How Does A Digital Signal Processor Work?

All digital data (including digital audio) is represented and stored as binary numbers (1 and 0). Audio processing (such as EQ and ANC) requires operating on these 1 and 0 to achieve the expected results. A microprocessor like DSP is needed to manipulate these binary numbers. Although you can also use other microprocessors, such as CPUs, DSP is usually a better choice for audio processing applications.

Like any microprocessor, DSP uses a hardware architecture and instruction set.

The hardware architecture determines the operating mode of the processor. DSP typically uses architectures such as von Neumann and Harvard architecture. These simpler hardware architectures are typically used for DSP because they are sufficient for digital audio processing when paired with a simplified instruction set architecture (ISA).

ISA determines the operations that a microprocessor can perform. It is basically a list of instructions marked by operating codes (opcodes) stored in memory. When the processor calls a specific opcode, it executes the instructions represented by the opcode. Common instructions in ISA include mathematical functions such as addition, subtraction, multiplication, and division.

A typical DSP chip using Harvard architecture will include the following components:

Program memory stores instruction sets and opcodes (ISA)

Data storage - storing values to be processed

Computing Engine - Executes instructions in ISA and values in data storage

Input and output - use serial communication protocol to transfer data to and from DSP