided by calculations are performed on 16-bit samples with, for example, 24 bit accuracy.