How Loud is Your Worship?
The answer is perception based and not quite simple. It’s influenced by many factors. We can measure sound pressure, and that’s useful, but correlating that to human-perceived loudness can be confusing. Here’s some clarity on the topic. A bit of tech talk is necessary first, then some practical conclusions.
FREQUENCY
Loudness is the subjective perception of sound pressure, and it’s influenced by the frequency content and duration of sounds. Our ears are definitely not flat - they’re quite nonlinear. We’re most sensitive to midrange frequencies and less sensitive to higher and lower frequencies. This is especially true at lower listening levels and become less true as loudness increases. So the tonal response of our hearing is rather complex and dynamic. Figure 1 is a well-known chart, going back nearly a century, that demonstrates this behavior by showing the required SPL (in decibels) for different frequencies to sound equally loud:
Fig. 1:
Sound level meters use weighting filters to compensate for this nonlinearity, allowing them to relate better to human hearing, usually with an “A” or “C” mode switch. A weighting relates to modest listening levels (largely ignores bass energy) and C weighting takes more low frequency information into account, better relating to louder listening levels. Measuring SPL with music that includes lots of sub-bass energy will measure higher on the C scale than on A.
As illustrated, if listening to a midrange sound of 1000Hz(1kHz) at 70 decibels, we would need to turn an 80 Hz sub-bass tone up an additional 10dB to sound equally loud, and an 8kHz high frequency sound up about 6dB to sound equally loud. While these may sound about the same “loudness”, they measure 70dB SPL, 80dB SPL, and 76dB SPL respectively. And again, this curve changes shape at different levels. Notice that at very loud levels (particularly 100, 110, and up) our ears are closer to flat (the “loudness” button on some home and car stereos compensates by boosting the lows and highs and is intended to be engaged at lower listening levels). There are also less commonly used weightings of B, C, D, and Z (which is essentially flat, or no weighting). Having no weighting reference indicated on a measurement implies an un-weighted measurement. Clearly, the non-flat frequency response of our ears is a huge factor in measuring loudness.
TIME
Human hearing averages levels in very short durations. And live music is constantly changing in level, sometimes several times per second. A very loud repetitive sound (such as a rock snare drum) may not sound as loud as a longer or sustained sound (keyboard pad or rhythm guitar) even if it measures, very briefly, at a much higher sound pressure level. For this reason, meters should have a time selection - at least a “fast” and “slow” response option. Fast, Slow, and Peak, are common modes. Do we want to know the instantaneous level of the snare strike (a few milliseconds) or the average music level sustained over a few seconds? Both are useful. But slow response metering often correlates best to human listener perception.
DECIBELS IN SOUND PRESSURE LEVEL MEASUREMENTS
dBs can be relative or absolute and are used in acoustics, electronics, optics, and several other fields. Turning a vocal up 3dB is an example of a relative dB change. We don’t care about the absolute level of the vocal before or after, only that it changed by +3dB. Absolute dBs refer to a known reference (0dB SPL), and are denoted with the suffix “SPL”. 0dB SPL is the theoretical lowest human hearing threshold, and sound pressure levels are expressed in decibels referenced above that. A face-to-face conversation may be 60-70 dB SPL and a diesel truck might be 90dB SPL. Here’s a chart of some common SPLs:
Fig 2
A meaningful SPL measurement must denote: 1) the weighting curve, 2) the time constant, and 3) the reference level. The statement “I mixed at 92dB” is simply invalid (but not uncommon). It’s missing the weighting (or implies no weighting), the time factor, and the reference level. Properly expressed, this might look like “92dB-SPL C Slow”. Stating only “92dB” and comparing that to “92dB” in another venue or application is also invalid.
Someone once made this analogy: “saying 100dB is like saying I got paid 1500 money last week. What is the value of the money?” That’s really meaningless.
There are different types and qualities of sound level meters (hardware, software, integrated handheld, mobile app, etc.). Any good meter should at least offer a choice of weighting and time modes. There are also Type I and Type II meters, which are built to different standards (Type I meters cost more). Any meter is only as good as the measurement microphone in use, whether external or integrated. Also, measuring absolute loudness requires calibration by a “microphone calibrator” tool. The variables explained above must be taken into account in any valid measurement. Other considerations:
microphone type - there are big variations. Specifically designed omnidirectional condensers are used for measurements.
microphone position – the front row, especially near a live drum kit, will probably measure louder than the back row. Sometimes, engineers measure multiple listener positions across the venue and average them.
Measurement duration - measuring during several loud songs will result in higher measurements than during a quiet song or a sermon (in most cases). Measuring over time is very useful.
CONCLUSIONS
Say you’re mixing around 94dBA Slow SPL on an arrangement with full sub-bass energy. Switching the meter to C weighting may result in readings of 10dB higher, or more, because C weighting factors in more low frequency information. Neither is necessarily wrong, as long as the user understands the difference.
Some SPL measurement tools offer data logging. These store periodic SPL values (every few seconds, minutes, or whatever is preferred) as a data file. Want to know if today’s worship set was really that much louder than last week’s? Review that report, assuming the measurements are comparable as discussed above.
Another way of expressing loudness in sound reinforcement, which I believe relates best to human perception in live worship environments, is Leq (Equivalent Sound Level). This averages the level over a stated period of time. For a 1-hour worship experience, a 20-minute music set bouncing between 90-100dB SPL followed by a 40-minute sermon ranging 78-86dB SPL might product an Leq in the range of 84dB Leq. Listener exposure of 84dBA over a one hour duration is not unsafe. Measuring during a music moment will usually read higher, and measuring during speech will usually measure lower. Leq relates well to human sound perception and exposure.
Measuring loudness is often misunderstood and expressed incorrectly, making measurements between venues or applications improper. Appropriate measurements, expressed correctly, can be compared. Comparing measurements of the same type is ok, but comparing measurements with different weighting or time factors is misleading.
If someone feels an experience is too loud, no SPL measurement will convince them otherwise. But often “it’s too loud” really means something else like “I don’t like the musical style” or “I can’t understand the words” and the listener may not even realize this is what they really mean to convey. These may or may not be actual loudness issues. Often, its a spectal issue - the midrange or high midrange is just too prominent (and we learned above that humans are far more sensitive to those frequency ranges).
Some other factors, which are out of the mixer’s control, are:
1) direct-to-reverberant ratio changes across the audience area. Indoors, a listener further from the loudspeakers hears relatively more room reverberation (a lower D/R ratio) and may perceive the lower intelligibility as tougher to understand and, sometimes, mistake this for an inappropriate loudness. Changing level will not fix that.
2) some people are hypersensitive to sound. It’s a real thing. Hearing protection or a quieter seating area (if available) are possible solutions.
Relative measurements aren’t useless. Say you measure one day with an un-calibrated meter and/or you don’t notice which weighting or time modes are selected. If you measure again at the same location with unchanged meter settings you can compare that difference meaningfully. While the actual numbers aren’t absolute, the difference is useful. Stating that “this song is averaging about 4dB louder than the last one”, for instance, without knowing the absolute SPL is fine.
Mixing live sound (including understanding loudness) is all about listener perception. When mixing live worship we are managing “feel”. The responsibility of the mixer is to make it feel “right” while conveying the stage performance across the house. This has a highly subjective component, and judging loudness is similar to judging mix balance...train your ears, get away from the booth, walk the house, listen, and know your room. The subjective components sometimes become debatable and thats when leadership has decisions to make and communicate. Sound level meters are a useful objective reference tool when understood, or a set back when they aren’t, and don’t replace good listening skills.