Distortion and Subjective Audio Quality
Mark Irwin, The London College of Music
Abstract.
The audio industry has always relied on the 'golden ears' of record producers and engineers.
Despite the industry moving almost exclusively to digital platforms there has recently been a renaissance in the manufacture and sales of 'classic' styled Valve and early transistor hardware, particularly at the 'front end' of the recording chain (Microphones and microphone amplifiers) and signal processors. There has also been a large number of re-issues of landmark units as well as many digital emulations (as 'plug-ins') for computer based recording platforms. According to conventional technical specifications these 'classic' devices should not perform as well as modern 'chip' based technology.
However, their popularity amongst professionals and the growing 'pro-sumer' and hobby markets indicates that perhaps there are qualities that the technical specifications are not measuring - or are we just being fooled by clever marketing and audio mythology into believing that these machines can really add that elusive 'fairy dust' to our recordings.
This paper seeks to gather the thoughts of a wide variety of end users as well as audio designers and academics.
Research Methodology.
My approach to researching this essay is not primarily scientific, as there is already a wealth of technical and scientific research into the design of audio hardware and the measurement of distortion. I have looked at the recent scientific research into the subject, including the papers "Auditory Perception of Nonlinear Distortion"1 by Earl R. Geddes et al and "The Effect of Nonlinear Distortion on the Perceived Quality of Music and speech Signals"2 by Chin-Tuan Tan et al. This body of work proposes new scientific methods for measuring distortion in audio. However, my argument assumes that what we need to measure here may be far more subjective in nature, and, although new ways of measuring distortion and its effect on the listener are essential, the real issues for audio production are far more complex.
I will also avoid re-addressing the well worn and documented 'Analogue vs. Digital' debate. I have asked for a preference from my sources, but as Nick Howes Head of R&D of Yamaha states: "It all ends up on CD or DVD or MP3 anyway, so who cares...". There are, however, practitioners who still prefer the sound of analogue tape for its particular brand of distortion and evidence that analogue recordings are imparted with qualities unique to the medium. Neither shall I look in detail at thorny issues such as digital Jitter, again these are subjects that have been thoroughly researched and debated.
My research methodologies are twofold, consisting first of a review of existing literature and, secondly, of a series of interviews with producers, engineers and audio designers. I have also questioned students on music technology courses here at Westminster University and at Thames Valley University, as well as academic staff and technicians. Some of these interviews were conducted face-to- face, whilst others were via e-mail. All revolved around twelve questions as laid out below.
Distortion & Subjective Audio Quality.
1. Do you prefer Valve or Integrated Circuit outboard/consoles/microphones?
2. What is the reasoning behind your preferred choice of hardware?
3. Do you prefer analogue or digital recording and why?
4. Do you think that modern re-issues of landmark audio equipment sound the same, if not what is the foremost difference between these units?
5. Do you feel that digital emulations (Plug ins) of classic outboard sound the same as the originals?
6. Do you prefer digital or analogue equalisation and why?
7. Do you have a favourite compressor? If so what is it and why do you favour this particular unit?
8. What do you feel creates 'warmth' from an audio processor?
9. Do you believe in 'musical distortion'?
10. Do you trust your ears or the technical specification?
11. Do you worry about signal levels when mixing/mastering and do you feel that some modern records are often mastered at too high a level?
12. Can you hear a difference between Analogue to Digital & Digital to Analogue converters, if so do you have a preference as to what you use?
Contents:
I have broken down this essay into the following sections based on the main areas covered by the questions listed above:
* Valves vs. Solid State, distortion & warmth
* Equalisation, compression & plug ins, re-issues
* Analogue & digital
* Mastering & broadcast
* Ears vs. technical specifications
* Conclusions
* Bibliography/interviews/respondents
Valves vs. Solid State, distortion & warmth.
What is distortion in audio recording?
In their paper The Effect of Nonlinear Distortion on the Perceived Quality of Music and Speech Signals3, Chin-Tuan Tan et al. defined the two broad types of distortion produced in audio:
" 1) Linear distortion. This involves changes in the relative amplitudes and phases of the frequency components present in the complex signal. Such changes are typically perceived as changes in timbre or tone quality (colouration). In principle, distortion of this type can be compensated (within limits) by linear filtering, and such filtering can be applied either before or after the transducer or transmission channel whose properties are being studied.
2) Nonlinear distortion. This involves the introduction of frequency components that were not present in the input signal. The effects of nonlinear distortion are difficult or impossible to compensate by subsequent processing. Signals subjected to nonlinear distortion are often perceived as distorted. It is perhaps unfortunate that the same word is used to describe both the physical process and the subjective impression. The effects of nonlinear distortion may be described as harshness or roughness or in terms of the perception of sounds that were not present in the original signal such as crackles or clicks."
It is widely understood that the human brain does not respond to sound in a linear way and that capturing acoustic events accurately but -perhaps more importantly- convincingly is a complex task. A great deal of research has been conducted into the nonlinear nature of human hearing. The human hearing mechanism clearly produces both linear and nonlinear distortion. For example it has been shown that we still perceive a fundamental tone even when it is missing and all is present is a series of related harmonics. Human hearing also produces linear distortion; frequency response is dependant on level and is far from flat at any volume; therefore any audio processor that measures a technically 'flat' frequency response is unlikely to sound 'flat', or even, to a human operator. There is of course a more complex physiological/emotional response involved. Past good and bad experiences and an emotional attachment to certain technology may also be factors that affect our choice of equipment.
It is widely recognised that, in addition to the nonlinear nature of hearing, the audio equipment we use also adds nonlinearities to the signal.
Linear distortion is characteristic of many devices that audio engineers use because of the unique 'tone'. Valve, and optical compressors or tape saturation are examples of this.
However, nonlinear distortion may also have subtle and pleasant effects if it is of the right kind. As audio professionals the questions are: how do we choose and what do we choose?
Valves or Solid State?
Edison invented the vacuum tube, or Valve in 18834 as a product of his work on the development of the electric light bulb. John Ambrose Flemming experimented with this early Valve in 1904 and produced the 'Flemming diode'. This device was a rectifier, which had the ability to convert voltages from alternating current to direct current but did not amplify the signal.
American scientist Lee de Forest further developed this work with his invention of the first 'triode' Valve, which could act as an amplifier, and patented the 'Audion Tube' in 1907.
The first transistors were produced during and after World War Two as the result of work by William Shockley, John Barden, and Walter Brattain at Bell Laboratories in the USA. These transistors were first patented in 1947/48 and could act as both rectifiers and amplifiers.
Transistors had obvious advantages over Valve's or tubes. They were smaller, cheaper to produce, more efficient, less prone to mechanical damage. They also produced far less heat.
Early transistor amplifiers were certainly inferior to Valve devices, but, as designers improved and developed their designs, transistors as the basis for solid-state circuits began to dominate the market.
By the 1980's Valves had become almost obsolete, except in niche markets such as guitar amplifiers and audiophile Hi-Fi systems and some areas of specialist studio outboard.
However, despite this market dominance Valves seem to be making a comeback, particularly in the worlds of professional audio and audiophile Hi-Fi design. This leads one to question what it is about this early technology that draws practitioners to these particular devices.
What is musical distortion?
It is widely understood that some combinations of sound are pleasing to human hearing. Musically speaking, some sounds create a basic tone, or 'fundamental', while others ('harmonics') define the musical instrument itself and give it its own unique 'sonic signature'.
I would argue that analogue processors in particular also impart harmonics that can be pleasant. When creatively applied, these offer distinctive additions to the audio producer's art.
In his seminal paper on the subject entitled "Tubes Versus Transistors - Is There an Audible Difference?"5 Russell O Hamm states that: "...anyone who listens to phonograph records closely can tell that tubes sound different from transistors. Defining what the difference is, however, is a complex phsychoacoustical problem."
In his paper he presents meticulous scientific research into what the difference is between, Valves, Transistors, and Operational Amplifiers.
His argument hinges around the different way that each of these devices distort and how this affects the 'musicality' of the listening experience.
He states that, whilst operated in their linear range with harmonic distortion of 10% or less, all of these devices will indeed have very similar characteristics.
However he notes that this is rarely the case and that amplifier's, particularly pre-amplifiers (microphone amplifiers) are often operated far outside this range. Under these circumstances he shows that Valves, Transistors and Operational Amplifiers behave in markedly different ways.
To summarise his results on harmonic distortion:
Valves produce strong 2nd and 3rd harmonics alongside 4th and 5th harmonics at a lower level.
Transistors are distinguished in the production of strong 3rd harmonic distortion with all other harmonics being present but at a greatly reduced level. Hamm also found that these amplifiers produce square wave symmetrical clipping when overdriven.
Operational Amplifiers have a very similar response to Transistors, with the added disadvantage of a complete suppression of even harmonics.
Hamm proceed to relate these results to the musicality of the end result, and states:6 "there is a close parallel here between electronic distortion and musical tone colouration that is the real key to why tubes and transistors sound different"
This is not unknown to musicians and acousticians; the character of an acoustic instrument is defined by its harmonics. Dominant odd harmonics (3&5) produce a 'closed' or 'stopped' sound. The addition of the 4th and 5th harmonic changes this to a more open or 'brassy' sound. Harmonics above the 7th add 'edge' or 'bite' and an excessive amount of these upper harmonics will produce a 'harsh' or dissonant tone.
As Dr Harvey Rosenberg of New York audio Labs also adds:7 " Transistors produce much lower total harmonic distortion than tubes, but the type of high order/odd order distortion produced is so irritating, so musically unnatural, that when we compare this to the larger amount, but more benign, even order harmonics of tube circuits, we, the musically refined reach the conclusion that tubes make more mistakes, but the right kind, while the transistor makes less [sic] total mistakes, of a fatal kind."
At this point one should mention, despite the weight of evidence that this argument is not as black and white as it seems. Much work has been done by designers to produce transistor amplifiers that create a sound much closer to that of a Valve device.
In his article for dB magazine Eric K Pritchard8 discusses the development of Transistor based 'tube emulators', which may be the way forward. He agrees with many of my other sources and states that: " the tube sound is often more subtle than the oscilloscope display. Low-level harmonics are not visible, but are audible. Distortion meters do not consider individual harmonics, but our ears do. Standard audio tests do not tell the whole story."
Therefore, it seems that there is 'fairy dust', and it exists in the way that harmonic distortion is produced when audio circuits are driven hard.
It is also obvious from my research that many practitioners see the distortion as a natural and desirable part of the process. As one academic and musician, Simon Zagorski-Thomas puts it: "the aim of the recording process is the creation and manipulation of musical distortion. I don't believe in 'transparent' recording. There's no such thing as a 'true' sound, whether recorded or otherwise. Even without the recording process, the listener's position and surroundings determine what's heard. The term 'distortion' implies there was something 'pure' in the first place. If the recording process is one of manipulation rather than reproduction then it can be viewed as a series of choices about which types of distortion are desirable and which aren't."
It is also clear from answers to my questionnaire that it is the quality of design that is important. As audio hardware designer Johnny Reckless states: "I could write you an essay about the fact that it is not Valve versus solid state that matters; rather the design philosophy, signal integrity, circuit topology and details of implementation. A good solid state design is vastly preferable to a poor Valve design and visa versa." The audio designer Ted Fletcher agrees: "Both Valve and integrated circuit systems are fine and acceptable provided that they are properly designed and used correctly. There should be no difference in normal use, only misuse. It is allowable to misuse Valves in microphones say, to create pleasant distortion."
However, in Hugh Robjohns article 'Tubular Tales' in Sound On Sound9 it is clear that some designers are not so happy with solid state devices in audio: " Vic Keary, who argues very coherently that every stage of the audio chain - amplification, compression, mixing and equalisation - all took a step backwards when solid- state technology became commonplace in the late 1960's. Consequently, none of his products have transistors or IC's in their signal path at all, and the few solid state components that are used are restricted to serving the power supplies - where they have distinct practical advantages."
As a past owner of one of Vic's legendary 'Chiswick Reach' compressors and an early version of his 'Culture Vulture' - Valve distortion unit I certainly must acknowledge his extensive understanding of Valve technology and audio design.
Valves still have their supporters and the ranks seem to be growing but John Watkinson10 is still unimpressed, though open minded on the subject: "Valve microphones are making a comeback. I don't mind. It's nice to have a piece of audio equipment to keep your hands warm. Also the heat from the Valve keeps the condenser capsule nice and dry. Do they sound better? Under some circumstances maybe they do."
Phase.
Of course distortion is not purely a one-dimensional effect. Sound consists of an event in time as well as space. It is widely understood that phase is an issue in recording studios particularly in terms of recording, equalisation, and monitoring. Ted Fletcher relates this to the question of audio warmth: "warmth' in an audio signal is usually a particular juxtaposition of phase relationships across the audio spectrum; it is personified by slight but smooth phase lag at both ends of the audio spectrum. More particularly, what causes 'lack of warmth' is any incidence of sharp phase shift that has been artificially introduced, whether by design (with effects) or accidentally by poorly designed equipment."
This would suggest that 'warmth' is not merely a question of adding harmonic distortion; rather, that some of this distortion exists in the time domain, and therefore the 'shape' of the distortion envelope may be as important as its contents. This definition would also place 'warmth' in the linear distortion category.
This line of thought relates well with accepted thinking on the importance of phase in equalisation and mixing theory.
Our spatial (or stereo) hearing mechanism relies on two distinct Head Related Transfer Functions (HRTF's): Inter-aural Timing Differences (ITD's), which rely on the phase lag between signals to localise sound, and Inter-aural Intensity Differences (IID's), which rely on the level differences between signals in order to locate audio events.
Both the mechanisms are frequency specific, with ITD's working better below 500Hz and IID's at frequencies of above 1kHz. There is obviously a hole in response at around these frequencies and some overlap at the cross over points. Although HRTF essentially defines the specialisation of human sound decoding it also demonstrates why both phase and frequency response are important in producing compelling and believable audio. Distortions in phase response, will clearly lead, too less realistic reproduction in the audio chain.
It seems that the evidence shows that there is a positive advantage to high quality Valve or solid-state circuits in pre-amplifiers and microphones and that Valve designs can be guaranteed to add predicable and pleasant colouration to the recording chain. It was also widely agreed that in the case of guitar amplifiers, Valve designs were still superior to solid state ones, largely for their unmatched ability to produce dynamic and musical distortion.
The majority opinion from those questioned was that 'classic' and modern Valve or high quality transistor microphones, pre-amplifiers, consoles, and outboard devices do add important musical distortions and therefore continue to be important production tools. Several sources felt that 'analogue' devices also provide a more 'musical' and forgiving 'front end' in the digital recording chain. There is still a feeling amongst practitioners in particular that digital recording can produce a 'sterile' sound. This belief seems to be supported in the results of my research into digital equalisation, digital processors can be built to be accurate and theoretically they can also be completely transparent. However it seems that 'transparent' is not what users want in most circumstances. Audio producers prefer tools with colouration (of the right kind) that can be relied on to do their job and are easy to use.
To use a fairly obvious artistic analogy, the producer relies on a palette of sounds in his production paint box, clearly an individual producers style or signature is partly dependant on his or hers' choice of tones.
There was also a strong belief amongst several respondents. Particularly those from, the design side of the audio industry. That it is simply a matter of time before digital emulations can provide the complexity and warmth of analogue units.
Equalisation, compression, plug-ins & re-issues.
Equalisation
Equalisation was essentially invented as a technical sonic correction device. To overcome the poor frequency response exhibited by early microphones and to even out distortions from disk cutting lathe heads. Early equalisers used Valve circuits although the bulk of units were designed after the advent of the transistor and are therefore solid state. The majority of surviving large format consoles were also designed during this period and it was for these consoles that a great deal of research and design, into equalisation design, was conducted.
Equalisers or 'tone' controls developed from simple fixed frequency, fixed bandwidth or cue width (and shape) to the modern multi-band parametric equalisers. This is the professional standard for analogue equalisers today.
Any discussion on equalisation has to include the great British console designer Rupert Neve. Sadly he seems to have retired in the USA and is not easy to get in touch with. I do have some quotes on the subject:11 "in 1977 Geoff Emmerick showed me that he could hear the difference between two identical channels on a recently delivered new console. After some hours of listening with him I agreed that I could hear a subtle difference. When we measured it I found that out of 48 channels, three had been incorrectly terminated and displayed a rise of 3dB at 54kHz".
I also remember, attending a lecture by the legendary designer, who demonstrated, how very high frequency sound effects what we hear.
Neve summarises it as:12 " It seems that frequencies well outside of the normally accepted limits of human hearing can be perceived."
Neve built his equipment with huge bandwidth. His consoles and outboard are still highly sought after. His design philosophy, would suggest that high bandwidth, and high quality design is an important issue in both analogue and digital hardware. This is supported in the answers I received. Many practitioners praising the quality of high-resolution digital recording.
The central question here is, as to whether more traditional analogue equalisers and QED, consoles are preferable over digital ones.
Digital devices offer a huge advantage in order of functionality, flexibility, and features. However, it seems that the analogue equaliser, still has its place. There seems to be a strong view that, poorly designed digital equalisers are an issue in modern audio hardware and software. Michael Kemp from Sintefex states:13 "an analogue equaliser has, as part of its complexity, non- linearities that are habitually exploited by recording engineers... the prime limitation of digital audio...is that its effects remain simple. It is from the complexity of analogue processing that much of its musicality derives."
In his article Kemp discusses the differences between analogue and digital equalisers. Analogue equalisers, are, in his opinion never perfect and suffer from complex non-liberalities. So much so that even inserting an analogue equaliser 'flat' is noticeable.
Digital equalisers, however, are far more transparent in use. Kemp argues that this leads to operators, over applying digital equalisers, in a vain attempt to create the complexity, and richness, of an analogue equaliser. He also believes that this also occurs with digital compression devices.
This of course will result in very poor sonic results and Kemp states that simple digital equalisers suffer from:14 " distortion in frequency response as the frequencies processed tend towards the Nyquist limit...the digital version reaches its limit at 22kHz in an increasing crush - like the bonnet of a car hitting a brick wall."
Kemp believes that the majority of simple digital equalisation algorithms suffer from un-musical distortions:15 "Digital audio will only start sounding good when the digital processing starts to get as complex as analogue processing always has been."
He makes a strong case for better design, and higher sample rates in digital equalisation. Kemp also suggests that with enough processing power, devices can be designed to convincingly re-create the compound effects created by analogue circuits. He also sees racks full of outboard processors, being replaced by one programmable digital system.
The overwhelming opinion, from the majority of sources I questioned, is that well designed analogue equalisers have distinct advantages in studio work. Many of those, who championed digital recording platforms, also had a preferred analogue equaliser(s).
Johnny Reckless states that: "Many early digital EQ's did not handle coefficient sensitivity; truncation and dither correctly and thus gained a reputation for a harsh sound. This is was because they were poorly designed - there is no implicit reason why digital EQ should not be transparent."
From this point of view, it occurs to me that different types of equaliser are suitable, for different tasks. Specifically designed precision digital equalisers for their transparency (if properly designed) and others modelled to match the complexity of analogue equalisers. The recent release of Focusrites' 'Liquid Channel' would seem to be a clear move in this direction:16 "The Liquid Channel promises to be a producer's dream, providing accurate renditions of a wide range of hard-to-find or unaffordable preamps and compressors, in an easy-to-use format and with respectable settings."
It would seem however that analogue equalisers can impart a distinctive sound, and will no doubt continue to be used.
In the replies to my questions, practitioners often sited analogue equalisers as being their first choice where available. Console choice, where stated was often pragmatic but digital consoles were often referred to as 'not bad', considering its digital' or as doing 'A' or 'B' well but no good at; 'C'. 'C' would often be equalisation and dynamics. Virtually none of the users surveyed specified a budget, or mid range digital console as their platform of choice.
As Pip Williams sums the subject up: "...The sound is generally sweeter and more musical...I find myself using more cut or boost with digital equalisers."
Plug-Ins
The overwhelming weight of opinion, as to the ability of software emulations to match up with classic hardware, is that the majority of plug-ins do not sound the same as their namesakes. Those from all sides of the industry shared this opinion. More importantly this view was held by those who worked almost exclusively in the digital domain, as well as those who preferred to work with analogue hardware, or recording platforms.
Many of those questioned cited at least one particular plug-in that they felt worked well. Although this choice seemed to be a personal one with a wide range of choices being made. It should be said that this research was limited to processing (equalisation and dynamics mostly). Rather than classic keyboard, and synthesiser emulations. Which in my, and others opinion, do seem to provide very convincing emulations, of the originals.
The defining feature of this debate, seems to be the complexity of the programming required to emulate analogue devices, and the level of understanding from the programmer of the way that these 'classic' units operate.
As Ted Fletcher states: "in most cases, digital emulations cannot emulate analogue equipment properly for the simple reason that any renowned equipment has highly complex effects on the sound, many of which are outside the knowledge of the digital programmer."
Nick Howes adds:" not even close. To untrained ears, maybe, to anyone who uses the originals, no. However we are developing virtual circuit modelling which does come close."
Most practitioners felt that high-end plug-ins are of an acceptable quality, but few felt that they came anywhere close to, convincingly recreating the sound of the original devices.
Music producer Mike Howlett says of re-issues and plug-ins: "some are good and function in a similar way, but generally most fail to replicate the 'colour' of the originals."
My own personal experience, having worked in a large format analogue studio for several years, and latterly as a ProTools convert has convinced me that equalisation and dynamics processing, are the weak links in the digital chain. However, I am increasingly convinced by the quality of acoustically sampled reverbs such as the very impressive Altiverb( and as mentioned above, the rise in quality of soft synthesisers and classic keyboards.
It seems that the answer is better, leaner programming and much more DSP. After all many of us have moved from studios with multiple dedicated processors, all doing fairly simple tasks, to studios that often rely on one big multi-tasking processor(s). It is hardly surprising, under these conditions, that we cannot produce the audio complexity, the richness of auditory experience we have become used to.
Compression
Compression has always been a somewhat arcane and often misunderstood technique for dynamic control in audio. My own experience in teaching music technology has demonstrated to me how often this simple subject is misinterpreted.
Compressors evolved out of early 'Limiting Amplifiers' devices designed to dynamically limit levels to tape (early tape formulations did not handle 'hot' signals well and therefore operating levels were strictly adhered to). These units (Valve and transistor) were 'hard limiters' with few controls. Soon more flexible and sophisticated devices, allowing a more subtle dynamic control, began to be developed.
Early pop producers, such as Joe Meek did much to make compression 'fashionable' as an audio effect. Through ought the history of audio production compression has been seen as one of the foremost (alongside equalisation) sound shaping tools.
Many of my respondents named a favourite hardware or software compressor(s) including:
* UREI 1176
* Joe Meek (SC2 & VC1) TFPRO P8
* WAVES C1
* Bomb Factory LA2
* DBX 1066, 160, 1960
* Aphex Compellor
* SPL
* Focusrite
* Fairchild
The majority of sources, chose at least one hardware compressor as a favourite with the 1176 still a firm leader. Again, there is a strong indication that a producer or engineers choice of compressor, is largely based on the signature musical colouration it imparts. This is also balanced with a demand for sonic clarity and quality, alongside that for ease of use.
It would seem that most choices are often made pragmatically. Dependent on the work in hand and the resources available.
Producer, and arranger Pip Williams thoughts on compressor choice are pragmatic: "I love the old Fairchild's, which fatten the sound of anything and are excellent over whole mixes.
I also really like the Drawmer 1960, which is superb on bass and vocals. I tend to always use one on lead vocals. It provides good control without hearing it working too much.
I think the Focusrite Compounder is a very good budget unit- ideal for a home studio where cost is important."
Re-issues
Re-issue devices, also often failed to live up to expectations, with a general feeling that they failed to sound as complex or quirky as the originals. As Johnny Reckless states: " the major difference in re-issues is probably because the discrete components from which they were built are no longer available. Modern re-issues ought to be more consistent, due to modern materials and production methods. Perhaps the variation and hand built unpredictability of the originals is part of their charm?"
However, not all the experts feel that 'classic' hardware is always a better choice. In his article, They don't make 'em like they used to, John Watkinson17 argues for a more balanced view. That colouration is a matter of taste and application: "some people suggest things were better in the old days; that somehow the equipment was better built and maybe even sounded better. Certainly the survival, or even renaissance, of Valve-equipped devices would suggest that at least some hold those views. I'm not so sure such things are automatically better, but nor would I refuse to use them in the appropriate circumstances."
The argument again comes back to the job in hand. There is a clear choice being made between musical colouration, from analogue devices, and a more transparent sound from modern re-issues and digital emulations.
Johnny Reckless paints a vision of future developments: "software models are (of necessity) simplifications of the performance of their analogue counterparts to limit processor cycles. Given sufficient time and processor power a digital plug-in should be indistinguishable from its analogue prototype. And one day of course it will be."
It seems that analogue processors still have a place in dynamic control, that is until digital hardware/software platforms can support much more complex programming.
When asked why, re-issues sound different, Andy East also adds: "the obvious reason is the construction methods used. For example, the re-issue Vox AC30 sounds close but not identical. The modern construction methods are completely different because of labour costs. Old Vox AC30's built in the sixties were hand wired on tag strip, now they use P.C.B's which often cause earth loops, impedance and resistance changes in the circuit. Also and equally important, the tolerance of the components used in the sixties could be as much as +/- 25%. Hence why so few AC30's sound the same. Today's tolerance levels in components such as resistors and capacitors are as little as +/- 2%. That's a huge difference. It's also worth bearing in mind that many of these circuits also rely on transformer coupling; this also has a major effect on the sound."
Analogue & digital
A short history of tape
Analogue tape recording, dominated the audio production industry for decades. In simplistic terms, analogue recording, revolves around recording sound onto a rusty plastic ribbon. Although contemporary 'high end' machines, using modern high-level recording tapes, achieve technical specifications very close to, and exceeding that of some digital platforms. Tape is still inherently unstable and non-linear in response. Tape is also far too linear when it comes to editing and copying, unreliable in use, and long-term storage, (*The jury may still be out on this, as some digital platforms are yet to be proved as far as longevity is concerned) and expensive in comparison to modern digital media.
The flexibility, and functionality of digital has led to analogue machines being all but replaced for most studio duties. Although there are, a hardcore of engineers in production and mastering, still championing the attributes of 1/2 inch, two-track tape for mixing and mastering
In 1878 Oberlin Smith (USA) comes up with the idea of magnetic recording on to wire. In 1898 Valdemar Poulsen (Denmark) 're-discovers' the idea of magnetic recording and patents the 'Telegraphone' the first ever answer-machine. This was followed by work at Bell labs in the 1930's and the development of steel tape recorders by Marconi and the BBC in 1931/2.
In 1932 AEG bought patents from Fritz Pfleumer (Germany) for paper backed magnetic tape. AEG went on to market the 'Magnetephone' recorder, recording the London Philharmonic. In 1945, USA and UK, investigators discover the Magnetephone in former German territory. The USA government seizes the patents.
In 1945 Former serviceman JT Mullin demonstrates the AEG machine, and Bing Crosby hires him to record his work for TV. Up until this point all broadcasts were live, as producers did not like the quality of 16-inch 33/1/3 RPM discs.
From 1945, Magnecord Corp start making wire recorders, and then move on to tape in 1949 with the PT-6. In 1946-47 Brush, introduce the 'Soundmirror' paper tape recorder and Amplifier Corp bring out the Magnephone.
In 1947, 3M bring out paper and plastic based metal oxide tapes - type 111 becomes the industry standard.
In 1948 Ampex, bring out the Model 200, 3M produce Scotch 111 gamma ferric oxide tape.
In 1949-50, Magnecord introduce experimental Stereo recorders and in 1948-49 Sony begins to produce recorders. From 1956 Ampex produce the first commercial Video Tape Recorder the VR1000 (later versions of these machines were used as 2 track digital audio recorders).
In 1958, based on Alan Blumleins patent Stereo LP's first appear on the RCA-Victor label. RCA introduces stereo tape in a cartridge format, which flops.
In 1962-64 Phillips introduce the stereo cassette.
From the Mid 1960's, 3 and 4 track 1" recorders developed. In 1966 the Beach Boy's Pet Sounds landmark multi-track album was released, and the Beatles Sgt Peppers landmark multi-track album followed by 1967.
In the late 1960's early to mid 1970's eight and 16 track two-inch recorders were developed. Eventually twenty-four track two inch machines developed and became the industry standard for music recording.
The first digital tape recorders arrived in professional recording studios in March 1978 with the advent of Sony's PCM-1600 machine. In 1982 the digital format was extended to the consumer market, with the release of the first Compact Discs.
Despite this analogue tape recording remained a competing professional standard for multi-tracking right up until recently. When hard disk recording became reliable, and relatively cheap. Today, perfectly serviceable, and very expensive analogue recorders, sit gathering dust in recording studio corridors, to make room for ProTools HD rigs. There is good reason for this. The advantages of a non-linear editing digital platform have revolutionised the audio production process. The majority of my sources felt that high resolution, high quality digital recording systems, are now, clearly the industry standard.
My original question, on recording platform was partly aimed at checking how many of the respondents preferred analogue recording overall. To separate them from those who recorded in digital but still had a preference for analogue elements in the recording and mixing chain. I also suspected that analogue tape would still have some supporters as a medium, despite digital being a clear market leader.
Most of those involved considered digital recording, particularly at high resolution, to be a viable and far more convenient platform than analogue.
However there were also some who felt that analogue saturation or 'tape compression' provides a unique sonic effect without a parallel, in the digital domain.
Andrew East - ex technical manager at Vox and Music Producers Guild Chairman states: "I prefer analogue recording as I feel it offers greater headroom in levels and distortion that can be recorded and used in the end result. Also, the better bass response.
Digital distortion results in clipping of the audio file that causes samples in the process to be dropped and therefore unusable recordings.
However in today's industry with ever decreasing budgets, the record company often dictates my choice! Compromise is the key word here."
Magnetic tape responds well to strong signals and therefore distorts near zero levels. As each cycle of a wave will pass through zero, so analogue tape will naturally distort.
Early designers reduced this distortion by adding a bias signal. Using a very high frequency alternating current signal injected as part of the recording process. Which helps to overcome, the resistance of the tape at low levels. Despite bias, analogue tape will still distort, and will distort more as the level is increased. Producing a signature 'punch' and 'warmth' to the sound of the recording.
Essentially small amounts of tape distortion seem to be flattering. Most analogue recorders produce around 70dB of dynamic range, and can be driven into distortion by up to 3%, before the sound becomes very unpleasant. This in comparison with the human hearing range of 120dB, and 16 bit Digital recordings 90dB, is fairly limited. This means that tape naturally compresses the signal particularly when overdriven. This effect generally results in the loss of high frequency information, resulting in a 'thicker' and 'warmer' sound than the original source.
"my reference for all things audio is still the sound of analogue, and in all my work with digital processors and converters, I try to achieve this character for my recordings." Mick Glossup18.
This view made me wonder, if this whole debate could be a generational one. Because older producers and engineers have grown up with analogue devices are they unable to make a truly un-biased comparison. However, in my research, many of my respondents are younger (mid 20's) musicians and budding music producers. Who have learnt their craft in an entirely digital age, and who generally seem to share the views of the older practitioners and designers.
However some producers choose to mix both formats, Pip Williams states:
"like many producers today, I like to record tracks with 16 or 24 track analogue and then transfer to Radar or ProTools. This is to give me the benefits of analogue tape compression and general warm sound, plus the usability and editing bonuses of digital, along with its other advantages. (Speed, ease of backup)."
Mastering & broadcast.
In Thomas Lund, and Soren H Nielsen's Audio Engineering Society papers: Level Control in Digital Mastering19, 0dBFS+ Levels in Digital Mastering20 and Lund's subsequent article, Distortion to the people in Resolution Magazine21 Lund demonstrates, how mistakes in the way we measure mastering levels are translated into distorted playback, on both professional and consumer CD players, playing commercially available CD releases.
As Lund states:22 "several golden ears in the industry feel that the best sound in pop and rock was produced between 1982 and 1995. Despite higher resolution in converters and DSP, lower jitter and probably a better overall understanding of the digital media, we seem to be on a declining rather than inclining sound quality slope even though the people buying records and films may not be conscious of it."
Lund's article starts with asserting that CD mastering has no level restrictions. That record production has become a 'loudness war ' at the mastering stage resulting in distorted CD releases.
When mastering turned from analogue, (1/4-inch and 1/2-inch two track tape) to digital level measurement was, and still is to this day, based on measuring digital audio on a peak sample-by-sample basis. This method of calculation seems have a tendency to miss transient distortion by ignoring peaks of short duration.
Maximum, mastering level is therefore 0dBFS (or full scale). Engineers simply have to avoid hitting 0dBFS for too long in order to avoid going over level.
However, Lund shows that: recordings with measurable and audible distortion can be created by mastering at high levels. That this distortion is not being indicated by current industry standard 0dBFS metering. That this distortion will result in gross digital distortion from both professional and consumer CD players.
He also demonstrates how the use of multi-band dynamics processors has become a tool not only for spectral balance, but also good at simply raising the overall loudness without increasing the peaks. Nielsen & Lund state that:23 "counting consecutive samples as a level barrier favours distortion and aggressive spectral shaping in order to achieve loudness."
Johnny Reckless is more damming; "Nearly all-modern pop/rock records are mastered way too high with transient distortion and no dynamic range. A lot of modern CD's sound dreadful compared to those being mastered 20 years ago."
Edwin Pfanzag of Salzburg University adds; " I would say that almost all the modern (pop) records are mastered with ridiculously little dynamic range and by applying multi-band compression are introducing loads of distortion."
Many of those questioned felt that record companies demand for louder records to compete on radio and TV has resulted in degraded sound.
This has to be a serious concern for an industry that surely needs to compete on quality in this digital age.
You could argue that as long as the buying public do not notice, then why should we worry. Surely, we as consumers can also demand the best quality achievable, at all stages of the recording chain.
Personally, I can hear this distortion on records, and in broadcast, but advertisers and radio stations also all seem to be culprits in debasing the quality of audio.
As Nick Howes says: " I think too much modern music is over compressed. This is particularly true and annoying in TV adverts."
Finally I have a quote from mastering engineer Bob Ohlsson:24 "...the biggest problem has been the wholesale replacement of skilled audio operators by an ever heavier-handed use of compression. Broadcasting is often un-listenable because nobody with their hands on the controls is being paid to listen anymore."
It would seem that this problem exists in a range of formats: CD's, analogue, and digital broadcast are all affected. It seems to me, that the professional audio industry has always sought to maintain high quality at the production stage, in order to allow for losses further downstream. This agenda seems to have been subverted in a lossless digital domain where impact through apparent loudness, despite heavily compressed broadcasting, is the primary goal. Leaving pure audio quality a poor second.
Now that this issue has been publicly aired, in both technical papers and the specialist audio press. One would hope to see further debate and effort towards establishing a meaningful measurement standard for, mastering digital audio, which will take account of this distortion.
Pip Williams, sums up the problem: "it's ludicrous to try and pump the level of CD's like we used to with vinyl. This has become the norm with dance music, where in clubs the track just has to pump at a constant volume."
Ears vs. technical specifications.
The resounding answer to this question is that without doubt, most audio professionals trust their ears over technical measurements. This is hardly surprising considering that most engineers and producers rely on their auditory judgement to make a living and have trained for decades by listening for noise and distortion.
However there is a bigger question being posed here: are technical specifications in fact misleading?
Many respondents felt that they trusted recommendations from fellow users rather than specifications. Some said they simply did not understand what the technical information meant, and were often confused by the differing measurement systems used, by different manufacturers.
The most vocal responses from my research were from the audio designers. Ted Fletcher has a very strong view: "technical specifications have always been the simplest of guides to how the equipment might possibly sound. With modern equipment they are very nearly useless, and often highly misleading."
Johnny Reckless comments on how he assesses audio performance: " I have learned to use both in complement. Neither is sufficient alone. Anyone who thinks otherwise is a fool. Preferably I will use several people listening and several types of measurement equipment to gain a balanced appreciation."
It would seem that there is a strong case for a new system of distortion measurement that considers the complex interaction of equipment and listener. The recent Audio Engineering Society papers cited above would indicate that work on this is ongoing, and promises to produce a more meaningful measurement of both linear, and nonlinear distortion in Audio.
The problem of course, will be, persuading manufacturers to adopt this new measurement system.
Conclusions.
In conclusion: I feel that it is obvious from my research that there is a good reason for the renaissance of 'classic' Valve and solid-state hardware. Well designed, microphones, pre-amplifiers, and outboard do seem to impart a sound that is complex and musically flattering. In the audio production process the 'fairy dust' these devices produce is something that many practitioners feel helps to ameliorate the lack of colouration or 'warmth,' imparted by digital recording platforms.
It should also be noted, however, that this only applies to properly designed and engineered technology. It is clear that there is also equipment on the market that does not live up to the 'classic' badge. I am afraid to say there may be units, which, simply endeavour to mislead the user. One or two devices, said to be built, with Valve technology are clearly solid state with a Valve or two added, largely for cosmetic effect, rather than sound engineering reasons.
When it comes to equalisation and dynamic processing, analogue units still seem to win hands down over many (though not all) of the available digital processors. Again this is mitigated by the fact that this technology is still developing. It is clear that as processing power increases so will the quality of digital processing and the complexity of design.
Plug-ins also seem to be largely considered inferior to the analogue processors, they often seek to emulate. It would also seem that this is partly a question of processing capacity, as well as an issue of design. Often users choose plug-ins for convenience rather than sonic excellence. However, again some of the 'high end' plug-ins do seem to attract some praise. My research seems to indicate that a lot more work needs to be done by software developers into the peculiar complexities of analogue processors, if they really wish to faithfully replicate these 'classic' devices.
Although there are still some, analogue tape 'die-hards' the majority opinion accepted that the various, distinct advantages, of digital recording platforms outweigh those of analogue tape. Particularly with the advent of high-resolution 24bit 96kHz digital recording.
However some practitioners still see tape as a very useful tool for certain applications. The analogue sound from landmark recordings and hardware is more often than not, still the benchmark that modern digital systems and recordings are measured against.
The 0dBFS issue, also draws a strong conclusion in support of a more accurate measurement system in mastering and greater awareness within the industry, and beyond of the dangers of digital distortion, in both music production and broadcast. This is a subject, which has recently received a great deal of publicity at the Audio Engineering Society, and in the trade press. It is hoped that this will result in a change for the better, in mastering practice.
As far as the: technical specification vs. ears question is concerned, my research shows that little faith is put in the present use of technical measurements by manufacturers. When it comes to distortion, there is a strong case for a more subjective and meaningful set of benchmarks. A system that acknowledges the different modes of distortion produced in systems would indeed be a step forward. Giving us a better idea, of how the device in question, will perform in practice.
Overall I have found this research fascinating and surprising. I hope to add to the debate with some further work in the near future. As Thomas Lund says:25 " so if the public doesn't care, why should we? Because pride in our industry, craftsmanship and the conservation of musical talent tells us to be concerned. If we believe that audio quality makes a difference..."
Bibliography/references.
1. "Auditory Perception of Nonlinear Distortion", Earl R. Geddes et al. presented at the 115th convention of the Audio Engineering Society, 2003.
2. "The Effect of Nonlinear Distortion on the Perceived Quality of Music and Speech Signals", Chin-Tuan Tan, Brian C.J. Moore and Nick Zacharov, Journal of the Audio Engineering Society, November 2003.
3. P-1012, p-3, 4,"The Effect of Nonlinear Distortion on the Perceived Quality of Music and Speech Signals", Chin-Tuan Tan, Brian C.J. Moore and Nick Zacharov, Journal of the Audio Engineering Society, November 2003.
4. Vol. 18, ( 2,"The Tube is Dead Long Live the Tube", Mark Wolverton, American Heritage of Invention & Technology, Forbes Inc, Fall 2002. Sourced from; http://www.inventionandtechnology.com/2002/02/tube.shtml
5. P1012, p 4, "Tubes Versus Transistors - Is There an Audible Difference", Russell O Hamm, Journal of the Audio Engineering Society, Vol. 21, No. 4, May 1973.
6. P6, p3, "Tubes Versus Transistors - Is There an Audible Difference", Russell O Hamm, Journal of the Audio Engineering Society, Vol. 21, No. 4, May 1973.
7. P2, p8,"Is the Truth About Distortion Distorted", Dr Harvey Rosenberg, New York audio Labs, This email address is being protected from spambots. You need JavaScript enabled to view it., April 2004.
8. P2, p1,"The Tube Sound and tube Emulators", Eric K Pritchard,
dB magazine, July/August 1994. Sourced at; http://www.pritchardamps.com/pritchardamps/tech_talk/db_magazine.pdf.
9. P184, p 2, "Tubular Tales", Hugh Robjohns, Sound On Sound, SOS Publications Group, June 2004.
10. P 59, p 6, "They don't make 'em like they used to", John Watkinson, Resolution, S2 Publications Ltd, May 2002.
11. P 28, p2, "Analogue Audio EQ", Rupert Neve, Line Up Magazine, October 1991.
12. P 28, p6, "Analogue Audio EQ", Rupert Neve, Line Up Magazine, October 1991.
13. http://www.ffliquid.com/press.html
14. P 66, p 11, "Sampling Equalisers", Michael Kemp, Resolution, S2 Publications Ltd, March 2003.
15. P68, p 3, "Sampling Equalisers", Michael Kemp, Resolution, S2 Publications Ltd, March 2003.
16. P 67, p 9, "Sampling Equalisers", Michael Kemp, Resolution, S2 Publications Ltd, March 2003.
17. http://www.mcdsp.com/users/usr_glossop.html
18. "Level Control in Digital Mastering". Thomas Lund and Soren H Nielsen. Presented at the 107th Convention of the Audio Engineering Society, September 1999.
19. "0dBFS+ Levels in Digital Mastering". Thomas Lund and Soren H Nielsen, Presented at the 109th Convention of the Audio Engineering Society, September 2000.
20. "Distortion to the people". Thomas Lund, Resolution, S2 publications Ltd, March 2004.
21. P 1, p 4-5,"0dBFS+ Levels in Digital Mastering". Thomas Lund and Soren H Nielsen, Presented at the 109th Convention of the Audio Engineering Society, September 2000.
22. P 60, p 6,"Distortion to the people". Thomas Lund, Resolution, S2 publications Ltd, March 2004.