AUD-03

DSP System & Gain Structure

ANSI/AVIXA A104.01 · IEC 60268 · ISO 3382 · ANSI S12.2 · ANSI/AVIXA D401.01

01 · Gain structure

Source type

Source output levelⓘ

dBu

Preamp gain

DSP input trim

DSP nominal output level

dBu

Amplifier input sensitivity

dBu for full output

System reference levelⓘ

dBu

Converter resolution

02 · Acoustic echo cancellation

Room RT60ⓘ

AEC tail length

Microphone channels

03 · Latency budget

Sample rate

Converter latency, each directionⓘ

samples

Processing block size

Serial processing blocks

Audio network latencyⓘ

04 · Matrix and I/O

Analogue inputs

Analogue outputs

Network inputs

Network outputs

Available card slots

Channels per card

Awaiting input22 fields left: Source type · Source output level · Preamp gain · DSP input trim · DSP nominal output level · Amplifier input sensitivity · System reference level · Converter resolution · Room RT60 · AEC tail length · Microphone channels · Sample rate · Converter latency, each direction · Processing block size · Serial processing blocks · Audio network latency · Analogue inputs · Analogue outputs · Network inputs · Network outputs · Available card slots · Channels per card

Capacity and layout

The drawing appears when the inputs are complete

01
Trace gain structure and headroom stage by stage, size AEC tail length against room reverberation, budget end-to-end latency, and check matrix I/O and card capacity.
02
Tip. Set gain structure once, at the input. Every stage that follows inherits the headroom you establish there, and a signal that clips at the preamp cannot be recovered by any amount of processing downstream.
03
+4 dBu — Nominal line level
04
20 dB — Typical headroom
05
≥ RT60 — AEC tail length
06
30 / 45 ms — Conversational / lip-sync limits

DSP system design

A DSP is where a signal chain either holds together or quietly falls apart. Gain structure decides how much of the available dynamic range survives to the loudspeaker, echo cancellation decides whether a conference call is usable, and latency decides whether talkers hear themselves. All three are set at design time and are painful to correct on site.

Gain structure and headroom

Every stage inherits the headroom established at the input. A signal clipped at the preamp cannot be recovered by anything downstream, and one set too low buys noise instead.

Use cases

+4 dBu nominal for professional line level
20 dB headroom above nominal is best practice
Microphone level typically −60 to −40 dBV
Signal-to-noise target of 25 dB or better

Formula

Voltage gain (dB) = 20 × log₁₀(V_out ÷ V_in)

IEC 60268 · ANSI/AVIXA A104.01

Echo cancellation and latency

The AEC tail must cover the room’s reverberant decay or echo returns to the far end. Latency accumulates through every conversion and processing block in the chain.

Use cases

AEC tail length at least equal to room RT60
One reference per independent loudspeaker zone
Under 20 ms for live reinforcement
Under 30 ms end-to-end for conversation, 45 ms for lip sync

Formula

Total latency = Σ(conversion + processing + network delay)

Coverage variation ≤ 6 dB (AVIXA ACR)

Processing resource counts and block latencies differ substantially between DSP platforms. The figures here are planning values for sizing a system; confirm channel counts, AEC instances and per-block latency against the specific platform before committing to a hardware list.