DSP — Digital Signal Processing

Digital Signal Processing

For a deeper dive into DSP algorithms and processor hardware, see dsp-fundamentals.

A DSP (Digital Signal Processor) is both a technology category and a class of hardware device. As a technology, DSP refers to real-time mathematical manipulation of digital audio signals — converting sound from analog to digital, applying algorithms to modify or analyze the signal, then converting back to analog for output. As a hardware device, a DSP is a dedicated processor designed to perform these operations with deterministic, low-latency performance that general-purpose computers cannot guarantee.

In AV systems, "the DSP" typically refers to a standalone audio processing unit between microphones/sources and amplifiers/speakers, handling routing, mixing, equalization, dynamics, conferencing functions, and acoustic correction in one box.

What DSPs Do in AV Systems

Signal routing and mixing — A DSP takes dozens of inputs (microphones, line-level sources, Dante audio streams, AES3 feeds) and routes them to dozens of outputs (amplifiers, zones, conferencing systems, recording) with full matrix mixing. Routing changes are made in software, not by rewiring.

Equalization (EQ) — Parametric, graphic, and Baxandall filters correct room acoustics, tonal imbalances in speaker systems, and feedback issues. Room correction systems include automatic EQ adjustment based on measurement microphone data.

Dynamics processing — Compressors, limiters, noise gates, and automatic gain control (AGC) manage signal levels. Limiters protect speakers from clipping; AGC maintains consistent loudness as speakers move closer to or farther from microphones.

Acoustic Echo Cancellation (AEC) — The critical algorithm for conferencing. AEC removes the loudspeaker signal from the microphone feed so remote participants do not hear their own voice echo. Requires a reference signal from the loudspeaker output. See aec.

Feedback suppression — Automatic notch filters detect and suppress feedback frequencies before they build into full-room feedback.

Delay and time alignment — Delays align speaker arrival times in distributed speaker systems. In a room with front speakers and delay speakers, the delay speakers receive a time offset equal to the speed-of-sound travel time: feet of separation / 1.125 = milliseconds of delay.

Major DSP Platforms in AV

DSP Configuration

DSP processors are configured using proprietary software — Biamp Tesira Software, QSC Q-SYS Designer, Shure Designer, Extron GlobalViewer. Configuration involves placing processing blocks in a graphical signal flow, setting parameters, defining control interfaces, and saving presets for different room modes.

Most professional DSPs support remote control via TCP/IP or proprietary APIs — essential for integration with room control systems (Crestron, AMX) that need to change DSP routing or levels based on room state.

DSP in Networked Audio Systems

Network-native DSPs like QSC Q-SYS Cores and Biamp Tesira accept audio inputs and outputs via Dante or AVB, eliminating point-to-point AES3 or analog connections to every device. A single network connection carries all audio in and out, with routing configured in software.

This architecture scales to large facilities — a corporate campus with 50 conference rooms may route all DSP processing through a pair of redundant Q-SYS Core processors at the IDF, each room connecting via Dante over the existing network.

Common Pitfalls

• Double AEC — Running AEC in both the hardware DSP and the software conferencing client causes voice clipping and artifacts. Only one AEC stage should be active. See aec.
• AEC reference not connected — In DSP signal flows, the loudspeaker output must be explicitly routed to the AEC reference input. Without it, AEC cannot function. This is a frequent commissioning error.
• Insufficient delay in distributed systems — Delay speakers not timed to the front-of-room speakers create a "double voice" effect. Calculate: feet of separation / 1.125 = milliseconds of delay needed.
• EQ without acoustic measurement — Tuning by ear without a measurement microphone leaves uncorrected room modes and tonal problems. Use a calibrated mic and real-time analyzer.
• Configuration not backed up before firmware update — Some DSP platforms reset configuration during major firmware upgrades. Always export the design file before any firmware update.