Decibels & levels

The decibel is not an absolute value but a logarithmic ratio to a reference. Only the suffix (dBu, dBV, dBFS, dB SPL) tells you what the number refers to. Without a suffix, “dB” just describes a difference, such as “3 dB louder”.

For quantities such as voltage and sound pressure the formula is 20 × log(a/b), for power 10 × log(a/b). That's why double the voltage means +6 dB, but double the power only +3 dB.

Sound & frequency

Sound travels through air at 20 °C at about 343 m/s, roughly 0.6 m/s faster per degree Celsius. The wavelength follows from λ = c / f. It explains many effects in practice: low frequencies with wavelengths of several metres bend around obstacles and are hard to control, while high frequencies are already shadowed by people and scenery.

An octave is a doubling of frequency. Human hearing covers about ten octaves from roughly 20 Hz to 20 kHz, with the upper limit dropping noticeably with age. That's why analysers often work in third-octave bands, i.e. three bands per octave.

Frequency ranges for mixing

The boundaries are fluid, but this breakdown has proven useful as an EQ guide:

The fundamental of the speaking voice is usually around 100–150 Hz for men and 200–250 Hz for women. For intelligibility, though, the consonants between 2 and 4 kHz matter most. A high-pass at 80–120 Hz on speech mics costs almost nothing and reduces footfall noise and plosives.

Loudness & limits

Audience (German standard DIN 15905-5): For events with amplified sound, the guide value is a rating level of 99 dB(A), averaged over 30-minute intervals. Measurement requires a meter of at least accuracy class 2. Other countries have their own rules.

Staff (German Noise and Vibration Occupational Safety Ordinance): From a daily noise exposure level of 80 dB(A) hearing protection must be provided, from 85 dB(A) it must be worn. For peak levels the action values are 135 and 137 dB(C). These values match the EU noise directive.

Every 3 dB increase halves the permissible exposure time for the same hearing dose. Eight hours at 85 dB(A) equals one hour at 94 dB(A). The exact obligations are set out in the applicable standards and regulations; if in doubt, their current wording applies.

Microphones

Dynamic microphones use a moving coil, need no power, handle high levels and are rugged. Ideal for vocals on stage, snare and guitar amps. Condenser microphones need phantom power, resolve more detail and respond faster to transients. Good for overheads, acoustic instruments and speech in quiet surroundings.

Proximity effect: Directional microphones boost low frequencies the closer the source is. It makes voices fuller, but also explains the boominess of singers who practically swallow the mic.

Phantom power P48: 48 V DC, carried on signal pins 2 and 3 against ground. It doesn't harm dynamic mics on balanced lines. Take care with ribbon mics on faulty cabling, unbalanced devices and when plugging with phantom on.

3:1 rule: When two mics pick up the same source, the second should be at least three times as far away as the first is from the source. This keeps comb filtering from time differences small.

Cables & connectors

Balanced lines carry the signal twice with opposite polarity. The input takes the difference, so interference arriving equally on both conductors cancels out. That's why 50 m of mic cable works fine, while an unbalanced jack cable starts humming after a few metres.

Hum usually means a ground loop: two devices are connected twice, via protective earth and audio ground. The fix is a DI box or an isolation transformer with ground lift, never taping over or disconnecting the protective earth.

Digital audio

The Sample rate sets the highest frequency that can be carried: just under half the sample rate (Nyquist). 48 kHz covers the whole audible range and is standard in live and video. 96 kHz mainly reduces latency and filter effects in processing.

The bit depth determines the theoretical dynamic range, about 6 dB per bit: 16 bits give around 96 dB, 24 bits around 144 dB. That's why with 24 bits it's no problem to leave some headroom instead of running just below 0 dBFS.

With digital consoles and networks such as Dante, latency adds up with every conversion and network hop. For in-ear monitoring the whole chain should stay well below 10 ms, otherwise your own voice sounds smeared.

Loudspeakers & room

Inverse square law: A point source loses 6 dB per doubling of distance. In its near field a line array loses only about 3 dB per doubling and therefore throws further back. You'll find the calculator here in the toolbox.

Polarity and phase: Polarity simply swaps plus and minus across all frequencies. Phase is a time shift whose effect depends on frequency. So a polarity switch on the console doesn't fix time differences between sub and top; that takes delay.

Comb filtering: When two identical signals combine with a small time offset, some frequencies cancel and others add up. It happens with two mics on one source or two speakers playing the same programme. Delay, distance or deliberately separate coverage areas help.

Haas effect: If a second sound arrives within about 5 to 30 ms of the first, the ear localises the source at the first signal, even if the second is louder. That's why a delay line is fed a few milliseconds later than the pure propagation time. See the Delay Calculator.

Reverberation: The reverberation time RT60 is the time it takes for the level to fall by 60 dB after the source stops. According to Sabine, roughly RT60 = 0,161 × V / A (room volume in m³, equivalent absorption area in m²). For speech, under 1 second is comfortable; churches and empty halls are often 2 to 4 seconds. An audience absorbs a lot, so a full hall sounds much drier than at soundcheck.

Formulas at a glance