siren polish


Obtaining useful sound measurements that accurately depict what one can expect to experience in an actual specific room setting independent of where the speaker was modeled and measured is hard, VERY hard actually, I’ll even say close to impossible.  And yet, properly conducted measurements are perhaps the most valuable criteria available for determining a speaker’s sound quality independent of real-world, by ear listening.


Background Information


Each room is different and no speaker will propagate sound consistently the same way from one room to another unless the rooms are identical from dimension all the way down to exacting furniture, flooring type, ceiling type, and wall type.


Therefore, most speaker manufacturers take sound measurements in an anechoic setting or, at the very least; they should take measurements that are representative of anechoic conditions.  Ground plane measurements, for instance, are a surprisingly good approximation for anechoic conditions when an anechoic room/chamber is not available.


It may seem ironic that real world conditions are far from anechoic and yet to design a speaker well, anechoic equivalent measurements are required.  In fact, our physical sense of balance relies on echoes and if put into a room entirely without echoes (anechoic) most people feel off-balance and disoriented.  Thus, a good listening environment must have at least some echoic properties. So, it may seem strange that better speakers are typically modeled and measured under anechoic or anechoic equivalent conditions.  One might even wonder why anechoic measurements are used at all if the anechoic environment is not desirable for actual listening.


Anechoic or equivalent sound measurements are a necessary part of speaker and crossover modeling and they also provide a relatively consistent finished result by which different speakers may be compared in lieu of auditioning each one by ear in the same room, in the same position, at the same volume, etc., etc.  In a room with echoic properties there are simply too many sound reflective surfaces to measure speakers consistently… in other words it is difficult to determine what sound is coming from the speaker itself as compared to a reflection off of a wall, ceiling, hardwood floor, chair, coffee table, bookshelf, fireplace mantle, and on and on.  When taking initial measurements for crossover modeling, speaker designers need to model the sounds coming from the speaker only, not the in-room reflective surfaces that vary wildly from one room to another.


More to the point: Speakers designed with flatter measured anechoic frequency response curves are more forgiving in their ability to be tuned to the varying multitude of real room conditions by an external equalizer.  Therefore, the apparent standard for measurements to be reported under anechoic equivalent conditions only makes sense.  “In-room measurements” on the other hand, are much less robust in their ability to be compared as room conditions typically vary considerably in their acoustical consistency as well as sound quality.


In a perfect world one would not even need to listen to various speakers to be able to discriminate which one would sound the best.  We would simply need to look at the measurement data and SPL frequency curves to find the best one.  I am sorry to say, the world is not perfect.


Where speaker measurements fail


When speaker measurements are taken for SPL vs. frequency curves, discrete frequency signals are sent to the speaker individually.  Conversely, when listening to music or other recorded content through a speaker, frequency signals are not discrete, they overlap—SEVERAL different frequency signals are played simultaneously. Thus, a flatter modeled or measured SPL curve may not be the best design if the crossover point is wrong or inferior components are used.  In fact, two identical speakers with nearly identical SPL curves but that differ in crossover point can sound completely different when playing actual listening content.  A crossover point that is too high will encumber the clarity and definition of the higher midrange when low bass content is played simultaneously.  A crossover point that is too low will encumber both higher midrange and high end clarity and definition when mid bass content is played simultaneously. Determining the best crossover point typically takes a lot of trial and error, perhaps hundreds of measurement frequency sweeps, several crossover prototypes and several iterations of internal damping tuning and A LOT of listening.


The larger the speaker enclosure and the more transducers it holds, the harder it is to measure comparative to actual listening.  Typically the longer the distance span between the two farthest transducers on a single enclosure the more difficult it becomes to measure the speaker at the conventional 1 meter distance away with a microphone.  This is because the axis from microphone to each transducer becomes less direct.  As it is customary to center the microphone on the tweeter axis the tweeter is unaffected (always direct for on axis measurements), however the mid and bass transducers that are farthest from the tweeter are affected most.  To compensate for this it is best to increase the distance of the microphone before taking measurements of larger speakers.


Some may argue that because lower frequencies are more “omnidirectional” it really does not change the measurements that much.  But one must remember that we are not talking simply about the axis angle of the transducers, we are also very concerned about comparative distances. The more unequal the distance between transducers to the microphone, the more unequal the SPL measurements will be as well.  And in the high end world class, a 1dB to 3dB variation from reality is something to be concerned about.


Yet, increasing the measurement distance is also problematic because with each doubling of the distance the measurement room must essentially quadruple in size.  So for the really large behemoth speakers this is a real issue as the cost of a proper chamber begins to approach, or even exceed the $1 million mark.  Few speaker manufacturers can or will justify that kind of expense.  So, we now have what are referred to as “in-room measurements” and though they may not be completely meaningless, they are much less robust for comparing specs than anechoic measurements.  With in-room measurements there will be a room resonance in the lower bass frequencies and one room cannot be compared to another (as I have already discussed). So, with larger behemoth speakers it is much more difficult to compare specifications.  Unfortunately, I am beginning to see advertised measurements of smaller speakers, even mini monitors touting unusually low bass specifications and then I read the “in-room measurements” phrase.  Putting it as politely as I know how… In my mind there is no reason for smaller speakers to list in-room measurement specs except to deceive the buyer.  So, buyer, be aware.