|Elliott Sound Products||Cinema Sound|
Cinema Sound System Setup
Rod Elliott - © 2012
I strongly suggest that the interested reader look at Lenard Audio - Cinema Sound before reading this. Much of the material there is a collaboration between John Burnett and myself, and is based on our research experiences when developing possibly the largest sound system ever created for commercial cinemas, and applying that research to a cinema installation in Sydney.
I have left out most of the history and many other details, so I could concentrate on the one major problem - the sound system and the contentious X-Curve alignment procedure. While many people may consider the sound in their local cinema to be 'good' or even 'very good', in reality this is probably not the case. Because we are watching the movie while listening, we become absorbed in the plot, action and dialogue, and the sound usually has enough dynamics to reinforce the overall experience.
The situation will be found to be very different if the cinema-goer dons a blindfold, and just listens to the sound. Without the image, the deficiencies in the sound quality become very apparent. In this day and age, we have to wonder how this could be - the sound should be superb, and far better than most people can ever hope for with their home cinema system.
The Dolby® CP650 processor (and others of its ilk) is a very versatile piece of equipment, and provides everything one could ever need to set up a high quality cinema system. This being the case, why is it that so many commercial cinemas sound so ... mediocre?
To understand the reasons, we need to examine the setup process in some detail. There is also an absolute requirement that we should understand general acoustics principles. One of these (not often voiced, or at least not in these words) is the key to understanding what goes wrong ...
You cannot correct time with amplitude!
Equalisers affect the amplitude of different frequencies, but cannot do anything to correct for room effects caused by reflected sound. Some background is needed before your understanding of these concepts becomes clear. Most of the problems in any enclosed space are the result of reverberation and/or echoes, caused by insufficient acoustic damping, combined with flat, hard and parallel walls. In general, most cinemas have enough absorbent, diffraction and diffusion surfaces to make the sound acceptable for the audience, however taking measurements to 'align' a cinema sound system is another matter.
Any claim that it is (somehow) possible to 'align' a room or other space using EQ is much like the concept of foreign languages. Most people know that speaking loudly makes no difference if the other person doesn't speak your language. Speaking slowly doesn't help either. Despite this, people persist in doing exactly these things - speak slowly and loudly and the other person must understand. Right? WRONG!
We can phrase the above a little differently too ...
You cannot equalise a room!
The above is a very common misconception, and any claim that 'room EQ' is even possible should be treated with suspicion and/or contempt. Any room has reflective, absorbent and diffractive characteristics, depending on furniture, acoustic treatment, wall, floor and ceiling materials, etc., etc. If you tried to 'equalise' a room, you are also 'equalising' the coffee table, so if it's moved (or some books are stacked on it), you would need to re-'equalise' the room ... every time something changed. This is a silly concept, but is easily proved with a microphone, simple (PC based) spectrum analyser and a pink noise source. I've done it, and yes, I can see the difference if I move the mic, coffee table or listening chair.
If I were to apply equalisation to 'correct' the response, the result is simply awful! I've done it as an experiment! It doesn't work, simply because the microphone 'hears' things that our ear/brain combination knows are irrelevant and are ignored. Microphones connected to analysers are dumb, and cannot differentiate between things that are audible and those that are not. Multiple microphones only make matters worse, never better.
There is, however, an exception. Room EQ can be applied at very low frequencies, where the wavelength of the frequency is large compared to the room dimensions. This is a region where the room is in 'pressure mode', and normal wave propagation cannot apply because the distances are too small. It's also worth mentioning that because of this phenomenon, contrary to common 'wisdom' you can have deep bass in a small room, and claims to the contrary are simply nonsense. After all, headphones can manage extremely deep bass, and their enclosed space is tiny.
As explained in this article and as anyone who has tried knows only too well, sound system measurements are fraught with difficulties. A lack of understanding by installers and misleading (or incorrect) instructions ensure that few cinemas will ever sound their best. What we can expect is that many cinema theatres will achieve an acceptable level of mediocrity, but will completely fail to delight any audience.
There are other issues as well - poorly designed and/or implemented speaker systems and underpowered amplifiers being just two of them. Because this article is primarily concerned with the practice of equalisation, these other problems will not be discussed here except in passing. Suffice to say that it is not uncommon for subwoofers to fail with alarming regularity in some theatres, because they aren't really subwoofers at all. Vented enclosures tuned to perhaps 35Hz, with no high pass filter to prevent excessive excursion at frequencies below cutoff, are almost designed to fail - especially if EQ is used to attempt to achieve 25Hz at the low end.
Throughout this article, I shall use the terms 'align' and 'calibrate' in single quotes. This implies that they are just words, and their normal meaning does not apply. True calibration implies that the system meets a proper (and realistic) standard, and that the same calibration of two or more different (but similar) items will result in very close correlation between them. Alignment has a very similar meaning, but neither term applies to the process of 'calibrating' a movie theatre.
Despite my very negative opinion of the X-Curve and the whole process of cinema equalisation, it must be taken in context - this is 2012 (at the time of writing), and no longer 1970. While measurement capabilities have improved, the X-Curve seems to be set in stone, even though it's arguably well past its use-by date. When introduced, the original 'Academy Curve' (which preceded the X-Curve) was an attempt to solve known problems, and provide the cinema goer with some consistency. The X-Curve followed this same principle, but allowed for the fact that many of the previous issues were (more or less) solved. Excessive noise and distortion were no longer dominant issues once magnetic strips were added to the film for sound (rather than the earlier optical sound-track).
In this respect, the whole process must be put into perspective, and the historical reasons considered. However, we now have the ability (but perhaps not the will) to create sound systems that are vastly superior to those that came before. Most of the original issues are now just memories, but there is still a dogged belief that the X-Curve is still relevant, and that far-field measurements are somehow useful. They weren't before, and they aren't now.
There are two types of echo that occur inside a room, large or small, but some effects don't become audible unless the room is large enough. Reverb is known to most people - it is immediately audible in a tiled bathroom, and there is a noticeable 'enrichment' of tonality within that environment. Many people like to sing in the shower for just that reason - the reverb makes them sound better!
In a larger room intended for the reproduction of speech, excess reverb makes the sound difficult to understand. Indeed, towards the back of a long room, the level of reverb commonly exceeds the level of the direct sound (this is known in acoustics as the 'far' or reverberant field).
The other issue is commonly referred to as a 'slap' echo. A sharp noise causes an almost immediate and very distinct echo, often followed by a 'flutter' effect as it dies away. Anyone who has tried to use a simple digital echo/ reverb unit will have heard this effect. The reverb is a series of rapid repeats of the original sound, dying away to nothing. If the walls are non-parallel but reflective, you may hear a 'chasing flutter' that moves from one end of the theatre to the other. Slap echoes occur in small rooms, but our hearing mechanism cannot respond to the very short delay times - typically only around 10ms for a 3.5 metre room.
These effects are all caused by time - namely the time it takes for a sound from the speakers or other source to hit a hard surface and be reflected. In a typical room, there are many surfaces that will be subjected to the original sound, plus sounds reflected off other surfaces that have in turn reflected yet more sound. This is the essence of reverberation, and while the effect can be pleasant, too much causes a serious loss of intelligibility. However, some reverb is essential or the sound is completely flat and lifeless. There is a balance between too little and too much reverb, and few theatres suffer from an excess. Some of the newer theatres may have very little reverb - not quite anechoic, but very dead.
Figure 1 - Direct Sound vs. Early Reflections
The early reflections are those that bounce off walls, ceiling or floor, and arrive at the listening (or measurement) position a short time (within a few milliseconds) after the direct sound. The time delay is determined by room size, listener position and path lengths, and the relative amplitude at various frequencies is dependent on the surface treatment and its effectiveness at the frequency of interest.
Figure 2 - Reverberation
Reverb is a longer term affair, and is measured in seconds. The standard reference for reverberation time is RT60, being the time in seconds until the SPL (sound pressure level) has decreased by 60dB. Once a room has been excited by a sound (a loud impulse is used to perform reverb measurements), the sound will bounce from one surface to the next until it is finally absorbed or dissipated. While it can be argued that a cinema (for example) should be completely free of reverb, this is not possible - especially at low frequencies. As a result, it is a fact of life that some reverb will be present, and it is essential to ensure that it is well balanced across the frequency range, and is not excessive at any frequency or location within the theatre.
Reverb is important for another reason too, and it has nothing to do with the soundtrack of the film. Humans are not used to total silence or anechoic rooms. We expect to hear certain characteristics that are congruent with our visual surroundings. In an open field, we do not expect to hear reverb because we can see far into the distance. When in a room, we do expect to hear reverb, and it should agree with our visual impression of the room's size. Large cathedrals are expected to have significant reverberation, small carpeted rooms with heavy curtains or drapes are expected to have very little. An interesting article on this topic is entitled If these walls could talk, they would whisper, published by The Guardian in the UK. The reporter describes his experience in an anechoic chamber as "disconcerting" - a hint of understatement I suspect .
However the sound system must be capable of representing the sound that is congruent with the picture and not incongruent with the picture. This is explained on on the Lenard audio site and needs to be explained again from my point of view as well. A scene of pirates on the open sea must not be contaminated by the reverberation imposed by the walls and ceiling of the cinema, for example.
Another effect that is disconcerting is when the film action is set in the open, yet you can hear the reverb of the studio sound stage. Again, this is incongruous, and confuses our senses. We see an open field, and we expect the dialogue to sound flat, with no reverb. The same would occur if the action were set in a large cathedral and we heard no reverb at all. Although common in some early films, most modern movies get these things right (but by no means all!).
The following is somewhat contentious, and should be considered as comment rather than established fact. Psychoacoustics is not an exact science, and there are many differing opinions of what is right and what is not. My opinion is that if a theatre is built that confuses our senses by having so little reverb that the patrons are disconcerted, this may be bad for business. Architects and acoustic engineers will generally strive to ensure that the room has the right balance of reverb to ensure patrons feel comfortable. Perhaps surprisingly, this is not difficult to achieve. Most theatres do sound somewhat 'dead', but not so much so that it causes discomfort. Once the film starts the difference is academic, as the soundtrack supplies the ambience needed to place the audience into the film - to become a part of the experience. This is the purpose of the surround speakers - to immerse the audience in a sound field that is congruent with the image. It would be very easy to add reverb to a completely dead theatre electronically, and the artificial reverb can be faded down with the house lights. While this would give the best of both worlds, it is not really necessary to go to such lengths. This could easily become a subject unto itself, but I expect you have the idea by now.
Desirable though it may be for the comfort of patrons, early reflections and reverb may cause problems with the reproduction of sound. Our hearing mechanism discards most early reflections, but only if they arrive within ~30ms. Of much greater significance, these effects create major problems for measurement systems that are incapable of separating direct and reflected sounds (something that humans do very well). Because of these problems, the cinema industry has a procedure for setting up the sound system in theatres. Installers use multiple microphones, most or all of which are at or towards the back of the theatre, in some cases well within the reverberant field. Naturally, these tests are done with no-one in the theatre, so only the furnishings and fittings provide damping and/or diffraction/ diffusion.
At the beginning of the setup process, a vague attempt is made to calibrate the SPL (sound pressure level) in the theatre area against a sound level meter. The CP650 instructions inconveniently neglect to specify that the meter MUST be set for flat response - most people use sound level meters with the A-weighting filter applied. This is completely wrong for making any kind of calibrated measurement. The instructions also fail to specify the type of measurement microphone, so some installers may use directional mics, others omni-directional, and some might use a mixture.
Why does the type of microphone make a difference? Because almost all directional mics have an inherent bass rolloff so response is unpredictable. If the sub is equalised using a mic with bass rolloff it will end up far louder than it should be. Directional mics will also favour sound coming from directly in front, and the installer is directed to aim the main microphone (mic 1) straight up towards the ceiling. Far more reverb than direct signal may be picked up by this mic, making any measurement pointless at best.
The first step is to set the Dolby processor to '7' (reference level), then adjust the amplifier gain(s) to achieve 85dB SPL in the reverberant field where the microphone(s) are located. Surround amps are adjusted so the level from the surround speakers is 82dB SPL (3dB lower than the main speakers). The sound level meter must be set for C-weighting (more-or-less flat, at least from 100Hz to ~10kHz). No specification for the meter is provided, so presumably it's ok to use a cheap and nasty meter costing less than $100.
Then the real fun starts. By using equalisation (bass and treble controls, a 27 band graphic equaliser and a parametric equaliser for bass), the system's frequency response is adjusted one channel at a time until each meets the industry standard 'X-Curve'. The system includes a 'real time analyser' (RTA), that displays the amplitude of each 1/3 octave frequency band. This is similar to a spectrum analyser, but the display is divided into separate vertical bars for each frequency.
Figure 3 - Cinema X-Curve
The X-Curve is named from eXtended range curve and is defined by ISO Bulletin 2969, although the more correct term is 'eXperimental', since that is exactly what it was in the beginning. It is intended to provide compensation for the fact that sound in a reverberant environment sounds louder than under anechoic (no echoes) conditions. While there are different compensation factors intended to be applied to rooms of different sizes (the standard curve is for a 500 seat auditorium), this doesn't happen. The CP650 processor has no facility to select room size, and the installer is expected to make the response fit the curve regardless of room dimensions. I think that after about 40 years it's fair to say that the experiment has failed and should be terminated.
One major factor that the X-Curve claims to address is that reverb at high frequencies is usually minimal, and that the majority of reverberation will be in the low to mid frequencies. Even the air itself will absorb the highest frequencies, and the typical theatre furnishings will generally reduce the reverb time at high frequencies to comparatively low values. If a speaker system is 'calibrated' to flat response in the far (reverberant) field, it will sound overly bright and harsh. The 'X-curve' allegedly compensates for this, but the process is flawed because you can't 'calibrate' a loudspeaker in the far field! However, this is the 'standard', warts and all.
Figure 4 - Pink Noise SPL Build-Up Over Time
The curves shown above are adapted from Ioan Allen's paper "The X-Curve: Its Origins and History" , and shows how the SPL increases over time when the room is excited with pink noise. If an attempt were made to equalise the response so it was flat, it is obvious that significant treble boost would be needed. During normal programme material which is generally transient by nature, the reverberation never has enough time to build up as shown, so the treble boost makes the sound shrill and harsh. No-one seems to have noticed that pink noise and programme material are completely different from each other, so we have the X-Curve.
|Notice something very interesting in the above curves. The red curve is the first arrival, and is the main stimulus for
what we hear with normal programme material. As opposed to pink noise. It's flat! There is no equalisation needed, because there is no reverb yet. Most sounds
will have passed before we hear any reverb, and for that reason, the theatre must be properly damped to minimise reflections.|
We don't listen to pink noise as a rule - it's not something that anyone is really hanging out to enjoy, so why on earth should we equalise a system so that pink noise supposedly looks right on a real-time analyser (without actually listening to it)? Apart from anything else, applying EQ won't make pink noise sound better anyway. The EQ certainly doesn't make the programme sound better, because it was performed with a completely inappropriate stimulus, and in the reverberant field.
The X-Curve specifies that the response should be flat to 2kHz, as measured by microphones of unspecified type*, located two thirds of the distance from screen to rear wall, in locations that are also not specified. After 2kHz, the response is supposed to fall at 3dB/octave up to 10kHz, and at ~6dB/octave above that, to the maximum of 20kHz (although 16kHz is commonly the realistic upper limit). As noted elsewhere, there is neither the requirement nor the capability to measure actual reverberation time (RT60). There is supposedly a 'small room' X-Curve, which rolls off at 1.5dB/octave, and a different version again that rolls of at 3dB/octave above 4kHz. These are just as flawed as the full sized version of course, and are therefore just as useless. Other variations exist as well, so as a standard it has rather too many variables with no clear direction.
* Actually, the mics are specified, and should be as per the SMPTE-202M document. This is quite specific about the characteristics of the microphones to be used. They must have omni directional response up to the highest audio frequency to be used, and calibrated for random incidence - not free field calibration ... see Brüle & Kjær or similar for a description of the differences. The Dolby instructions are written on the basis that the installer somehow 'knows' this.
If a standard curve is to be applied, then it can only be applied to a standard room. Since there is no facility (or requirement) to measure RT60 before 'calibration', we can be assured that the end result will be far from standard. No two rooms will ever be exactly the same, unless they are truly identical in every respect - right down to the floor covering and light fittings. There is a suggestion in the manual that the system should be checked by ear after calibration, and that small adjustments may be needed for different room sizes. This suggestion is not at all prominent, and some installers might perform a very basic listening test at best.
For the time being we shall ignore these 'minor issues', read the manual for the processor, and follow the instructions as they are written. Having done so, the installer can cheerfully head back to the office with another calibration completed to everyone's (dis) satisfaction. Interestingly, anecdotal evidence (from several independent parties) indicates that even if the same installer returns to re-equalise the same cinema with no changes to the system, the EQ will be different. This will happen as many times as the system is 'aligned'. If the exact same results can't be achieved every time for the same theatre and sound system, then the procedure is (by definition) fatally flawed. No-one seems to have noticed!
The process goes wrong for a number of reasons ...
To cover these points in order is essential ...
1 - Reverberant Field
A microphone picks up all sounds that impinge on the diaphragm, and gives each variation in sound pressure equal 'weight', regardless of origin or time of arrival. This causes measurement readings to differ markedly from what we hear - especially in the reverberant field. Because we have stereo ears, we are able to focus on a particular direction, ignoring (to a large extent) sounds that arrive from different directions.
Microphones cannot do this, so every sound at a given level is reproduced equally, regardless of its point of origin or how long it has been delayed before reaching the diaphragm. In addition, even small diameter measurement microphones show response anomalies when sound arrives off axis. Unspecified microphones are just that - unspecified, so no-one knows or seems to care about the possibly severe frequency response errors they might introduce.
The installation manual specifies that the measurement mics should be in the reverberant field, as this is claimed to be an important aspect of the procedure. This practice guarantees that the end result will be less than satisfactory.
The pink noise sound source is a relatively constant sound, and reverberation has plenty of time to build up the overall measured SPL. The film sound track will normally consist of speech, music and sound effects, almost none of which are constant. Most are highly transient in nature, so high frequencies in particular will be quieter than expected, reducing clarity. That's exactly what occurs when you roll off the top end of the audio frequency spectrum - is anyone surprised that this happens?
2 - Microphone Placement
Since no-one specifies where the mics should be placed, nor that calibrated test mics are used, it's up to the installer or setup technician to determine what mics should be used and where they are placed. Moving a mic just 300mm in the reverberant field will change the frequency response - dramatically in some cases. It is thought that by using a number of microphones (each giving a bad reading as described in #3 below), that this somehow will result in a 'good' reading.
Not so. An infinite number of averaged bad measurements simply provides another bad measurement - the number of times you do it is immaterial if your methodology is flawed in the first place. Proper loudspeaker measurements are taken in an anechoic environment, using calibrated microphones and a test process that is documented to the smallest detail. No-one in his right mind would even consider that just placing any old mic somewhere in the room and taking that as gospel was a sensible approach. This, however, is the documented procedure for 'calibrating' a movie theatre!
3 - Microphones vs. Ears
Our hearing mechanism has evolved over hundreds of thousands of years, and is specifically designed to give a very low priority to reflected sound that arrives within around 30ms after the direct sound (but only if the sounds are the same - otherwise we might ignore an important auditory warning sound while someone is speaking). This enables us to hear clearly, even in the presence of nearby reflective surfaces. This ability is enhanced by our ability to detect direction, so any signal that is essentially the same as the first, arrives within the 30ms window and comes from a different direction barely registers at all.
Microphones cannot (and do not) do this. All sounds are registered equally regardless of direction (for an omni-directional microphone at least), with the microphone responding only to the relative amplitude and phase of any two (or more) signals. A microphone measurement of a loudspeaker in a typical listening room or theatre will respond to each and every reflection, giving a highly unrealistic representation of the true performance of the loudspeaker system.
Figure 5 - 500Hz + 530Hz Example
In addition, consider a microphone and measurement software subjected to two or more frequencies at the same time, separated by perhaps 20 or 30Hz as shown above. They can't tell the difference and will average the SPL to obtain a value that is not matched by what we hear. Even though people's ear/brain will hear the effect easily (we hear it as a modulated tone), the measurement system will give a reading based on the average composite sound which may not match the audible characteristic at all. It has no ability to know that there are two different frequencies involved, each with possibly quite different reverb characteristics. Since the 'calibration' is performed using pink noise, this exact issue can potentially exist in reality, because true pink noise in a reverberant environment effectively contains all frequencies at once.
4 - Empty Theatre
Human beings (en masse) have fairly good sound absorbent properties, and our hard surfaces (heads) also act well as diffusers. An empty room (even a theatre with plush seating) will show very different characteristics when full and empty. There is absolutely no provision to even try to compensate for these effects!
5 - Equalisation
Where a loudspeaker is deficient in some region of the frequency spectrum, it is sometimes (but by no means always) possible to apply (preferably modest) EQ to correct response anomalies. If we have three front speakers (left, right and centre), it is reasonable to expect that they will be either virtually identical, or at least very similar (ideally, left, right and centre speakers should be identical).
If frequency correction is applied to a loudspeaker, identical correction should be applied to all others that are the same. As noted above, left and right speakers will be identical for all intents and purposes, and the centre channel should ideally be likewise.
When EQ is applied differently to an array of (identical) speakers, the imaging is destroyed, and the overall listening experience almost always creates listener fatigue, loss of clarity and focus, and simply sounds wrong (not unreasonable, because it is wrong). The setup process 'calibrates' each speaker independently, and all three front speakers (even if physically identical) will have different EQ applied. This approach is just plain silly - it can't ever work properly!
Many theatres don't have 'real' subwoofers, so the same loudspeaker may be expected to handle both bass and lower midrange. In such systems, the left and right speakers will be close to room corners, and will have greater bass output than the centre speaker. In such cases, it may be necessary to apply slightly different bass EQ to the centre speaker, but left and right speakers must be the same to preserve imaging. Where any loudspeaker driver is expected to handle both bass and lower midrange, expect high levels of intermodulation distortion if there is significant LFE (low frequency effects) material along with speech or other lower midrange programme material.
6 - Reverb Time
Attempting to equalise a room to a standard frequency response simply doesn't work (see below for all the reasons). Attempting to do so to compensate for reverberation time that has not been measured is pure folly. The X-Curve allegedly accounts for reverb, but there is no process for measuring (or even estimating) how much reverb exists, so any attempt to equalise for an unknown quantity is guaranteed to fail, even if it were possible in the first place.
7 & 8 - Listening Tests
Any listening test is subjective, and different people will hear different things (even from the same system). However, an installer should be someone who is interested in good sound reproduction, and as such should be able to make an informed opinion as to whether the system has potential, exceeds expectations, or is best suited to land-fill. That there isn't a single suggestion anywhere in the CP650 (or any other that I'm aware of) setup manual to listen to the system (other than to check for rattle, buzz or other signs of major component failure) indicates that the entire process is determined by equalisation alone, and if this meets the 'standard', the system is supposedly fine.
The entire official 'calibration' technique is completely unsatisfactory on all respects, as should be obvious.
The surround speakers get the same treatment, and will almost always have different EQ applied to left and right surround channels even though they are (or should be) the same. While this is not as great an issue as with the main speakers, it is still incorrect to apply different equalisation to each bank of speakers because the ambience can be destroyed by different amounts of EQ (radical or otherwise).
Likewise, if dedicated rear channels are used in a theatre, the same will apply.
As should now be quite apparent, the vast majority of issues are due to reverberation within the theatre, and are therefore the inevitable result of time delays and reflections. The solution isn't at all difficult to understand. Remember ...
You cannot correct time with amplitude!
... Yet this is exactly what the industry standard setup procedure recommends to 'align' or 'calibrate' a movie theatre.
We can add to this ...
Microphones and ears respond to sound completely differently !
Nearly all of the response aberrations measured by the microphone(s) in the reverberant field are the result of time - assuming that the speakers have a respectable frequency response to start with. There is a time delay between the direct sound being reproduced until it reaches the mic, and additional time delays before the reflections start to arrive. Since the mic cannot differentiate between the direct and reflected sounds, it will show a frequency response that is completely at odds with what the audience will hear.
If the installed speakers are wrong, near field (i.e. microphone as close to the speakers as practicable) measurement can be used to correct any minor anomalies, and correct for screen attenuation of high frequencies for example. The equalisation capabilities of the processor are sufficient (in most cases) to make appropriate corrections, but even here there are difficulties (again involving early reflections, at least in part) that can give a very unrealistic indication of the actual frequency response of the loudspeakers. If there are serious issues with the speakers, they must be replaced or repaired. In some cases, a simple change of crossover frequencies can rescue an otherwise mediocre speaker system. A truly bad speaker system is unlikely to be able to be saved, and should be replaced.
Assuming that the operator understands exactly which aberrations are likely to be caused by reflections (and ignores them as required), the loudspeakers can hopefully be equalised to sound at least passable, and if this is not possible, the installation should be halted at that point. Once an EQ curve has been defined for any one speaker, all speakers of the same type must be equalised identically.
To do otherwise will cause the effects referred to above - loss of clarity and focus, and listener fatigue. For anyone who has been to a movie theatre recently, you should now have a very good idea as to why the experience may have been less satisfying than should have been the case.
Of course, most home theatre systems are not equalised at all, and in many cases can sound far better than one's local cinema (mine does!). This is a guaranteed way to force people out of real cinemas and get the movie on DVD as soon as it's released (or obtain a pirate copy well before DVD release).
A 'real' cinema should give the audience a better experience in every respect than a home theatre system, and if they fail to do so, loss of patrons will continue at an ever increasing rate. The cinema patrons will always like the large screen and ultra-sharp picture that film produces, but if the sound is ruined because of poor alignment technique they will ultimately be driven away. Simply increasing the volume isn't the right answer!
In his article "The Mythical X-Curve" . John F Allen writes ...
While a directional speaker that sounds right to the ear in a living room may indeed exhibit a flat upper frequency response with a real-time analyzer and pink noise, such will not be the case when a speaker is in a room the size of a theatre. When equalized with pink noise to show a flat response in a theatre, speakers deliver sound with too much treble. The resulting sound is unnatural, way too bright and impossible to listen to. This, again, is due to the far greater reverberation of the larger room being included in the measurement. Since there is more low frequency reverberation, the lower frequencies appear to have a greater amplitude than the higher frequencies. Looking at such a measurement on a real-time analyzer, the higher frequencies appear to be rolled off.
The X-Curve was an attempt to normalize the shape of such a measurement in a large room. It resulted from measurements made of theatre speakers after they were equalized to sound the same as a set of studio monitors placed at the console position. When the two sets of speakers sounded as close as they could, the theatre speakers exhibited a frequency response that was basically flat from 100 to 2000 Hz and rolled off at a rate of 3 dB per octave above 2000 Hz, when playing pink noise and measured on a real-time analyzer. Below 100 Hz, the X-Curve showed a roll off of these lower bass frequencies. But this primarily due to the weakness of the older theatre speakers in the bottom octave. Rolling off the bass a little would help prevent these systems from being overloaded and damaged. It was also noted that larger theatres would exhibit a somewhat steeper high frequency roll off, and that smaller theatres would exhibit a slightly reduced roll off of the high frequencies. This finding was officially noted in 1990. Beyond that, there have been few additional guidelines to aid technicians in the interpretation of these measurements and the equalization of cinema systems.
It is incomprehensible that after all this time (22 years at the time of writing), the same processes are still used and recommended for cinema sound system alignment. While the X-Curve is still something that needs to be considered, it should not (must not) be used as the standard for system setup. No real effort has even been made to adapt the equalisation curve to account for room size, let alone taking even the most rudimentary RT60 reading, although it is required for THX.
While some installation technicians (as noted by John Allen) will use their own judgement, most will simply follow the instructions. The results are predictable and can be heard all over the world - cinema sound that is indistinct and lacking clarity, and producing listener fatigue because the EQ causes an overall loss of focus and image.
In Dolby's defence, setting up a sound system is not an easy task, and they have attempted to provide a process that will give an acceptable result. Given that few installers will have the specific skills needed in acoustics and electronics, the process described is designed to make the system EQ as painless as possible. With the appropriate background knowledge, it is obvious that the methodology is flawed and can never work properly.
The comments and recommendations in this article are not in isolation - John Allen (of HPS-4000) and Ioan Allen (Dolby Laboratories) have both presented papers to the industry (International Theatre Equipment Association and SMPTE, plus industry magazine articles) that state much the same thing. Both have extensive experience in the cinema industry. My involvement is more recent, yet it was immediately apparent that the established standard alignment procedure was simply wrong.
In my opinion, the industry has had more than long enough to get its act together and scrap the X-Curve, yet nothing has changed. There are still far too many people in the industry who continue to think that this fatally flawed system is 'right', and there is enormous resistance to change.
The CP650 processor I worked with is one of the latest Dolby processors, but it still dictates a setup and alignment procedure that has been demonstrated to be in error, defies logic and ignores basic acoustic principles.
Fairly obviously, it is imperative that the sound system (where 'system' means left, centre, right and all surrounds) is reasonably free of audible defects, bearing in mind that there is no speaker system that is actually free from colouration. The system needs to sound well balanced and free of audible discontinuities across the range, before any attempt at equalisation is made. Speaker EQ only works if the amount of EQ needed is small and doesn't require sharp filters to boost or cut any frequency.
The main things that are missing (or simply wrong) from the alignment process are many, but one of the main ones will always be difficult. It is vitally important that the installer listens to the system. Not the rather cursory listening test suggested in the CP650 manual for example, but a comprehensive listening test that has defined objectives. With experience, it is possible to isolate many problems with no test equipment at all, other than a pink noise generator (already built into the processor) and a pair of trained ears, whose owner knows what to listen for.
This part is critical, but it is surprisingly easy to demonstrate problem areas and teach someone what to listen for, and how to do so accurately (within limits of that person's hearing of course). An instant reference is available using a set of headphones. Even relatively cheap headphones have far fewer frequency aberrations than any typical speaker system, and the instant comparison allows the operator to listen for specific differences - typically frequency peaks or dips. Peaks are the worst offenders, because they tend to be far more audible than dips. The latter can limit (or completely ruin) clarity and definition, but can be harder to isolate.
Major (severe) peaks or dips indicate that something is seriously wrong with the loudspeakers, and these problems cannot be fixed with EQ. The only way to address this kind of problem is to have the supplier identify the cause and fix the loudspeakers.
Amplifier racks should ideally be co-located with the main speakers. It is (IMO) unwise to have the amps in the projection room and have to run long heavy-duty cables all the way to the back of the screen. Only a single send is needed for each speaker stack - the electronic crossovers must also be in the same rack as the amplifiers. For anyone to think that the use of passive crossovers is alright is unthinkable in this day and age. Only a fully active (preferably 4-way) speaker system can do justice to a well recorded film sound-track, and be capable of the dynamics needed while retaining sensible amplifier power.
Of course, this approach does have some issues, since cinemas may have many screens operating at the same time. When the amps are not located in the projection booth no-one knows if there is a fault, but this is a fairly easy problem to solve with the application of a bit of technology to allow any amp/ speaker combination to be monitored individually, and raise an alarm if an abnormal condition arises. Swapping out a faulty amplifier is admittedly a little more difficult though.
4.1 - Reference Level
To obtain and maintain the correct reference level is an experience in itself. Unlike the broadcast or professional public address industries, there is nothing in the setup procedure to calibrate the power amplifiers in their own right. There are no details (or procedures) provided to allow the amp's level (volume) controls to be set so that a measured output reference level is obtained for a reference level output from the decoder. The alignment will normally be done with amp level controls set to maximum (as suggested in the installation manual), but this is only a suggestion and may not happen. At some time after alignment, levels will be changed. Will the projectionist be able to return to a known (and calibrated) setting should someone fiddle with the controls? If every setting is recorded in the projection room log, perhaps. In general ... probably not.
What happens if an amplifier fails and is replaced by another with different gain? Now we have a real problem, because there is no process to define the speaker voltage referenced to line level (from the processor). To include a procedure that sets a specific gain structure to the entire B-Chain¹ is not difficult, and needs to be included. This would allow re-alignment of amplifier levels using nothing more sophisticated than an AC voltmeter - a patchable VU meter would work just fine. Agreement of the reference level is another matter of course - some will argue for 0dBm (775mV), others for 0dBV (1V) and others for +4dBu (1.23V) as is common in professional public address and many recording studios. It actually doesn't matter which standard is used, so long as the reference level information is kept in the projection room log, or labelled on the amp rack.
Note 1 - The B-Chain is that part of the processor that handles the signal sent to the speakers. The section that handles the analogue and digital signals from film is known as the A-Chain.
4.2 - Power Amplifiers & Crossovers
Many of the systems available today still (inexplicably) use passive crossover networks. A cinema installation is a professional application and can be very demanding at times. There is no reason at all to use a passive crossover for any system, even for the smallest theatre system. Electronics can be produced at such low cost that every system should be fully active, and use electronics crossovers for everything other than the surround speakers. Surrounds are used in relatively large numbers and are not usually expected to have the same response or definition as the main system, so an exception is more than reasonable in this application. The surround speakers are expected to be able to provide the same SPL, however few can even come close.
By using electronic crossovers, each amplifier has a somewhat easier task, and power requirements can usually be reduced for each amp. This approach gives a system that is capable of being louder and cleaner than an equivalent passively crossed loudspeaker, all other things being equal. This topic is discussed at length in the ESP article Biamping - Not Quite Magic (But Close), and is recommended reading for those who have not used electronic crossovers.
By applying this approach, the installer has total control over each frequency band in the system, so reliance on passive crossover networks being right is eliminated. Even if a network is (theoretically) right, it may not be right for the specific conditions encountered in a theatre environment. Because an electronic crossover can achieve 24dB/octave filter slopes easily and cheaply, each driver in the system has greater protection from out-of-band frequencies - especially important for high frequency compression drivers. (Use of 24dB/octave Linkwitz-Riley crossovers is mandatory for THX certified systems.)
Ideally, each individual driver should have its own amplifier. This affords maximum control over the driver's resonance and creates a very robust overall system. On the same topic (driver control), the power amplifiers should be located as close as possible to the speakers. Very long cable runs can add significant resistance to the speaker circuit, resulting in large power losses and loss of driver control. While it may seem more convenient to have the rack in the projection booth, the losses associated with this practice can become unacceptably high unless very large diameter wiring is used.
Subwoofers pose another set of problems, and these are often not addressed at all. Many subs use vented enclosures, and while these can give very good performance, the bandwidth must be limited to prevent all signals below the box tuning frequency from being amplified. There have been many cases where certain sound tracks have caused subwoofer failures in multiple cinemas, and this is the direct result of allowing frequencies below the enclosure cut-off frequency to be amplified and sent to the subs. Added EQ (by an installer not familiar with the box limitations) will increase the likelihood of driver failure dramatically. This is easily fixed with a professional electronic crossover, but the system processor is unlikely to have any such provision.
Although references are few, all sub amps should be fitted with adjustable limiters to prevent excessive power at any frequency. Many of the power amplifiers available are more than capable of destroying any loudspeaker ever made - regardless of its claimed power handling. Failure to limit the power to a safe value will ultimately cause failures, and it is guaranteed that these will occur at the most inconvenient time possible (yes, Murphy really was an optimist ).
That many of the installed systems are grossly underpowered is another issue again. The typical average SPL is expected to be around 85dB in the theatre, but peaks can be a great deal louder. If a system is struggling to get to 85dB and we expect peaks to reach 105dB, this is a ratio of 20dB. The amplifiers need to be able to produce 100 times as much power to reach 105dB from the 85dB reference level. A 200W amplifier that just reaches 85dB needs to be upgraded to produce 20kW to achieve the 105dB level. No driver made can withstand so much power, and high efficiency loudspeakers are the only way to keep power requirements to a reasonable level.
Very few people will say that cinemas sound excellent. Some sound very ordinary indeed, and not necessarily because the sound system is inherently bad. Properly set up, many systems are capable of providing a satisfying experience. Not excellent perhaps, but certainly better than they sound now. Some will be seriously underpowered, or will be of a design that will never work properly given the requirements of a cinema system. Even so, they can still be made to sound at least passable if properly set up.
All frequency response variations that are caused by reverberant field energy are time related, and are caused by reflections from walls, ceiling and floor - each with its own time delay. Every surface and every surface treatment will affect the amount of reflected signal at any given frequency. Because all of the variations are displaced in time from the original sound, none of the problems so caused can be corrected using (amplitude based) frequency response modification.
In case you think that perhaps digital delay might help, the answer is (in general) "No". A highly reflective surface could be tamed by having a speaker (or multiple speakers) mounted on that surface, providing an anti-phase signal to cancel the echo ... this might work in some (limited) cases, and even then only at low frequencies. However, the cost and complexity to do so is disproportionate to the benefits, and it is simpler, cheaper and usually more effective to treat the surface as needed. Treatment may include absorption and/or diffraction/ diffusion. Properly applied, these can make the problem far less of an issue.
Frequency response variations caused by deficiencies in the speaker system can sometimes be equalised or corrected electrically by other means - for example phase reversal of drivers to account for electrical phase reversal in crossover networks. Any equalisation must be performed by someone who knows (or knows how to calculate) which frequencies are affected by reflected sound from nearby surfaces. The measurement mic will give spurious (and useless) results at a number of frequencies based on the distance from the sound source to the mic and any surrounding surfaces. Anything that seems completely wrong (and that your ears tell you is not the case) is almost always the result of reflections causing the microphone to provide incorrect data.
Using multiple microphones will not help, and in most cases will make matters even worse than using a single measurement mic. Multiplexers are suggested for some parts of the EQ process because allegedly using multiple mics is somehow 'better'. These (along with the extra microphones) should be left in the cupboard where they belong. Where one microphone can give a bad reading, many microphones will simply provide many bad readings.
Whenever loudspeaker measurements are performed by the designer, only one microphone is used in an anechoic measurement area, and is commonly placed as close to the speaker as practicable to minimise the influence of reflections. This is admittedly difficult with large cinema systems, but the established methods used at present simply don't work, so a new approach is essential.
All equalisation should be as gentle as possible - a speaker system that requires radical EQ to sound even passable has no place in a theatre or anywhere else, and should not be used. As mentioned above, all identical loudspeakers must be equalised identically - regardless of what the measurement microphone may indicate. It is then essential that the installer carefully listens to each speaker to ensure that they sound the same (a mono source directed to each speaker in turn is needed). This should be done with pink noise, dialogue and music, and careful adjustments made to ensure that dialogue (in particular) is clear, crisp (but without excessive sibilance), and has no "chesty" resonances. While such resonance may sound ok if you listen to talk back radio announcers, it has no place in a cinema. All such effects are applied on the soundtrack where they are needed - they must never be part of the overall sound.
Where it is obvious that one of the speakers sounds different from the others, find out why! It may simply be that a high frequency horn is behind the masking screen (the black material on either side of the screen itself), or there may be a faulty loudspeaker in the array. Other things can influence the sound as well, and all potential physical causes must be examined before resorting to equalisation. EQ is the last step in the setup process - not the first! Ever !
4.3.1 - Screen Loss
There is one place where EQ is absolutely essential, and that's to compensate for what's known as 'screen loss'. Because the main speakers are located behind the projection screen, all sound has to pass through the screen itself before it gets to the audience. This isn't a problem for bass which passes straight through, as does most of the midrange. High frequencies are heavily attenuated, and the presence of the screen can also interfere with the normal sound propagation in the horn (almost all theatre systems use horn drivers for upper midrange and treble). Theatre screens are always acoustically 'transparent', but the degree of transparency can never be as great as one might like, because too much light would be lost. It has been determined that the screen loss of 'typical' screens is in the order of 6dB/ octave above 5kHz .
The key point here is well balanced, meaning that there will never be sharp peaks, and the all-important 'intelligence band' (my terminology) from 300Hz to 3.4kHz must be free of colouration and distortion. This range should be flat, but not if radical EQ is the only way to achieve the flat response. This frequency range provides the listener with all the dialogue detail needed to understand what is said, and for this reason (not at all coincidentally) is the frequency range used by the telephone system.
This, very unfortunately, means that subjective assessment becomes an important part of the installation process. The idea is to make the system sound good, not flat - while these conditions will coexist in a very well designed system, many systems will never sound good if an attempt is made to equalise them to be flat. As soon as subjective assessment becomes part of any installation process, problems are created. Each individual will have a slightly (or sometimes radically) different idea of what sounds 'good', so any installation needs to be verified by consensus - a number of people should agree with any change, and should agree when a system is sounding as good as it can.
This is at odds with the idea that a single person can come into the theatre, set up a few microphones, perform a 'standard calibration' and equalisation process, pack up and leave. Now it seems that we need a few extra people who know what systems should sound like, so they can argue amongst themselves until consensus is reached. However, this is not necessarily true.
The key to understanding what sounds genuinely good (as opposed to what some people may think is good) is education. It doesn't take very long to demonstrate good and bad sound to someone who has the capacity to hear the difference. It is not especially difficult to let a new installer know what to listen for, and what to do about it ... or what can be done about it. People in the industry need to understand how our hearing works, how microphones make a complete hash of things if set up incorrectly, and how to measure a system properly. With education, an installer will know quite quickly when nothing more can or should be done.
Education appears to be the missing element in the process at present. While some installers simply follow the manual, others take more care and use their knowledge and judgement to perform the setup. Those with the education (probably self taught) will generally get the best out of a system, while those who simply follow the 'rules' will make a few systems sound better, others worse, and the remainder will be pretty much unchanged (but different). The human mind can be strange at times - if something sounds different after it has been messed with, it will nearly always be perceived as 'better'. This can even happen if it is demonstrably worse!
For the (quasi-religious) fanatics of X-Curve alignment and anyone else who doubts this material, please do yourselves a big favour. At the next installation you perform, first verify that the sound from the speakers (with no EQ at all) is as it should be. If the speakers don't sound right, get the people who installed them to come in and correct the problem(s) before continuing. Sounding 'right' means that voices should be clear and intelligible, music should sound like music, without harshness or shrillness that hurts your ears.
The only thing that should sound like a goat pooping on a tin roof is ... a goat pooping on a tin roof! In other words, the speakers, without any treatment from the B-Chain processor, should sound as they should ... 'right'. If you would be happy to have the sound you hear in your living room, then they are probably ok. If your only music system at home is an MP3 player, AM radio or a pair of computer speakers, please find employment in another industry - you are totally unsuited to setting up a sound system. (Yes, I am quite serious.)
The installed system needs to be verified as capable of reaching the required levels at all frequencies, without distortion. This alone will defeat many systems - they are often undersized, sometimes by an astonishing margin. A system that cannot achieve the required SPL can't be made to do so without major upgrading. This can be an expensive exercise, and one the theatre owner(s) my be unwilling to undertake.
After the speaker installers have done their best, send all test signals to the centre speaker. If needed, apply the minimum EQ to obtain a reasonably flat response, as determined by using a single omni-directional measurement mic positioned close to the speaker, and preferably on axis with the high frequency horn (this assumes identical left, centre and right speakers). Listen to the speaker carefully, and make corrections as needed to make it sound right. Compare the speaker response to that from headphones (not ear 'buds' - proper headphones). The important part of this process is to make the speaker sound right - measured response is secondary to sound quality. Do not rely on the microphone - it only tells you a part of the story. While a useful tool, it can mislead you in any number of exciting ways. Only use the one speaker at this time!
Most important of all, ignore every instruction in the processor setup manual regarding mic positioning and equalisation, except where it helps you to work through the menu system to apply the most basic EQ possible.
Using a CD or DVD with known clear dialogue, listen up close, in the 'prime' seating area, go right at the back of the theatre, etc. The sound should be excellent at any location in the theatre. Now do the same with music. Listen for colouration in each location. If there is none up close and lots further away, the room is bad and should be corrected with acoustic absorbers and/or diffraction or diffusion material before you continue - unlikely but possible. Unless you have a good background in acoustics, it is best to engage a professional. Acoustics can be a black art, and the best solution isn't always the most obvious.
If EQ was applied to the centre speaker, apply exactly the same EQ to the left and right speakers, again assuming that all three are the same. Play a film sound track, listen from every (sensible) location in the theatre. As before, the dialogue should be clear and have excellent intelligibility, no matter where you sit.
Listen carefully to sounds that pan across the screen, and to sound that is supposed to come from a particular location. Make sure that it comes from the place it's supposed to. Listen with your eyes closed, and verify that you can locate the precise point where the sound seems to originate - verify that this makes sense in context with the on-screen action.
During the course of this exercise, make notes that will help you to remember - our auditory memory is notoriously short.
After having done the above tests, if you still have doubts that you have created an auditory masterpiece, perform the setup exactly according to the processor instructions. You can even tell the speaker guys that they can make the speakers sound horrible again if that's what you started with.
When the setup is complete, go back into the cinema and use the same material at the same volume (this is important) as you did before. How's the sound now? Better than the first test?
Listen very carefully to the same dialogue and same music. Listen from the same locations within the theatre. Does everything still sound as it should, with pinpoint accuracy of location, completely clear and intelligible speech? Do all music passages completely fail to hurt your ears? Do the high frequencies have sparkle, giving the same clarity as before, and without any harshness?
If there is just one "no" in any of your answers to the above, you have proved the point. While the 'official' setup process is unlikely to produce a catastrophic failure (although this is certainly possible), the chances of it producing a better result than the method described are almost nil. The critical thing is to know what to listen for - once you know that, the rest falls into place.
It is useful to provide some specific examples of what goes wrong when we attempt to take a measurement of a loudspeaker system under non-anechoic conditions. Anechoic chambers are used when accurate response measurements are needed on any sound reproducer. That a movie theatre is non-anechoic is obvious, and some reverberation is necessary as discussed above.
When a microphone is used to take a measurement, the direct sound from the loudspeaker is the first to arrive. Early reflections are those that bounce off walls, the ceiling or floor, arriving shortly after the direct sound. A path length difference of only 345mm causes a 1ms delay - it is safe to assume that the vast majority of early reflections will have to travel a great deal further than 345mm in a typical theatre - even a small one.
If we assume for the sake of simplicity that the first reflection from the left speaker is from the left wall, it may have to travel 2 metres further than the direct signal in a reasonable sized theatre. This represents a delay of 6ms, which is well within the 30ms limit where our hearing mechanism rejects such sounds as being spurious. In addition, I added a 7ms and 10ms delay, each at a lower amplitude than the one before - again, this is typical, but doesn't apply to any specific theatre. Actual figures will vary, but the effect is the same. Because a microphone cannot reject spurious signals, it will tell us that the frequency response curve looks something like that shown below.
Figure 6 - Frequency Response With Early Reflections (6, 7 & 10ms)
High frequencies will always be attenuated more than low frequencies. This is not simply because the high frequencies are more easily absorbed, although this is true. Another factor is that the high frequencies are more directional, so far less original signal even gets to the side wall to be reflected. This effect has been included in Figure 6, by attenuating the delayed high frequencies at 6dB/octave, starting from around 500Hz. The same is done to each reflection that is added to the direct sound. Each reflection used in the above is at a different level, as will commonly (but not always) be the case. The first reflection (6ms) is 6dB lower than the direct sound, the second (7ms) is 12dB down, and the third (10ms) is 16dB down. If by some horrible chance all reflections were at the original level and/or have more high frequency content, the graph will look a great deal worse (and yes, that is possible).
Now, if we use multiple microphones and a multiplexer (a device that allows one of several mics to be selected), then the problem should go away - this is the reason that a single mic is not recommended in the setup process, right? With more microphones to choose from, the number of possible combinations is now increased dramatically, but the net result is not improved at all. The setup manual says to use a microphone multiplexer, but goes into little detail (well, none actually) as to how this should be set up in its own right, other than to "select sequence mode" when calibrating the subwoofer. Suffice to say that sequencing will not achieve anything that is dramatically more useful than a single microphone. While the multiplexer does allow the installer to select any mic in the group, who's to say which one is right? All of them? None of them? (The answer is actually "none of them".) As stated above, if one mic takes a bad reading, multiple mics will take multiple bad readings. There is no magical number of bad readings that constitutes a good reading.
As you can see, the single mic response looks appalling. Selecting a different mic will give a different appalling result, and any attempt to equalise to make the response appear flat will be an unmitigated disaster. The end result will sound absolutely dreadful, and you will not have solved the problem at all - only created another far worse problem. Fortunately, it is easy to fix - simply reset the EQ to flat.
It may be worthwhile here to add to the original statement that defines the process ...
You cannot correct time with amplitude, and ...
Throwing (expensive) technology at the above makes no difference whatsoever, because ...
You still cannot correct time with amplitude !
It is very difficult to understand how companies with vast technical resources could have failed to see that the entire process they recommend is fatally flawed. While the end result may comply to a standard is of no consequence. The standard itself is flawed, and until it is totally reconsidered and changed to match reality and established acoustics principles, theatres will continue to sound the way they do now - not necessarily bad, but certainly not good either.
It is almost as if there were a global conspiracy to ensure that no cinema should sound so dramatically better than its competition as to raise any questions from the patrons. While the X-Curve and everything connected to it is an attempt to ensure acceptable sound, the industry should be striving for outstanding sound - acceptable simply isn't acceptable when exceptional can be achieved with very little additional effort.
Many of the speaker systems commonly used can undoubtedly be aligned to provide excellent performance (independent of the processor), and those that can't have no place in a cinema. There is little doubt that some of the systems rely on final calibration to correct response anomalies, but this can usually be fixed without incurring a severe cost penalty. To rely on the processor to correct any loudspeaker problems is not the right approach, and the industry needs to set minimum standards for the equipment that must be met - before any equalisation is applied from the B-Chain processor or other projection room common equipment.
What is the likelihood of change? Unfortunately, the prognosis is not good based on the reactions that have been heard so far. Many of those in the industry appear to have a vested interest in maintaining the status quo, and allowing reality to impinge doesn't seem to be an acceptable option. There are notable exceptions of course, but they haven't been able to force a change either.
There is one full sized cinema sound system in Sydney (Australia) that I know of (because I helped design and install it) that was not set up according to the standard X-Curve, and the speaker system itself is calibrated to sound at its best with no external equalisation. When standard Dolby calibration was performed on the system, the results were very disappointing indeed. A signal fed directly into the amplifier racks sounded really good, but film sound tracks using the processor sounded ... well, wrong. Poor definition and imaging (especially for the all-important vocal range), and strange dips in frequency response had converted excellent sound into merely mediocre - just as one might expect from any other theatre with an otherwise very good sound system.
After a (rather painful and frustrating) tour through the CP650 processor's menu system, the EQ settings were removed. All tone controls were set to flat, the subwoofer parametric EQ was disabled, and all equalisation for the left, centre and right speakers was returned to flat response. It is notable that each of these very important speakers had different EQ settings, even though their tonal balance is virtually identical without any EQ at all. Equalisation was also removed from the surround speakers, which although fairly ordinary have acceptable response for their purpose (and sounded better without EQ). The surrounds are actually the weak link in the system, but funds are not available to upgrade them.
After the equalisation was removed (other than correction for screen losses), the system was back to sounding the way it should. There have been a great many comments from patrons - including film professionals - that this particular cinema had the best sound of any theatre they have attended. There isn't a single seat in the theatre where the sound is too bright or too loud, and likewise no seat where dialogue isn't absolutely intelligible. In short, the sound was excellent at any seating position.
This installation showed that use of the X-Curve and extensive equalisation is not only unnecessary, it creates problems that didn't exist before. There is a sensible requirement that the speaker system should be properly aligned in its own right, but once this is done, attempting to apply room equalisation will do far more harm than good. For more information about the system itself, see the Lenard K4 theatre system - this is the basis of what we installed at the cinema.
The system itself is a 4-way fully active design, with all drivers horn loaded for maximum efficiency. Each loudspeaker driver has its own dedicated amplifier, and all amplifiers, crossovers and other system electronics were built by John and me. No part of the system uses off-the-shelf assemblies. The final system is relatively inexpensive, compared to purchasing all the equipment from normal industry outlets. This approach has the added benefit that individual sections can be tailored to suit their exact purpose, with a minimum of compromises.
In case you were wondering, John Burnett (Lenard Audio) and I have worked together on many projects, including the K4 in the cinema in question. The installed K4 system includes the ESP P125 4-way 24dB/octave crossover networks, P84 third octave bass equaliser and P127 power amplifiers, plus a P48 EAS subwoofer equaliser circuit to obtain sub-bass extension to 20Hz. There are also other parts of the system that were custom designed to provide additional functionality that is not found in other cinema systems. It is very pleasing to have worked on an installation such as the one John and I set up - I've not been to a movie theatre anywhere that sounds as good!
Unfortunately, the system has been de-commissioned and is no longer in use, as the cinema was bought out by another chain.
As is obvious from advertising material and theatre posters, the Dolby based systems are not the only ones available. Although there appears to be a huge amount of information on the Net, much of it is duplicated, and the majority is directed at home theatre rather than cinemas. Little detailed technical information is available unless one has access to the equipment.
Dolby SR, SRD, etc.
Dolby SR and SRD are just two of a whole family of formats. For more information see the Dolby website.
Although the THX® system uses a Dolby processor, it has different (and it would seem closely guarded) setup requirements. Lucasfilm (the creators of THX) will relieve you of a large sum of money to have your equipment and theatre certified as THX compliant, but it would seem that many cinemas will cheerfully claim to be certified, even though they haven't parted with a cent to have the work done. Much of the work needed to make a cinema THX compliant involves ensuring that minimum acoustic criteria are met, covering reverberation, sound transmission (through walls, floor and ceiling), ambient noise, etc. In my opinion, the standards set appear to be perfectly reasonable (from the few snippets I have been able to find). There is no reason for any new theatre to be built that does not comply, as it seems to be sensible acoustic design.
THX also insists on a minimum sound quality standard from all installed equipment - especially loudspeakers and crossover networks. That any installer would consider using anything less is cause for some concern, but there is a vicious circle effect ... if the sound is bad or mediocre, fewer patrons will attend. Fewer patrons means less income, making it hard to justify spending a lot of money on a good sound system. If the sound remains bad ... (and so it continues).
All in all, it would seem that the THX requirements are a very good starting point, and to refit an old theatre or build a new one without applying proper acoustic treatment and installing a decent sound system is a recipe for disaster. In some cases, old theatres may already have acceptable acoustics (not ideal perhaps, but acceptable), and little or nothing may be needed in addition to what's there. Few existing sound systems will be re-usable, but that depends on the age and general condition of the equipment. The financial burden ultimately decides what is possible, because motion picture theatres are no longer the "cash cow" they once were.
Sony's SDDS (Sony Dynamic Digital Sound ®) system uses its own proprietary processor, and like the Dolby system it has provision to equalise the loudspeakers. Not having played with one, I can't comment on the setup process in any real detail. The manual does provide a glimmer of hope though.
Because the SDDS system has provision to interface with the Dolby CP500 (and above) processors, in many cases the Dolby processor will maintain control over the overall system equalisation. This ensures that all films will sound as mediocre as each other if the standard setup is used.
The SDDS processor does have full equalisation built in, and it must be pointed out that the instructions are at odds with those from Dolby. In general, the instructions are in quite close agreement with my recommendations above. Sony rightly points out that you cannot equalise the room with its reverb and reflections, and suggests moving the measurement mic(s) closer to the speakers if measurement results appear wrong. It is also recommended that all speakers of the same type should use the same EQ settings - this is a very good start, and is consistent with reality.
Unfortunately, it is likely that install technicians used to the Dolby system will use the procedure they know, rather than follow the instructions.
Digital Theater Systems. The latest processor (XD10P at the time of writing) has full equalisation facilities, as well as all normal decoding facilities. I finally have access to the calibration procedure, and it seems to be reasonable - at least on the surface. Although the unit has an inbuilt RTA (real time analyser), it is suggested that this should only be used for quick checks. There is a complete section describing how to 'calibrate' the room using the in-built graphic equaliser, and the X-Curve is prominently displayed as the ideal response.
It does have the ability to measure RT60 reverb time as required for THX certification. Full 'calibration' is recommended to be performed with an external fully calibrated RTA and microphone. However, it is recommended that once the centre channel is 'aligned', the EQ should be copied to left and right - again, assuming that they are the same as should be the case. Likewise for the surround speakers.
There are also some suggestions for verifying that the speakers are up to the task, and this includes a proper listening test. Info is pretty sparse on exactly what to listen for, so I suspect it is assumed that the technician will have reasonably good knowledge of how a cinema system should sound.
There have been a number of other competing formats for digital and/or multi-channel audio for cinemas, but many of them have died because of lack of support or technical problems. I don't propose to even attempt to list them here, as they are not relevant to the current discussion - namely system equalisation.
That cinemas should sound consistent is beyond any doubt. That the original work of Ioan Allen (who joined Dolby in 1969 - see Interview) was ground-breaking is not disputed. Allen pushed the boundaries of cinema sound in many areas, and for the first time, there was a move to maintain some kind of standard so that cinema goers could expect to hear the dialogue clearly, and experience the movie more-or-less as was intended when it was dubbed, mixed or otherwise dealt with at the production sound stage.
For various reasons, the methodology used and decisions made were flawed - taking measurements in the reverberant field is pointless, and can only ever yield mediocre results at best. However, even mediocre is certainly better than 'really bad', or perhaps patrons complaining that "the sound was total crap!". In some cases, it is probable that mediocre was a huge leap forward.
When equalisation is used, it is illogical and obviously completely wrong that more or less identical speaker systems (left, centre and right) can (and usually will) have radically different EQ applied at the end of the 'calibration' process. If any EQ is needed at all, it should obviously be the same for all speakers of the same type. In addition, and perhaps most important of all, remember that ...
You cannot correct time with amplitude!
Reverberation is time related, and there is absolutely no form of equalisation that can be applied that will change it. None whatsoever! To continue with the pointless pretense that the process works just means that there will be no improvement of the sound quality in cinemas, regardless of further advances in loudspeaker performance.
The technology to make excellent sounding speakers has existed for many years, but loudspeaker driver and cabinet costs and the sheer size of the systems needed for a decent sized cinema mean that it becomes a very expensive exercise to outfit a modern multi-cinema complex with the best that can be made. However, these costs must also be put into context - a modern film is an extraordinarily expensive undertaking, and may only last a month or so at the box office.
The cinema (individual or complex) will be there for countless films, and a genuinely excellent sound system becomes just a comparatively small part of the overall setup or refurbishment cost. If properly designed, ongoing maintenance should be minimal - a well designed and implemented sound system can last for many, many years without a single failure.
Once the myth that the X-Curve is somehow a good idea can be finally laid to rest, and the silly reverberant-field 'room equalisation' nonsense is stopped, there is no reason at all to prevent the cinema from being all it can be. There are already a few people who advocate abandoning the X-Curve and all attempts at room EQ, but as yet they are a more or less silent minority.
Looking through websites and forum pages is instructive. Many explanations for the X-Curve are simply regurgitated from some other website, and some are almost identical to each other. Industry professionals on forum sites often ask questions that clearly show that they have no idea of what the X-Curve is, why it is used, and what it's supposed to do. Very few point out any of the serious deficiencies that have been described here.
The references cited here are just some that may be found on the Net, discussing cinema processing, equalisation (not just for theatres) and many other similar topics. These are the ones from which I made specific notes, but there are countless others that discuss the general principles.
|Copyright Notice. This article, including but not limited to all text and diagrams, is the intellectual property of Rod Elliott, and is Copyright © 2012. Reproduction or re-publication by any means whatsoever, whether electronic, mechanical or electro-mechanical, is strictly prohibited under International Copyright laws. The author (Rod Elliott) grants the reader the right to use this information for personal use only, and further allow that one (1) copy may be made for reference. Commercial use is prohibited without express written authorisation from Rod Elliott.|