The Need to Panic: the disease 'Loudness War' is spreading!

Introduction

Hello again my little Darlings! Today I will stand witness, not for the prosecution but for crimes against good audio quality. I will present proof that something that once was healthy now suffers from an illness so severe that no one knows if it will ever be able to recover again. The "something" this article is about is called audio quality and the illness it currently suffers from has been known for decades as 'Loudness War'. Besides presenting numerous examples I also argue that this 'Loudness War' is partly responsible for declining sales of published music (-> CDs, downloads, etc.) and that it´s a crime against peoples' nerves by producing static. A true disgrace to the technical capabilities of the CD format and its possible achievable dynamic music produced right now exhibits a limited dynamic range comparable to the Edison cylinder from 1909. Roughly 15 years ago this illness was mainly contained in the country of "Pop/Rock/Rap/Soul" but in recent years has been spreading into the land of "Scores/Classic/Jazz". Like many other diseases it behaves without intelligence or emotion, its only motivation is greed. Symptoms are as follows:

• brickwall limiting
• awkward frequency response
• audible distortions
• absence of natural dynamic
• 'life' seems drained out of the music

The ultimate fate of patients affected is death causing in turn the demise of the music industry as we know it. Not all of the symptoms above need to be present though in order for the patient to be described as being suffering from 'Loudness War'. Another thing: as with every other malady mild and severe cases need to be distinguished. The patients used as examples will be subjects recently exposed to this abnormal condition, furthermore these patients will come solely from the land 'Scores'. So let´s start by having a look at two subjects where one is not affected and the other one is:

Subject 1: 'Lord of the Rings: Fellowship of the Rings' - not affected

Subject 2: 'The Dark Knight Rises' - severely affected (already treated with "ReplayGain" here)

Compare the two individuals above: Subject 1 (Birthdate: 2001) doesn´t show any signs of the horrible disease, dynamic peaks are intact, loud parts of the music are loud but not brickwall limited while soft parts are delicate. Frequency response (not pictured) behaves completely natural. Subject 2 (Birthdate: 2012) shows strong signs of brickwall limiting, absence of natural dynamic and an awkward frequency response (pictured further below).

History of the Illness

What are you saying? I´ve been too fast? Where does the disease come from, you want to know... Well, it has been in existence for quite some time. It first emerged in the '60s in an attempt to make music as loud as possible considering the very limited abilities of the then-in-use vinyl distribution format. First Phil Spector and then Motown Records tried to squeeze out every single bit of Loudness in order for their products to be perceived better than others while also combating the poor high frequency performance of then present radio. This continued for a while and wasn´t remotely as severe as it is today - the technical gadgets available to us today were simply not existing all those years ago. Surprisingly, when the CD was introduced in 1982 the affliction turned dormant for a while, probably because people had to learn how to work with a digital (absolute) format. But at the beginning of the '90s the 'Loudness War' returned with an unparallelled force and it has been worsening ever since. More and more people are reporting rather extreme cases that even got mentionend in articles from mainstream magazines.
Who is responsible? I call them Men In Suits (abbreviated as MIS) who mostly are managers (-> investors / bosses), their natural habitat being the executive floor in huge buildings containing the offices of music publishing businesses like Sony Music, Universal or EMI. These MIS firmly believe that something that is louder will be bought by more people, hence will be generating more profit so that they can purchase another Porsche. They hide their desire for profit under a decent looking mask and say sentences like "The public wants it so". In their defense they actually have a tiny point: if out of two musical objects one is louder than the other the louder one will be perceived as sounding better, this has something to do with how our ear works. Of course, MIS are using this fact soleley to hide their greed. Anyway, Mastering Engineers (abbreviated as ME) are the poor blokes ending up spreading the desease, sucking every dynamic out of recordings and turning them into frightful sounding monsters. They are actually fully aware of what they are doing, all the while claiming that they are unable to do something about it.

Examples

Now back to the symptoms, shall we? As described above brickwall limiting and dynamic mutilation are not the only things 'Loudness War' does to audio quality. Another very audible symptom are strange frequency responses prompting equally strange reactions in listeners. In an effort to sound well on an iPod or a cellphone the overall frequency response has been changed into something that barely resembles natural sound. How would an ideal frequency response look like? Well, it would have the form of a curve decreasing towards higher frequencies. A curve like that takes into account how us humans perceive sound: low frequencies (bass) can only be perceived with great difficulty by our ear, higher frequencies (treble) are less of a problem. At the same time our ear can hear exceptionally well at frequencies in the middle (1.000 to 3.000 Hz). Recordings not affected by the health impairment adhere to this concept not by countering our hearing sensivity but by presenting a pleasing overall sonic package. Have a look at two examples:

A perfect and idealized frequency response

Subject 3: 'Superman Returns' - extremely close to the ideal above

You can see how gradually the curve declines towards high frequencies. You can also see that Subject 3 'Superman Returns' (Birthdate: 2006) and its frequency response match the idealized response almost perfectly. The only exceptions are missing deep bass below 65 Hz, rendering the perceived sound to be "faster", and a little bump at roughly 14.000 Hz. That 'mountain' is responsible for the amount of air this score presents. Bumps or "Mountains" as small as these represent a clever method to infuse audio quality with something of a "sonic character". Following now are two examples of subjects suffering from an extreme case of frequency distortion with their condition being critical:

Subject 4: 'Salt' - critical deviations

Subject 2 again: 'The Dark Knight Rises' - awful

Subject 4 (Birthdate: 2010) is able to serve as a very prominent example how serious 'Loudness War' can affect audio quality. The deviations in frequency response cause such unbearable suffering for the patient that even its description prooves problematic. Bass and treble are extremely exaggerated while the mids are recessed. Frequencies around 1.000 Hz  on the other hand are amplified to counter the extreme gain at both frequency ends. Combine that with the effects of excessive brickwall limiting (not pictured) and you are watching a patient being close to death. A skeptical person might now argue that this frequency response is caused by the many synthesizers, drum pads and samplers employed for the production of this score. But one listen to the score crushes this otherwise valid argument because in places where the pure orchestra can be heard the response still presents itself as being heavily distorted. The effect on listeners is an audio quality that is shrill, piercing, lifeless, colourless and extremely unnerving. The medical state of Subject 2 is not that critical. Instead of sounding piercing it "only" sounds shrill, dull, flat and full of 'droning' bass, in short, equally unnerving. But not all is bad in the land of scores because even now there are still a few exceptions around suffering only from mild or serious but stable cases of 'Loudness War':

Subject 5: 'The Bourne Legacy' - stable

Subject 5 is dynamically not as healthy as its frequency response

Subject 6: 'The Amazing Spider-Man' - a healthy patient

Subject 6 is also dynamically in good health with only slight limiting

Subject 5 (Birthdate: 2012) presents an interesting case: while it may look like a serious affection on paper it isn´t like that in reality. This score is perfectly suited for comparison to Subject 4 ('Salt') because both movies contain similar themes, their music similar orchestration and the same amount of artificial elements. Additionally, both scores are composed and engineered by the very same persons (James Newton Howard (composing) and Shawn Murphy (engineering)). In case of Subject 5 however the frequency response is actually pretty balanced, the artificial elements are well integrated and sound non-intrusive. The slight imperfections can be treated fairly easy with an equalizer, just as well as its dynamic compression can be removed with another medicine that will be mentionend later in the section "Therapy".
Not many words need to be spend on Subject 6 (Birthdate: 2012) since it appears to be in very good health with the result that it sounds exceptionally well (A sidenote: Dear producers, composers and engineers - yes, I know that you read my blog - a review from me can be simple and positive as evidenced by those two words).

Subject 7: 'The Avengers' - an interesting case

Subject 8: 'Snow White and the Huntsman' - more or less affected like Subject 5

Subject 7 (Birthdate: 2012) is a very interesting case because it deviates in an entirely different direction by decreasing high frequencies even further, resulting in a warm and voluptuous sonic character. Apart from having the brickwall limiting symptom it showcases a very well designed and characteristic audio quality. Subject 8 (Birthdate: 2012) is an example of a patient not unlike Subject 5 ('The Bourne Legacy') with the exception that bass is much too strong and frequencies around 2.000 Hz have too much power (again to counter the strong bass). Like Subject 5 this can be treated fairly well with several medicines.

Subject 9: 'The Shadow' - a frequency response infused with character

Now let´s have a look at scores being called "remastered", the score for 'The Shadow' serving as an example for all other releases by small independent labels, in this case Intrada. A remaster can be described as something akin to the rebirth of a score, occasionally it´s easier to just re-create a patient instead of prescribing medicine to get rid of the symptoms. In many cases however "remastering" means: "make it louder". The potential buyer expects of course that the last, yet unheard detail may be uncovered... but in 90% of all cases he/she is sadly wrong. Ironically, with every current remastering the patients' condition gets more critical by adding more treble, bass and gain. Subject 9 (Birthdate: 1994, Rebirth: 2012) luckily isn´t such a case; it shows a complete lack of brickwall limiting (not pictured) and a characteristic frequency response caused by the employment of synthesizers which are producing strange noises working within the sonic context. In fact, overall audio quality is fairly high and has been improved over the first release from 1994. That one suffered from analogue, dynamic compression producing 'pumping' artifacts which are nowhere to be found here. This is a rebirth where you have MORE dynamic range than before. In my opinion the frequency response could have been a bit more on the warm side for the rebirth since a more voluptuous sound would have suited the golden tinted atmosphere of the score better. Of course, this is complaining on an extremely high level because this score has been treated very, very well during its rebirth.

Effects

During my introduction I´ve already stated some of the effects afflicted patients may have on listeners. Effects can be cumulative though that isn´t always the case. They are as follows:

• listening fatigue
• headaches
• ringing ears
• bleeding ears (unconfirmed)
• heightened anger
• high blood pressure
• wrecked nerves

The most interesting effect would be listening fatigue. How to measure if someone is affected by it? Do we really know that 'Loudness War' affected patients are the cause for creating fatigue for the listener? Before we talk about it we need to distinguish between psychological fatigue and physical fatigue. The first one is predicated on the perception of Loudness by the listener only. The latter one can be caused by two things: mechanical fatigue and biochemical fatigue. As suggested by James D. Johnston in his article "Listener Fatigue and Longevity" outer hair cells might be depolarized if exposed to high loudness over certain periods of time. Secondary, the inner hair cell firing rate is proportional to loudness, meaning that if something has more loudness our ear starts to protect itself by decreasing the amount of information it is sending to the brain.
Psychological or cognitive fatigue is more difficult to describe because it´s hard to measure. But it´s safe to say that with speech signals multiband compression flattens the signal and causes a loss of articulation, prompting the brain to work harder to process the information. Dynamically compressed transients contained in recordings affected by the 'Loudness War' crush the leading edge of the signal, thereby reducing intelligibility and again forcing the listener working hard to understand its content. On the other hand, slower dynamic compression motivates the listener to regard the music playing right now as background music - it will be tuned out and ignored because it isn´t a cognitive challenge anymore (not to be confused with the aforementionend challenge of increased processing power). All of this turnes music into noise. Rip Rowan described it transparently in his article "Over the Limit":

"WHY IS THE LOUDER IS BETTER APPROACH THE WRONG APPROACH? BECAUSE WHEN ALL OF THE SIGNAL IS AT THE MAXIMUM LEVEL, THEN THERE IS NO WAY FOR THE SIGNAL TO HAVE ANY PUNCH. THE WHOLE THING COMES SCREAMING AT YOU LIKE A MESSAGE IN ALL CAPITAL LETTERS. AS WE ALL KNOW, WHEN YOU TYPE IN ALL CAPITAL LETTERS THERE ARE NO CUES TO HELP THE BRAIN MAKE SENSE OF THE SIGNAL, AND THE MIND TIRES QUICKLY OF TRYING TO PROCESS WHAT IS, BASICALLY, WHITE NOISE. LIKEWISE, A SIGNAL THAT JUST PEGS THE METERS CAUSES THE BRAIN TO REACT AS THOUGH IT IS BEING FED WHITE NOISE. WE SIMPLY FILTER IT AND QUIT TRYING TO PROCESS IT."

Another result that you need to look at: the decline of sales for the music industry. Have you noticed the recent increase in sales for the old, very outdated vinyl format? It has been stated for example by Kristina Dell in a Time Magazine article that the perceived "warmer" sound of vinyl might have to do something with the necessity of using finesse during mastering in order to work around vinyl's physical limitations regarding signal level. Compare that to the brutal sledgehammer approach applied to CDs. Declining sales happened at roughly the same time as the Loudness of CDs went to extreme levels. The big labels (--> Sony, Universal, EMI, etc.) won´t hear any of it, for them the culprit causing declining sales is easily identified as being a filesharer.
They do have a point: sales have been declining since the year 2000, the same time when mp3 and filesharing rose to mainstream prominence. Since it´s then easy to come to the filesharers-are-guilty conclusion the result was that they were starting to lobby for stronger rights enforcement. But as ME Bob Ludwig states: "People talk about downloads hurting record sales. I and some other people would submit that another thing that is hurting record sales these days is the fact that they are so compressed that the ear just gets tired of it. When you're through listening to a whole album of this highly compressed music, your ear is fatigued. You may have enjoyed the music but you don't really feel like going back and listening to it again.".
So our subconscious (hello Sigmund) might be responsible for diminishing sales since it obvisously cannot stand 'Loudness War' affected patients anymore. When we stretch this argument to its logical conclusion it´s safe to say that 'Loudness War' is a danger to the artform 'music' itself. Contrast (dynamic) creates tension, an intellectual and emotional challenge if you will, if one removes that you are left with garbage that formerly was art. Another opinion to support the theory of diminishing financial returns caused by 'Loudness War' comes from ME Bob Speer, who wrote in 2001 that "The record labels blame digital downloads, MP3s, CD burners, and others for the lack of CD sales. While there is some truth to their constant whining, they only have themselves to blame for the steady decline in CD sales. The record labels need to reevaluate what they consider to be good music. Much of the music being produced today isn't music at all. It's best described as anti-music. It's anti-music because the life is being squashed out of it through over compression during the tracking, mixing, and mastering stages. (...) It's no wonder that consumers don't want to pay for the CDs being produced today. They're over priced and they sound bad."
Additionally, Nick Southall wrote in 2006 that "Compression will continue to be abused in the pursuit of loudness for as long as the recording industry believes that louder shifts units (...) Global album sales are falling year-on-year, far less mega-million-selling records are occurring (...) and I think this is because the clamour to make music louder has made it less loveable, and in the long run loveable sells more." I´m afraid that his call for action won´t reach the people responsible for it. As the recent financial crisis has shown managers in charge of multi-million-dollar companies are rarely interested in long term profit. They strive for an easy success, achievable during a short time period (this is actually taught in college).
All these theories and opinions are supported by the Evergreen Project. There the author Chris Johnson visually analyzed spectograms from a number of the most commercially successful albums; he found out that "the more strongly they sell, the more likely it is that they will have High Contrast characteristics," meaning a wide dynamic range. In particular, speaking about the album 'Greatest Hits 1971-1975' by The Eagles he continues that "it is gratifying, but unsurprising, to discover that the single most commercially important album in RIAA history contains some of the most striking dynamic contrasts pop music's ever seen."

Granted, he talks about pop music there - but if 'Loudness War' is bad for pop how much more difficult might the situation be regarding filmscores? Those consist of recordings made with an orchestra and while slight dynamic compression always has been a part of the sound of Rock/Pop/Soul music this can´t be applied that easily to orchestral recordings. On natural instruments dynamic compression tends to be very audible since it destroys timbre and character, it simply removes microdynamics vital for recognizing them. These instruments are part of the orchestration that most of the time has been carefully fleshed out - effectively dynamic compression or brickwall limiting destroys the message of the music, as a result rendering it unsuited for the movie it was written for - shouldn´t that be avoided?

Therapy

Medicating the aformentionend patients in critical condition for symptoms of 'Loudness War' is not easy and involves spending some time with them in order to ease their suffering. Therapy consists of a lot of research and also bears the risk of hurting them even further by destroying their data integrity. Naturally, another risk is the possibility of prescribing the wrong medicine. Of equally high importance is the amount of medicine to be dosed - too much of the treatment might destroy the subjects' sonic character, e.g. render it even more unrecognizable than before. Nevertheless, treatments so far available are:

• ReplayGain as a starting point
• declipping (for example with iZotope's DeClipper)
• ReLife (Terry West Productions)
• re-processing individual channels for battling stage symptoms (for example PLParEQ)
• re-equalization (for example with iZotope's Ozone)
• repairing stereo field with multiband stereo imaging (iZotope Ozone again)

As you can guess those treatments are extremely expensive. Luckily, not all of them need to be administered in every case. Physicians may therefore apply treatment as they deem necessary after extensive study of the patients' medical condition. A few examples of successful cases are presented as follows:

Remember Subject 4 ('Salt') - before treatment

Subject 4 after treatment

As you can see above on the before treatment / after treatment charts, Subject 4 has been eased of its pain: Amplficiation of frequencies around 11.000 Hz has been reduced considerably as well as the overbearing bass response while mids and presence area have been strengthened. The therapy also got rid of the effects of brickwall limiting by eliminating it completely. All in all, I can say that treatment has been a success - people listening to Subject 4 aren´t complaining anymore of headaches or bleeding ears, they now are actually able to recognize the music in a non-fatiguing way.

Subject 2 ('The Dark Knight Rises') before treatment

Subject 2 after treatment

The same is valid for Subject 2. You can see how much the overpowering, 'droning' bass has been reduced, as a result presenting sonic colours for the first time. Mids around 1.000 Hz have been strengthened, the same goes for higher frequencies at roughly 4.800 Hz. The amount of amplifications at 3.000 Hz has been straightened out as well. Sadly, the symptoms for Brickwall limiting (not pictured) couldn´t be rectified completely but at least ReLife was successful in easing them greatly.

Conclusions & Opinion

I called this article "The Need to Panic: the disease 'Loudness War' is spreading". You, my beloved constant reader, might think that I´m exaggerating. You might look at the charts and think "The differences look tiny to me, they can´t be as bad as described." Should that be the case I have to tell you that you´re mistaken. While these pictures don´t look like much they contain a powerful statement: gain differences for several frequency bands between 'ill' releases and 'cured' releases happen to be as big as +/- 10 dB (for example on 'Salt' ). Additionally, the effect of transients with natural dynamic can make a difference like night and day. For subjects like 'The Avengers' or 'The Bourne Legacy' the effect of sound being not differentiated may not be that apparent - on subjects like 'The Dark Knight Rises' or 'Salt' they indeed are easily recognizable and they of all are in desperate need of being cured. If I´m extrapolating the examples given here to the entire soundtrack market it leaves us with 50% of all scores sounding well enough while the other 50% are nothing more than distorted audio garbage sounding like a Wax cylinder from 1909.

Most of the time that crap (sorry) is unleashed to the public by the big majors (--> Sony, Universal, EMI) while smaller and independent labels like Intrada or LaLaLand generally are releasing the "good", unaffected half (with some exceptions). Just for the sake of argument let´s expand this situation further into the even bigger market of mainstream releases. How can people treat 'Loudness War' affected releases with respect if even their creators won´t? What if not only filesharers were responsible as it is so often stated by the music industry? What if the 'Loudness War' had an effect too? This would be quite ironic: cranking up the volume was done in order to boost sales and not for the opposite to happen. Is it just a coincidence that sales degenerated at the same time when music got extremely loud? The big labels won´t finance research into that matter - regarding every consumer as a potential thief is much cheaper and easier. (A Sidenote: I´m not defending filesharers here - but that argument alone is insufficient). Furthermore, trying not to make music louder would require balls, something those managers or financial advisors seem to lack considering the absence of daring decisions. I can tell you this though: I have stopped buying mainstream releases because they sound like shit. I´m not even downloading them illegally. While they might be available for free at some place on the net they still sound like shit. I can also promise you that I will stop buying score releases coming from major lables if their sound quality continues to be as bad as it is now, they are a disgrace to the technical possibilities the CD offers and I won´t accept music that has been destroyed on purpose. Right now I´m buying roughly 70 CDs, SACDs or digital downloads every year (not even including DVD's or BluRay's), most of them being scores and classical music - but I cannot guarantee that this will continue indefinetely.

Enough with the ranting. I could go on and talk about mastering engineers saying "No" to 'Loudness War' even if that would mean getting fewer assignments. As you can see above they are complaning - but are they doing something about it? You may remember the statement from ME Bob Ludwig; considering his opinion it's surprising that he continues to expose music to the disease by cranking up the volume using the excuse of "The customer won´t accept anything else." If he would be true to his word he´d stop taking on those assignments. To be fair to him and others: I´m not sure if I wouldn´t behave like them if I were in their shoes; it´s easy to see oneself as a victim instead of an accomplice - and I have been a victim of my own bigotry in the past too. Anyway, this situation has to stop. An orchestra - not even one augmented by synthesized sound elements - cannot be treated as the next best Pop music release and I just don´t have the patience to 'cure' their diseases. I consciously know about this and can subsequently try to get rid of it - but what about the tens of thousand other people being unaware? They simply stop buying these releases. So I urge the people in charge to act now before you´re broke.

Sources:

experience and 'The Loudness War: Background, Speculations and Recommendations',
Earl Vickers, ST Microelectronics, first released at the 129th AES Convention, 2010

Last update: 16.07.2013

More MD Recorders: Sharp MD-MT 180 & Sony MZ-N 510

Finally! A long time has passed since my last article and while I´m sorry for what appeared to be apparent inactivity I actually have been very busy. I´ve spent these two months with several 'new' MiniDisc recorders, more measurments regarding ATRAC... and of course: having fun! Yes, I may appear to be the most gadget-centric geek in the history of blogging  (don´t you sometimes have the feeling that it´s healthy to be delusional?) while in reality I tend to be an outgoing person with an occasionally quite egocentric attitude (WHO? MEEEE? - imagine Cher here). Guys, here in Germany it finally happened to be... Summer! The Sun was shining, the boys were wearing short pants revealing everything of secondary importance... well, if there was anything to be revealed... the flowers were smelling, the air was warm and people seemed to be relaxed which isn´t an easy feat for us stubborn Germans. Nevermind. I would like to talk about two MiniDisc recorders I purchased on eBay. One is the Sharp MD-MT 180 and the other is the Sony MZ-N 510, both purchased in used but decent condition. I also bought a Sony MZ-N1 - I would love to talk about that one too but one week after receiving this lovely looking MD Recorder it broke, I suspect either the ribbon cable to the magnetic writing head or the laser responsible for its malfunction. Both seem to be common flaws with this little machine; the ribbon cable could have been repaired by myself with a bit of soldering but the laser is unsalvageble. It is a MiniDisc recorder to be avoided at all costs, it just is too unreliable for its own good. Even if it would be well sounding (it was). The MZ-N 510 doesn´t have these reliability problems and neither does the Sharp.

The Sharp MD-MT 180 - ugly, don´t you think? Not even Instagram-like 'art' can hide this

Sony MZ-N 510 - my second NetMD Recorder and IMO with a superior design compared to the recorder above

I´ve acquired the Sharp because several people all over the net are talking about their high sound quality. According to them Sharp MD Recorders generally have a well rendered bass response and good technical qualities. Sharp themselves claimed their implementation of the ATRAC coding scheme to be the best on the market, superior to the one Sony originally designed and improved over the years. Why would Sony allow other companies to produce their own versions of ATRAC? Because when they were releasing the MD format to the market in 1992 Sony had hoped to increase market penetration by allowing competing manufacturers to use their patents without actually having to pay fees for them. In the end, three other companies apart from Sony produced their own ATRAC implementations: Sharp, Pioneer and Matsushita (Technics, Panasonic). That situation inadvertently led to different and confusing ATRAC version numbers: while the Sony MZ-N 510 uses ATRAC DSP Type-S, the Sharp MD-MT 180 uses Sharp-ATRAC 6.0. Both ATRAC versions feature full 24 Bit processing (-> WideBitStream), enabling them to capture audio material with a resolution higher than that of the Compact Disc in spite of using a lossy encoding scheme. In order to find out how the Sharp ATRAC performs and if it would offer an improved way of encoding music I decided to buy it. The Sony MZ-N 510 on the other hand was purchased to have a look at the last ATRAC version to be released by Sony and to find out how well the NetMD feature performs. When ATRAC DSP Type-R (Type-S is the same) was released to the market Sony claimed that an additional encoding step resulted in less distortions when compared to the predecessor ATRAC 4.5. According to them they achieved it by doubling the processing power of the ATRAC chip. Well, to me this sounds an awful lot like a flowery description of a classic 2-pass-enconding, yet how Sony would achieve this in combination with on-the-fly encoding is baffling to me. Anyway, let´s see how it performs (playback necessary for the charts below was performed by the Kenwood DM-5090 through its digital, optical output):

Sharp ATRAC 6.0

Sony ATRAC DSP Type-S (the same as ATRAC DSP Type-R)

The Noise Level of -400.0 dB (A) of ATRAC 6.0 is probably achieved by the Sharp muting its input, this muting can obviously be done to a specific channel only which in turn leads to the phenomenal Stereo Crosstalk performance. Sony's ATRAC obviously treats both channels as one entity; it also doesn´t mute its input. THD + Noise and IMD at 10 kHz seem to be much better compared to the Sony while IMD + Noise seem worse. The chart below shows another row of measurments (four seperate measurments for each ATRAC version combined into one respective output chart). Below that you´ll also find some graphs explaining some of the measurments:

Comparison of Sony ATRAC 4.0, DSP Type-S and Sharp ATRAC 6.0

Frequency response ((dB) A) of the several implementations

THD + Noise comparison

IMD comparison

As you can see on the frequency response graph Sharp's ATRAC does something odd to the high frequency spectrum. Further examination of the encoded multispectrum test signal reveals that high frequencies are not really "there", instead they have been replaced by quantization noise. Sony's ATRAC probably behaves the same even though it appears to be concealed better there. The THD + Noise graph also shows that the Sharp ATRAC has problems with distortions around 2.000 kHz, a frequency area where our ear is very sensitive. Even Sony's ATRAC 4.0 exhibits slightly lower distortions there, ATRAC DSP Type-S faires best with the lowest amount of distortion. On the other hand the Sharp ATRAC has the lowest low frequency distortions. The intermodulation distortion graph shows - just like the frequency response - some odd high frequency behaviour of the Sharp ATRAC again. With the Sonys the overall noisefloor is higher but much more consistent throughout. With them though one thing is crystal clear: the attention of the encoding scheme was shifted to frequencies more sensitive to our ears instead of higher frequencies when one observes measurment differences from ATRAC 4.0 to ATRAC DSP Type-S. While ATRAC 4.0 encodes high frequencies with less noise it partly fails at lower frequencies, with ATRAC DSP Type-S it´s the other way around. Sharp's ATRAC 6.0 obviously attempts to do the same but fails; high frequencies show odd behaviour while frequencies important to our ear show higher distortions as well. Which is evidenced with music too:

"Lady in the Water", Sharp ATRAC 6.0

"Lady in the Water", Sony ATRAC DSP Type-S

As you can see above Sharp's ATRAC 6.0 only rarely encodes frequencies beyond 16.000 Hz, it behaves exactly like mp3. The track above is only one example, I´ve seen it with every single track I´ve recorded with it. With every track the frequency amount or content above 16.000 Hz seems to be irrelevant to Sharp's ATRAC. Sony's ATRAC DSP Type-S on the other hand manages to encode content up to 18.000 Hz, sometimes up to 19.000 Hz and occasionally even up to 20.000 Hz. BTW, before I made any measurments at all I started with my examinations by listening to the respective recordings, then I took the step of looking at them while measurments were the last thing I did and only confirmed what I was suspecting: Sharp's ATRAC 6.0 isn´t very good. From what I was seeing above I expected ATRAC 6.0 to measure perfectly at frequencies important to our ear, the missing content above 16.000 Hz led me to expect an attention-shifting effect not unlike the one I experienced when going from ATRAC 4.0 to DSP Type-S.

Sharp MD-MT 180 - disappointing performance

Sony MZ-N 510 - NetMD equipped and a reliable recorder

How do these different ATRAC implementations sound? Do they sound at all, something they shouldn´t do? Yes, they do. Recordings made with the Sharp MD-MT 180 always sound flat and "grey", Sharp ATRAC encoded music played back on for example the Sony MZ-R 30 or the Sony MZ-R 900 somehow seem to loose their amount of reverb, turning them into something uninvolving without much differentiation. Recordings made with the Sony MZ-N 510 on the other hand sound... well, pristine and very much transparent. During the last two months I haven´t been able to trip the ATRAC DSP Type-S into making any errors whatsoever. Despite using lossy compression the latest Sony ATRAC seems to be fully transparent, I´m unable to discern any difference between a 24 Bit .wav original and its ATRAC DSP Type-S encoded counterpart. Sony's ATRAC 4.0 in comparison softens the sound somewhat and can be prone to compression errors. The new James Horner score "The Amazing Spider Man" for example contains two tracks with fingersnapping and ATRAC 4.0 fails very audible at encoding them, creating strange sounding compression artifacts. ATRAC DSP Type-S doesn´t have any problems at all, recordings made with it are just perfect. It doesn´t matter if they are played back by the MZ-R 30, the MZ-R 900 or the Kenwood DM-5090. BTW, I double checked the transparency of ATRAC DSP Type-S with recordings made with the ATRAC DSP Type-R, the direct predecessor. For that I used the (now repaired) Sony MZ-R 501 from my boyfriend which he had in his posession for nearly ten years (now it´s mine ;)). ATRAC DSP Type-R is very much the same encoder, only the ATRAC3 decoding part is not as powerful and error resistant as the one contained in ATRAC DSP Type-S. Not surprisingly, both measured and sounded the same.

Sharp MD-MT 180 close-up

Sony MZ-N 510 close-up

But how do they sound with their headphone outputs during playback? Connecting my Sennheiser HD-448 to these outputs (they both lack line-outs) and using Sony-encoded music the Sharp easily beats the Sony. For one, the Sharp's output is considerably louder (boasting 10 mW into 16 Ohms compared to 5 mW for the Sony) and it´s also the more neutral output. The Sony's output isn´t very good I´m afraid, bass is too strong and treble seems muted while the stage is constricted. It appears to be better suited for recording and not so much for playback, considering that you - like me - have the same desire for neutrality. Attention: the playback quality for both gadgets is more or less caused by their built-in headphone amps and has nothing to do with D/A converters, ATRAC decoders or anything else. One thing however is for sure: Neither the Sharp nor the Sony are able to match the sound of the MZ-R 30 or the MZ-R 900 through their line-outs. While the MZ-R 30 does indeed colour the sound slightly the R-900 is one of the most neutral portable players ever released - even though both of them are clearly superior playback devices. You may remember that I´ve written during my last MD article that I would like to have a look at ATRAC3 performance... oh, you are confused by another ATRAC version? Yes, it´s an easy mistake to make so let me explain: ATRAC3 was released in 1999 together with the MDLP (MiniDisc LongPlay) addition to the MD format and shouldn´t be confused with any of the aformentionend ATRAC versions. Data rates for MDLP were 132 kBit/s (LP2) and 64 kBit/s (LP4) and with these you could double or quadruple recording time. But even when boasting the same name as the old ATRAC these two encoding schemes doesn´t have very much in common... and it´s something that´s very audible: LP2 sounds decent but is no match against more modern implementations of mp3, ogg, wma or against the first ATRAC. LP4 is just one thing in my opinion: useless, more recent codecs like AAC are much better suited for a bitrate this low.

... and the Winner is the Sony MZ-N 510

If this would be a classic contest between two contestants the MZ-N 510 would be the winner. It´s a very reliable recorder, with transparent and pristine recording quality. Playback quality on the other hand isn´t that good. Its NetMD feature is only suited for one thing: conveniently titling an MD. For music it´s almost worthless simply because the achievable sound quality suffers so much. You can imagine how I´ve used my several MD recorders during the last two months: the MZ-N 510 for recording, adding or substracting track marks and for titling an MD while either the MZ-R 30 or the MZ-R 900 were used for playback in combination with my FiiO. Seems complicated? Yes, it is. It would be much easier to just use my Sansa Clip+ for it will upload music much faster (with MD recorders I have to record it instead in real-time), is much smaller and supports more modern encoding schemes. But no, I won´t do it just because it doesn´t sound as well as the Sony MZ-R 30 and is not even close to the sonic perfection of the Sony MZ-R 900. Nevermind, should you consider buying a used MD recorder go for the MZ-N 510 - but use it as a flawless recording device only. Avoid the Sharp for it will inadvertently colour recordings while at the same time won´t encode as well as the Sony.

Last update: 03.09.2012