Decibels

The decibel (dB) is a dimensionless logarithmic ratio expressing the relationship between two quantities — power, voltage, or acoustic pressure. It's the fundamental measurement language of audio engineering because it compresses an enormous dynamic range (a ratio of 10,000,000,000:1 between the threshold of hearing and the threshold of pain) into a manageable scale of roughly 0–120 dBSPL.

The Mathematics

Power ratio (dB): dB = 10 × log₁₀ (P₁ / P₂)

Voltage or amplitude ratio (dB): dB = 20 × log₁₀ (V₁ / V₂)

The factor of 20 vs. 10 exists because power is proportional to voltage squared: P = V²/R. Doubling voltage quadruples power, so a 6 dB increase in amplitude is a 12 dB increase in power — but since we're measuring voltage, 20 × log₁₀(2) = 6 dB is correct.

Key relationships to memorize:

• +3 dB = 2× power (1.41× voltage)
• +6 dB = 4× power (2× voltage)
• +10 dB = 10× power (3.16× voltage) — perceived as roughly 2× louder
• +20 dB = 100× power (10× voltage)
• -3 dB = half power (the standard definition of a speaker or filter's -3 dB point)
• -6 dB = quarter power (half voltage)
• -20 dB = 1% power, 10% voltage

Reference Levels and Suffixes

Every dB measurement with an absolute value requires a reference. The suffix specifies the reference:

The dBu/dBV difference: 0 dBu = 0.775 V; 0 dBV = 1.0 V. The professional standard of +4 dBu equals approximately 1.23 V. The consumer standard of -10 dBV equals 0.316 V. The difference between +4 dBu and -10 dBV is approximately 11.8 dB — close enough that "12 dB difference between pro and consumer" is a useful rule of thumb. Connecting a professional output directly to a consumer input without attenuation will overdrive the input by ~12 dB.

Acoustic Decibels (dBSPL)

SPL references the threshold of human hearing: 0 dBSPL = 20 µPa (micropascals), approximately the softest sound a young person with healthy hearing can detect at 1 kHz.

SPL Reference Table

AVIXA DISCAS specification targets 70–78 dBSPL for speech reinforcement in conference rooms. AVIXA PISCOR provides SPL targets for intelligibility: the speech level should exceed the ambient noise floor by at least 25 dB for excellent STI. See ansi-avixa-discas and ansi-avixa-piscor.

Inverse Square Law

In free-field conditions (outdoors, away from reflective surfaces), sound pressure level decreases by 6 dB for each doubling of distance from the source:

ΔSPLdB = -20 × log₁₀ (d₂ / d₁)

A speaker at 90 dBSPL at 1 meter produces:

• 84 dBSPL at 2 meters (-6 dB)
• 78 dBSPL at 4 meters (-12 dB)
• 72 dBSPL at 8 meters (-18 dB)

In real rooms, early reflections and reverberation add energy, so actual decay is slower than the inverse square law predicts. Rooms transition from the "direct field" (dominated by inverse square law) to the "reverberant field" (dominated by room reflections) at the critical distance — typically 1–3 meters in conference rooms.

This is why ceiling speaker placement at regular intervals (distributed systems) outperforms a single high-power speaker in reverberant rooms: it keeps every seat in the near field of at least one speaker, maintaining consistent SPL and intelligibility.

Digital Levels (dBFS)

In digital audio, 0 dBFS is the clipping ceiling — no signal can exceed it. All useful signals are negative dBFS:

The -18 dBFS nominal convention provides 18 dB of headroom above the operating level before clipping — matching the ~18 dB headroom above +4 dBu in analog professional equipment (clip point typically around +22 dBu).

Most professional ADCs calibrate so that +4 dBu analog = -18 dBFS digital. Some broadcast converters use +4 dBu = -20 dBFS. Always check the converter's reference level before connecting analog and digital gear.

Power and Amplifier Decibels

dBW expresses amplifier power relative to 1 watt:

• 1 W = 0 dBW
• 10 W = 10 dBW
• 100 W = 20 dBW
• 1000 W = 30 dBW

Speaker sensitivity is specified as dBSPL at 1 meter with 1 watt of input (1W/1m). A speaker with 98 dBSPL/1W/1m sensitivity driven with 100 W produces 98 + 20 = 118 dBSPL at 1 meter (each 10× power increase = +10 dB). This is how you calculate maximum SPL: sensitivity + 10 × log₁₀(power).

dBm is power relative to 1 milliwatt. Originally used in 600 Ω telephone systems where 1 mW into 600 Ω equals 0.775 V — hence the historical connection to dBu. In modern use, dBm is primarily an RF measurement: +20 dBm = 100 mW, +30 dBm = 1 W. Wireless microphone transmitters output 10–50 mW (+10 to +17 dBm).

Gain vs. Attenuation vs. Level

• Gain — the amplification applied to a signal (in dB)
• Attenuation — negative gain; reducing signal level (often expressed as a positive number of dB loss)
• Level — the absolute signal magnitude at a point, expressed in dBu, dBFS, dBSPL, etc.

A preamplifier set to 40 dB of gain takes a -50 dBu microphone signal to -10 dBu. That's still below +4 dBu nominal — another 14 dB of gain is needed. Gain and level together define the gain-structure of the system.

Measurement Tools

LUFS (Loudness Units relative to Full Scale) is used in broadcast and streaming for perceived loudness normalization. YouTube, Spotify, Apple Music, and broadcast standards all specify target loudness in LUFS (typically -14 to -23 LUFS integrated). Understanding LUFS is increasingly important as AV systems feed streaming workflows.

Common Pitfalls

• Confusing dBu and dBV — Professional systems use dBu (reference 0.775 V), consumer uses dBV (reference 1 V). A +4 dBu signal is NOT the same as +4 dBV. The ~2 dB difference between references compounds with the ~12 dB difference between professional (+4 dBu) and consumer (-10 dBV) nominal levels.
• Ignoring headroom in digital — Setting peaks at -3 dBFS leaves no room for inter-sample peaks (ISPs), which can exceed 0 dBFS even when the sample values don't. True peak meters reveal ISPs; standard peak meters don't. Use true peak limiting for any output going to broadcast or streaming.
• Misapplying the 3 dB rule — "3 dB doubles power" and "6 dB doubles voltage" are frequently confused. When calculating speaker coverage or amplifier requirements, use the correct formula for the quantity being calculated.
• Not accounting for ADC reference calibration — Connecting +4 dBu to an ADC that maps +4 dBu = 0 dBFS (rather than -18 dBFS) immediately clips the digital output. Always check converter specs before connecting professional analog equipment.
• Treating dBSPL as linear — A room with 60 dBSPL background noise and 60 dBSPL speech signal is not "twice as loud." Two identical incoherent sources add 3 dB (to 63 dBSPL). The difference between 60 and 63 dBSPL is imperceptible; the signal-to-noise ratio is 0 dB, which is unusable for speech intelligibility.