Vocabulary

ADC, A/D

Analog‐to‐Digital (Converter); an electrical device meant to convert continuous, analog signals to numeric form. An input is a voltage or current, the output is a series of (usually binary) numbers which in some way represents the incoming signal to some sufficient level of accuracy.

ASIC

Application Specific Integrated Circuit; an integrated circuit that has been designed to perform some specialized task. Opposed to general purpose components, such as gates, CPUs, logic arrays and other which are meant to be used in a wide variety of different designs. Examples include the special purpose synthesis chips used in many common synthesizers and samplers. When used to implement whole algorithms, ASICs allow great processing speeds at a modest cost when large production volumes are involved; however, some versatility is lost because the functionality is hardwired on the chip and therefore cannot be changed by updating software. ASICs are often used as glue logic between more general purpose components and as helper chips to ease specific processing tasks.

DAC, D/A

Digital‐to‐Analog (Converter); an electrical device meant to convert a numerical representation of a time‐variable signal into analog form, usually for output into subsequent analog processing. The input is a series of (usually binary) numbers, the output is a (continuously varying) voltage or current which can be used to drive analog equipment.

DSP

1. Digital Signal Processing; the area of mathematics and engineering which concerns itself with analysis, synthesis and modification of signals by numerical means. This document is about audio DSP, but DSP as a whole also includes the processing of visual, seismic and other numerical data.
2. Digital Signal Processor; a numerical processing device especially designed and optimized for the implementation of common signal processing tasks, such as the fast Fourier transform, saturating arithmetic, cyclic buffer management, filtering and some vector operations.

FFT

Fast Fourier Transform; a loose term referring to a number of ways to implement the discrete Fourier transform more efficiently than a naïve interpretation of the mathematical definition would suggest to be possible. These methods take advantage of the symmetries of the sinusoid base in which signals are decomposed by the DFT to achieve superb performance compared to the straight‐forward dot product implementation. (Whereas using the dot product would result in an algorithm with a running time proportional to the square of the block length, FFT methods generally reduce this proportionality to block length times a logarithm of the block length. With large block sizes the difference in computation time is enormous.)