Magazine Salon AV 2001 No. 10, translated from Russian
Seven words about audio expert errors
Leave them! They are the blind leaders of the blind, and if the blind lead the blind, both will fall into the pit.
The Gospel according to Matthew, chap. 15, v. 14
Only audioexpertiza is able to direct the efforts of developers of audio equipment in the right direction. All their wanderings, frequent visits to dead ends, as well as returns to previously rejected ideas occur, primarily because of the miscalculations of audio expert examination. Being in most cases untenable, audio expertise deprived designers of audio equipment of the correct landmarks, and therefore their creative activity is similar to the wandering of travelers in the forest without a compass.
Those who believe in the possibility of making perfect technique without audio expert examination, only improving its parameters, probably do not realize that these characteristics were established as a result of the listeners’ assessment of the visibility of certain distortions of sound, that is again audio expert examination.
As it is insulting to us, we have to admit that the objective parameters of the audio equipment are secondary to the data obtained, as it turned out, by unreliable audio expert examination. It is because of this that the audio expertise eventually ended up in the center of the Gordian knot of all the problems of the audio industry.
In this situation, highly experienced manufacturers, desperate to catch the “Firebird” sound quality all their strengths decided to spend on achieving a comfortable sound. That is, they followed the path of eradicating audible and even “visible” flaws in audio equipment, rightly believing that it is easier to proclaim its merits in advertising than to get it in reality. Approximately so, fortunately, for the present outside the CIS, have achieved the “consumer qualities” of strawberries. As a result of such a policy of the companies, music lovers and audiophiles lost their worthy audio equipment and, moreover, proved to be helpless hostages of shameless PR.
To save them from information violence can only free and truthful audio press, which, from my point of view, as well as the Ministry of Health, is designed to warn the reader about the danger to the health of his soul’s comfortable sounding of music, and also advise buying the audio equipment that is really the best.
To this end, I think the magazines publish expert conclusions on the new technology. But then why does audio publishing still feel some embarrassment towards the reader? A fatal role here, if, of course, do not take into account purely commercial reasons, the inconsistency of the audio expert examination that they conduct is playing.
In this regard, I am forced to bring to the readers sad news: it is not the absence of scientific knowledge that is the reason for the inadequacy of audio expertise, but the inexcusably crude methodological errors in its conduct. There are seven such grossest and most common mistakes. In the article, I intend to list them and give advice on how not to admit them in the future.
It happens that when comparing the sound of two amplifiers or an amplifier and a “jumper”, the audio experts can not notice any difference between them. Such a result does not at all mean that there is no such difference. The reason may be a low resolving power, or, as it is usually called, the “transparency” of the reproducing tract.
In the first approximation, the transparency of the path depends on the length of the path overridden by the musical signal. Its length, in the first approximation, is determined by the number of electrical components, among which this signal wanders. Simple arithmetic tells us that the path length of the signal in the playback path can be about 0.5 to 20% of the path length in the audio path. Precisely because of this disproportion, audio expertise is forced to capture the difference in the sound of reproducing devices, which is approximately 1% of the named path . Truly, it is almost the same as to feel the “pea” under a lot of featherbeds, the number of which, moreover, remains unknown. It is unknown, by the way, because recording companies never report how much electronics they managed to use when recording a particular record or CD. And one more aggravating circumstance: the transparency of the path depends on many other factors that are difficult to take into account, for example, the weakening of this ability is often explained by the use of negative feedback, electrolytic capacitors, transistors, and so on. It is impossible to take into account the influence of all factors on the transparency of the path, therefore it is best to determine experimentally by it.
To assess the transparency of audio systems P. Quartrup proposed to use a method, which he called a “comparison method by contrast.” To get an estimate on the compared audio systems, you need to listen to a lot of sound recordings of different styles made using different technologies, trying to catch which of the systems reveals more differences between them. In the author’s opinion, the more accurate is the system that transmits more contrasts between different records .
Quortrup’s method, indeed, allows you to choose a sample of the equipment with the highest resolution, but it does not give an opportunity to estimate the scale of audible differences. The reason for this metrological nonsense in the random selection of phonograms. It is because of this accident that the size of the contrasts subjectively perceived by listeners can not be determined.
And still: since the records are random, they contain musical works most often aesthetically incomparable. They are incomparable primarily in the timbre of sound, emotional and spiritual content of music, so the method proved to be suitable only for detecting contrasts of technical quality of the recording, but does not provide an opportunity to evaluate aesthetic contrasts. In other words, the method does not cover the most important signs of sound for the perception of music.
Of course, all this is very distressing, but it is not worth throwing off a beautiful idea, only because of its undeveloped nature. The method proposed by the Quortrup should only be slightly improved. For example, to take not random phonograms, but having a definite initial contrast among themselves. That is to do approximately the same as in its time taken for 100 ° C difference in the temperature of boiling water and melting ice.
A certain initial contrast of sound can be at home to anyone who takes a pair of LPs for comparative listening, which are made from one primary phonogram, that is, master tapes, but released at different times, say, with a break of approximately 10-15 years . It is known that during this period there was a complete technical re-equipment of recording studios, and therefore we can rightly expect a quite definite difference in the sound character of the selected record on the discs we compare. It remains only to agree on which firm to give preference to. In my opinion, for these purposes, the RCA Victor records from 1955 to 1962 are suitable, and in order to contrast with them, use the repeated issues (One issue includes the records produced with the first original of the matrix. The case when the records are produced from the newly made first original, that is, after being re-recorded on the lacquer disk of the old master tape) of these programs in the early 70’s.
Why do I recommend RCA Victor records, not other companies? Because during the designated period this company was the world leader in sound recording quality. They did them, in comparison with other firms, on the shortest path and only in natural acoustic conditions, using the minimum number of microphones – usually there were not more than three. We will conditionally consider the reached then RCA Victor level “boiling temperature” sound quality sound recordings.
By the 1970s (already under the RCA trademark), the firm managed to reach the level of medium, that is, mediocre sound recordings. This level, which, by the way, was kept approximately constant for 7 years, is recognized as the temperature of “melting ice” sounding phonograms.
So, if you choose as the initial contrast the difference in the quality of the recording on LP RCA Victor of the first and later releases, say, one of the violin concerts performed by Yasha Heifets (in the late 50’s recording), then comparing the sound of these LPs on the tested audio system , You should hear contrasts on many signs of sound. If these contrasts are perceived by you as stunning, then the audio system is acceptable for use as a control reproducing tract resolution.
Naturally, LP is not suitable in all cases of audio expert examination, for example, how to be with testing CD players? This is not the easiest question I will try to answer in a separate article.
Whatever assessment of sound quality of the equipment can be considered the most blatant mistake, and not only because our psyche is more adapted to compare than without regard to evaluation. Whichever evaluation of sound is unacceptable primarily because the tested, that is, the reproducing part of the apparatus by the number of electroelements is about 1% of the sound path. Even if an unbiased evaluation of the sound, someone gets it, it will be correlated first of all with the recording path, or in other words, with the quality of the sound recordings used in the examination.
In order to assess the experts’ correlations with the tested equipment, its sound must be compared with the sound of the reference sample (the Optical sample is considered as a reference point of subjective testing). For this, the test sample included in the reproducing control path must be periodically “replaced” by the reference one. Comparison of the sound of the test sample with the reference one is called paired . There is, however, in the method of paired comparisons of the Achilles heel. It is not clear which sample should be considered the supporting one. There are two approaches to his choice.
The first approach. Use as a reference sample a commercially available, and therefore many known for sounding sample of audio equipment . Such an average sample (of course, in a certain price category) is taken as the standard of sound quality, with which the experts determine whether the tested sample sounds better or worse. (In beauty contests, sound is produced by ay diodeapparatus if someone has conceived such a conduct as a standard sample, One should nominate one of the contenders for the prize place). In small developed countries such as Denmark, Sweden, etc., where literally every kilometer you can find a Hi-Fi store, and if you want even to listen to such a sample, this approach can be considered reasonable.
In Russia, where Hi-Fi stores are located at a distance of not less than 1000 km, the idea of a standard sample conflicts with the surrounding reality.
The second approach. To use as reference the reference sample of audio equipment , which is simultaneously a component of the control reproducing path involved in the audio expert survey.
In order to imagine what this path and its reference components should be, let us recall that the sound engineers make their recordings “tuned” to the sound through the studio control channel and even harmonize with it.
If we are ready to fully trust the sound engineer’s taste, then it remains to be accepted as an indisputable truth that the recordings made by him will sound best not on any super-expensive or considered “ideal” audio equipment, but on the one that looks most like a sound control track. Hence, it is the studio control path, in theory, should become a control and audio expert. But here’s a catch – different recording studios used different control channels. By the way, it is because of this, in the first place, were so not similar in sounding sound recording of different companies.
What if, during the audio expert examination, each controllable track requires its own control path? The way out, in my opinion, can be only one: to generalize the quality indicators of the studio control channels of the best recording companies, reducing them to the requirements, which should be answered by the path used for the audio expert examination. To begin with, since something is better than nothing, I propose to adhere to the following requirements for such a path:
Of course, when selecting the components of the control path, it is impossible to take into account everything. Special troubles bring unexpected effects of their harmonization. That’s why we have to accept the fact that different expert groups use different sounding control channels. The difference of the routes led us to a dangerous line, namely, to the violation of the requirements of the main metrological principle – the unity of measurements (Unity of measurements is when the results of measurements of the same object performed in different places and at different times within the permissible error remain the same) . Fortunately, the way out of this situation is partly found by the audio expert. He, as it turned out, takes into account the errors in the sound of not only the tested, but also the reference sample. The expert does this almost subconsciously, comparing the sound of the samples with the aesthetic standards saved in memory. For details on aesthetic standards, see paragraph 5 of the article.
Is it possible, after reading the testing protocols of an audio device, to judge its sound quality, if the estimates in them are incomplete, that is, cover not all aspects of sound? I think that such assessments are likely to cause distrust, as they will surely remind the reader of the story of the three blind men, each of whom, feeling only one part of the elephant, one – the leg, the other – the trunk, and the third – the tail, Completely wrong judgments about what it is.
The correct idea of the quality of the sound of the audio system can be made only if the evaluation is made according to all its main characteristics, and not only audible, but also those that are confused in the wilds of our subconscious. Such a list is given in my book, specifically devoted to testing of audio equipment . Here I only list the most important among them, that is, those signs, without an estimate, which is impossible to formulate an idea of the sound quality of the device under test. I refer to them:
Evaluation of the sound quality of the test apparatus will be incomplete if its resolution is not determined by the same characteristics.
The reader probably noted to himself that I propose to evaluate the quality of sound about the same way as experts do at international beauty contests, selecting applicants for prizes.
The results of the audio expert examination should be questionable whenever the music program used is made up of sound recordings accidentally tucked under the arm.
To what extent the correct selection of phonograms is important for us, it is not difficult to guess, if we recall that the transparency of the recording path is much smaller than that of the audio equipment being tested. That is, the recording path always dominates to a greater or lesser extent, “obscuring” the sound of the device under test. There is another problem: the inevitable harmonization of the tested samples of equipment with the selected material.
Many people do not feel the complexity of the problems I have raised, because huge and often different in the composition of the equipment used, recording tracks, like the genie, hide in small-sized records or CD.
And nevertheless there is a main requirement to the phonograms used – it is an opportunity, with their help, to convey to the experts those aspects of sound that are being tested. Even the best reproducing apparatus will not be able to “pull” out of the record what is not in it. Unfortunately, there are no discs, even test ones, in which there are all aspects of sound that interest us at once. Therefore, to conduct testing in full, you have to use a large number of sound recordings.
How to select them? First of all, the typicality of phonograms, that is, the lack of specific features of sound, should be taken into account. Harmonization of the reproduction path with atypical records can make sounding on it records of other firms unbalanced. And only in the second place it is necessary to take into account the quality of recording, the nature of music (the composition of musical instruments) and the level of musical performance.
For reasons that you will have to repeatedly return, I advise you to focus mainly on LP. Use CD audio expertise, of course, you can, but with great reservations, in addition, it is a separate topic for conversation.
First, let me remind you of the main advantage of LP. Their mass production has long ended, so no surprises from them can not be expected. Although the LP recordings were often not the same, the differences between these experienced audio experts were well studied, which means that they can be easily taken into account during listening, unlike CDs, whose sound character is often unpredictable.
Typical, that is, having a certain average balance of sound recording, produced before 1970 RCA Victor, Columbia , EMI, Decca , Verve , and according to the tonal balance, the most perfect were the Deutsche Grammophon disks. All the firms I listed in the period from 1955 to 1962 did really wonderful records. Then Atypical released then Atlantic , ECM, Philips , Blue Note , etc.
On the frequently asked question, which music or which genre is more suitable for testing, I will say something like this: it does not matter how music is called, it is important how fully the musical means are presented in relation to the transmission of which through audio equipment is tested. In the process of audiosexpertation, records of singing voices, including the sound of the choir, as well as natural musical instruments, such as violin, cello, double bass, piano, timpani, harpsichord, jazz saucer, flute or clarinet and, of course, ensembles (chamber and orchestral) .
When assessing the quality of transmission through the equipment of musical emotions and spiritual content, one should use the records of the greatest musicians who managed to capture their game on the early LP. I mean Heifetz, Milstein , Schering , Michelangeli and Gould .
Organizers audio experts are mistaken when they are invited to work as professionals just hearing young guys. Even more they are mistaken when they believe that experts can be professional musicians or, worse still, psychophysical scientists. The organizers of the examination often do not take into account the fact that the audio expert is a completely independent profession. In addition, he must have an extremely thin ear, auditory memory, musical sensitivity and, most importantly, what is usually not enough for modern musicians, and especially scientists, is a musical culture and artistic taste.
Many people have a thin ear from birth. Due to this ability audiophiles distinguish the sound of network cables, equipment stands, etc. Those who have such a delicate rumor, it is customary to call “golden ears”. The fineness of the hearing is associated with the resolution of the audio equipment in much the same way as the visual acuity with the increase of the microscope.
However, to have a fine hearing for an audio expert is not enough. The expert should hear not only the difference of the compared sounds, but also determine which of them is better, and for each of its features. The psychological mechanism of preferences is based on the well-known method of triads . In accordance with this method, the subjective preference of one stimulus to another occurs when one of them is more similar to the third, the so-called reference one. The audio expert should use the aesthetic standards of sound that have developed in memory .
What could be simpler than this method? However, it was the formation of the right aesthetic standards among audio experts that proved to be the most difficult thing. If you think about it, it’s about educating their initially missing musical culture and artistic taste. Because of the complexity and, most importantly, the length of the process of such education, the role of aesthetic standards by the organizers of audio expert examination was practically leveled. As a result, manufacturers of audio equipment and audio editions have for some time lost the understanding of what, in fact, should be striven for. This, as well as the commercialization of questions related to sound quality, is the main reason why the “goldfish” eventually took millions of music from musicians, and instead left a broken trough called High End .
Descriptions of sounding ay diodeapparatury, usually published in audio editions , readers perceive as having art criticism gibberish. And all because you can not understand the description of sound in “free”, maybe even the perfect literary language. This is explained simply: the language of music is not translated into other languages .
But in this matter, nature partly confused us with maps. It turned out that there are some word-descriptions that readers associate with a completely certain nature of the sound. These words, the so-called associative adjectives, we really understand, because in our brain there was once a rigid associative connection between certain characteristics of sounding and some visual and tactile sensations. Such descriptive adjectives, existing in the form of polar pairs, found about 200 . Here are some of them: “high – low”, “light – dark”, “bright – dim”, “warm – cold”, “smooth – rough”, etc. It was these words that misled audio experts, forcing them to believe that they are able to talk about what they hear without difficulty.
Unfortunately, in order to make a full description of the sound, associative adjectives are not enough. Therefore, half a century ago, the acoustics of the old school reflected on the creation of a special dictionary covering all aspects of the sound of the dictionary. Theoretically, such a dictionary can be compiled. It is only necessary to stipulate in advance the contents of all the terms used in it, rigidly tying them to all the most important auditory sensations from sounding. The compilation of such a dictionary involved many: the first was L. Beranek , later – JG Holt , P. Quortrup, AES and even your humble servant .
If you ever compare all the dictionaries created in half a century, then, most likely, you will find yourself in difficulty, because you can not prefer one to another. To understand this issue, I offer readers a clue in the form of a riddle. The best among the dictionaries is the one with which you can most fully express the outline of the “elephant”.
The organizers of audio experts usually ignore the acoustic conditions in which the AU equipment is tested, and often do not even mention them in their reports. But testing is always done in closed rooms where the energy of the sound field near the ears of the listener is 70% composed of reflected sounds and only 30% of the energy of direct sound. Even if one does not delve into the specifics of the distortion of the music signal in the room, the mere fact that we are listening to it, mainly the reverberation field, should make us count the room as an equal link in the reproducing path.
It has long been known that the listening room, even to a greater extent than other links in the tract, changes the sound, and especially with respect to such signs of signs as:
Tonal balance (only speaker placement can change the frequency response at low frequencies within +12 dB).
Clarity ( articulation and connected sounding change, which affects the perception of the emotional and spiritual content of music).
Due to the ignoring of the acoustic conditions by the administration of audio testing, the rooms in which this testing is usually performed are chosen among those who did not manage to occupy other purposes. And this means that different groups of experts are forced to listen to audio equipment in unequal rooms, not similar in form, volume, not equally drowned , etc. Even in the first approximation, the rooms known to experts do not meet the requirements for the standard listening room recommended by IEC 268-13 .
Different acoustic conditions in the conduct of audiosexpertiza come into conflict with the principle of the uniformity of measurements and therefore cause a final blow to the reliability of its results.