We know that sound can be heard, although each person hears it differently depending on where they are in a venue, and it also varies a great deal with the response of each person's ear. It is true that the ear's response changes with age; although its response is clearly shown in the Fletcher-Munson curves, the most sensitive area is between 2KHz and 5KHz, and that is the area that will be lost first.
What will not change, either with age or with hearing loss due to prolonged exposure to high sound pressure levels, is the way sound is measured with analyzers of different kinds. So let's look at the different measurements.
Frequency response
To better understand how sound behaves, we first need to interpret frequency response graphs, which show the relationship between amplitude and frequency.
A response is considered flat within a 3dB window. A flat frequency response is that of a pink noise signal in which all frequencies are reproduced at the same level.
Measuring the output of a mixing console, that is, an electronic measurement, is not the same as measuring that same signal with a microphone, in which case the frequency response will reflect the response of the sound system and the acoustics of the room.
Frequency response has a resolution measured in fractions of an octave; in audio the most common is the third of an octave, which is the standard in graphic equalizers. I usually use 1/12 of an octave, since that is the smallest difference the human ear can perceive, and I think mine can too!
A higher resolution would make it harder to interpret, adjust and correct. These devices that display frequency response are called spectrum analyzers and they work in real time, which is why they are also known as RTAs (Real Time Analyzers).
The audible frequency range of the human ear lies between 20 Hz and 20,000 Hz (which means we can only hear sounds of more than 20 cycles or oscillations per second and fewer than 20,000 cycles per second). Amplitude refers to level (electrical or acoustic), so the greater the amplitude, the higher the level (volume), and vice versa. Amplitude is expressed in dB.
These cycles can be positive or negative and have a specific length and duration, called wavelength and period, respectively.
Impulse response
Another important graph which, though it may not seem so, gives us a lot of information and is very useful, is the impulse response. It shows amplitude and time: the first thing we can learn from it is the polarity of an audio signal; the second, no less important, is time.
So when we have several sound sources, we will know which one reaches us first, which has the highest level, and whether any of them is miswired. Needless to say, impulse response graphs are far more accurate than handheld polarity checkers (the ones made up of a pulse generator and a handheld detector).
This is very useful when setting up delays in sound systems, to delay the signal at a concert because of the long distances in the venue. We are talking about the delay towers found at large festivals, placed behind the FOH position.
Remember that these towers are time-aligned with the main system, but we can also use impulse graphs to delay the main system relative to the backline.
Phase response
Phase is one of the parameters that has received the most attention in recent years, and also the least understood. This is because analysis software such as Smaart Live or Mac Foh has come onto the market, and laptops are being used more and more in the live sound world.
A phase graph relates delay in degrees to frequency across the entire audio spectrum. We should know that not all the frequencies reproduced by drivers, loudspeaker cabinets or sound systems come out at the same time. This is due to the design of the driver itself, the active and passive filters, and the crossovers and processors.
The consequences of these delays can be seen below (a): if two signals come out at the same time with no delay, we have a phase relationship of 0º, so the resulting signal will be a summation; in case (b) there is a small delay equivalent to 90º, causing a smaller sum than in the previous case. Finally, in case (c), with a 180º difference, the resulting signal will have maximum attenuation, close to cancellation.
Next we see how a phase response graph is displayed on a Meyer SIM analyzer; on these we always have a second graph, the magnitude, showing amplitude versus frequency. Remember that filters introduce phase delays according to their type and order.
If we now place a microphone in front of a loudspeaker, we are making an acoustic phase response measurement. It is worth mentioning that manufacturers' measurements are usually taken at approximately four meters and then extrapolated to one meter; they are also usually done in reflection-free environments such as anechoic chambers, which must also be large enough to allow low-frequency measurements, given the long wavelengths involved.
Suppose a graph in which we see a practically flat response between 300Hz and 16KHz, while in the low-frequency area we see delay: this will be due to the presence of a high-pass filter.
We might also see a line at the top of the graph, which is what we know as coherence; it is always present and tells us the signal-to-noise ratio being used. It sets the threshold above which we will take the measurement, depending on the ambient noise.
Ideally, the manufacturer should provide a phase response graph for each loudspeaker model, as this information will be very useful when combining different cabinets.
Real-time spectrum analyzers (RTA)
Real-time spectrum analyzers measure the sound level at specific frequencies. The number of frequencies that can be measured can range from ten to more than 240. An analyzer's resolution depends on how many it can measure.
There are analyzers with different resolutions, ranging from one octave (it measures the level at only ten frequencies) to more than 1/24 of an octave (it measures the level at 240 frequencies).
The more frequencies, the higher the measurement resolution, so 1/24-octave analyzers are much more reliable than 1/3-octave ones.
While high resolution (1/24 octave) can reveal the acoustic cancellations characteristic of comb filtering, the same problems are not visible when looking at the same measurement at low resolution (1/3 octave). Another limitation of RTAs is that, when analyzing a sound system, the graph will show us boosts or dips at certain frequencies, but not the reasons for them.
Transfer function
This consists of comparing the test signal with the signal from the object being measured.
To perform a transfer function, the FFT analyzer requires two channels (two measurement points). One channel calculates the FFT of the reference signal, and the other calculates the FFT of the object being measured. The difference between the reference signal and the measurement signal gives us the amplitude-versus-frequency and time-versus-frequency response.
The advantage of the transfer function is that any audio signal can be used as the test signal (music, pink noise, a sine-wave frequency sweep), since variations in the test signal are canceled out when compared with the object being measured.
In addition, this process takes into account the latency of the measurement system used, sound cards, processors…
In practice, a measurement microphone placed at the mix position will be our measurement signal, while our reference signal will be the same pink noise we send to the system, taken at an intermediate point, such as an aux output on the mixing console, a splitter…
SPL
Finally, another factor to consider when taking measurements is sound pressure level (SPL), which is measured with a sound level meter.
In this measurement we must take into account the scale used and the weighting filter, which can be A, B, C or FLAT (linear), used for sound pressure levels from lowest to highest, respectively. The A scale is the one used by the relevant authorities when taking measurements.
A sound level meter on the A scale will not account for low frequencies in the same way as one on the C scale
Current regulations on mandatory maximum sound levels are around 105dBA in venues such as nightclubs or concert halls, while bars with music must never exceed 90dBA. Obviously, in most cases these maximum levels are exceeded by the mere presence of the audience in the venue, so in my opinion a serious, in-depth debate should take place between lawmakers, engineers and industry professionals to arrive at real, enforceable rules with a more realistic point of view.
