[gregorio]

There are countless threads throughout Head-Fi dealing specifically with (or mentioning) bit depths and/or sampling rates in terms of perceived quality. Comparisons or opinions on so called Hi-Rez recordings, in terms of say Hi-Rez verses standard-definition CD (16/44.1), being a common example. There are always subjectivists who state that they can clearly hear a difference/improvement with Hi-Rez, an argument which is (rightly) rebutted by the better informed rationalists on the grounds that the subjectivists have fallen victim to at least one type of logical fallacy. At some stage during the thread, the rationalists will likely need to point out that the subjectivist/s are not comparing like with like; they are comparing different masters, not different formats. This quote taken from a recent thread is a typical example:
 
Quote:

The only reason why it would sound better is that it has been mixed & mastered differently than it's CD counterpart.

I'm pretty sure we all are up for better music quality, however upping the settings that are beyond human hearing to begin with doesn't magically do that; that's only snake oil ... we can only achieve better quality music when audio / mastering engineers start doing their job properly.

 
I certainly agree with the first part of the quote and the first sentence of the second part but the last part ("we can only achieve better quality music when audio / mastering engineers start doing their job properly"), crops up fairly commonly here on Head-Fi in one form or another. IMHO, while the quote as a whole indicates an enlightened understanding of digital audio, this last part also indicates both a very narrow appreciation of "better quality" and a misunderstanding of the role of the mastering engineer. This, and some other aspects of mastering which appear to be generally quite poorly understood, is why I've started this thread.
 
"Better Quality":
 
Of course, the determination of quality is a personal, subjective opinion. Having said this, there are a number of determining factors members here would commonly agree on, which would therefore form somewhat of a consensus of opinion. A typical example is dynamic range. Heavily reducing the dynamic range, mainly accomplished through a process called "compression", is likely to have all of the following side effects: It increases the noise floor, it damages/smears the transients and it damages the stereo image (perceived width and/or depth). In any moderately decent consumer playback scenario, these side effects are usually perceived as distortion, a loss of detail and/or a poorer "soundstage". To those who listen to music critically and who have a better than "moderately decent" playback scenario, these side effects can (on occasion) legitimately be described as "night and day", especially if a significantly less compressed version is available for direct comparison. So, we can all agree (have a consensus opinion) that compressed recordings are "poor quality" and indicative of incompetent/poor mastering. ... Whoa, hang on a minute, not so fast!! There are a couple of fallacies in this statement which means that it's untrue, or at least, that it's only true sometimes:
 
1. While there are a variety of reasons people join Head-fi, probably the most common is to improve their playback scenario, either by looking for reviews/comparisons of specific equipment or occasionally by gaining a better understanding of how equipment works generally or by some combination of the two. While there are many different sub-groups of members, even those who appear polemically opposed, do have certain things in common, which makes them part of a bigger but still relatively small niche demographic. The consensus opinion, that which the vast majority of us can  agree, is NOT a truism, it's specific to this bigger though still relatively small niche!
 
2. If this statement really were true, why would we (mastering engineers) go to the additional effort of adding audio compression in the first place? The reason is simple; virtually without exception, some amount of compression is desirable: It helps to smooth out or tie together an otherwise disparate recording (a common weakness with multi-tracked or multi-sourced mixes), it helps improve presence and it reduces dynamic range (I'll come back to why we might actually want this!). Providing not too much compression is applied, we can achieve these aims without too much of the unwanted side effects. While some compression processors are specifically designed to achieve these wanted aims as transparently as possible, others are specifically designed not to! "Classic" compressors take advantage of the non-linear nature of compression to add a type of colour, flavour or character which maybe desirable aesthetically. TV/film audio post virtually always employs the most transparent compressions but with music recordings it varies from genre to genre and even sometimes within the genre. It's not at all uncommon, with many genres, to employ different types of compression within the same music mix/master, to achieve different effects. In fact, this has evolved to the point where the use of compression itself can be described as an art and there are a number of rather convoluted compression "schemes" and ingenious applications.
 
Coming back to wanting to reduce dynamic range not only goes to the heart of the role of mastering but also ties the two above points together.
 
The Role of the Mastering Engineer:
 
The role of the mastering engineer is much more than just the application of compression, it's to take the producer's studio mix and turn it into a mix which works as intended in the consumer playback scenario (which raises issues I may address in a further post) but I'll continue with the example of compression. Both at the mixing level and the mastering level there are reasons why reducing the dynamic range is often desirable, for now, I'll just deal with the mastering side.
 
A lot of dynamic range is only a good thing if our playback scenario can actually reproduce it. I've used the term "playback scenario" rather than audio equipment because it's not just about the cost/quality of equipment. For example one can have a very expensive playback system, capable of reproducing a very wide dynamic range fitted in a car but in practice, while driving, we have a relatively poor playback scenario which is incapable of a wide dynamic range (due to engine, road and other traffic noise). Either we crank up the volume to hear the quiet parts and deafen ourselves when the loud parts come along or we set the loud parts to a comfortable level and then can't hear the quiet parts. Another example would be the use of decent head/earphones in a train, bus, plane or metro. Of course in many situations a poor playback scenario is caused by poor equipment or even fairly decent equipment turned down low because the listener is only listening casually, maybe while they're working or doing something else or maybe because it's late and the neighbours have complained! In ALL of these scenarios, a reduced or even a highly reduced dynamic range is a good thing! Being able to hear the whole track in relative comfort is obviously "better quality" (even with unwanted side effects) than only hearing some parts of it. Here on head-fi discussion is mostly in terms of optimum playback scenarios or at least as optimum as practical but this is a niche! As mastering engineers we rarely get the opportunity of mastering for a single niche playback scenario, we have to master for a number of different scenarios and this means that for the more extreme scenarios, compromises are inevitable. We could master recordings to have as much as a 70dB dynamic range but who would pay for them? Yes, many here would cheer but in reality only a portion of what is a already relatively tiny number of people (who have the required playback scenario) would buy such a recording. Just recouping the costs of making such a recording is quite a risk, let alone making any serious profit. By the way, a 70dB dynamic range is about the maximum practical, we would in certain cases be able make much more but we are running into a couple of practical limitations: 1. The dynamic range of mics and recording environments and 2. The safe range of human hearing. To hear all the dynamic range of even a 70dB recording means that if you have an extremely quiet listening environment, say 30dB, the peak level of the recording would be at 100dB, which most people would find very uncomfortable. Most consumers optimal listening environment is more likely to be around 50dB and therefore a max dynamic range of about 40dB would be more appropriate and for those who commonly listen to music while travelling or doing something else, 20dB or less would probably appear "better quality". In other words, a mastering engineer who crushes a recording with compression could be achieving "better quality" and very much "doing their job properly", even if it sounds the opposite to you personally in your personal playback scenario! One other consideration, mastering engineers (like other audio engineers) generally have limited executive authority. We don't have the final say, we can advise or recommend but at the end of the day our employers (musicians, producers or labels) can and do demand we give them what they want and commonly, that is a master more compressed than we would ideally like.
 
G

 

[RRod]

Speaking for myself, when I say "man the mastering is bad on this", I mean the much more complicated statement "Why can't the artists and suits sign off on something a bit less compressed?", so don't think everyone blames you. What miffs me is that it seems like engineers are being forced to cater to scenarios that could be easily handled by just a modicum of DSP on the player side. It would seem trivial to have a world where portable players came set to a stock radio-like compression + loudness normalization setting that could be turned off if desired. So the masses aren't the wiser and the audiophiles can get back their dynamic range. Still, I know that the non-engineers in the room would want to hear a vetted final product, screwing it all up anyway. Sadly it doesn't seem like people these days care much about music, judging from answers I get to the question "what do you listen to?" from the lay public.

 

[watchnerd]

It's a lot more than what was mentioned above, as well.
 
EQing alone can have a huge effect beyond obvious timbral changes.
 
Want to move the image forward or backward in the sound stage?  
 
+/- 2 dB changes centered around 1600 Hz, Q of about 1.00
 
Want on increase 'transparency'?
 
Crank down the presence region, a high Q of about 4.00 for -5 dB at 5000 hz
 
Warmth?
 
220-250 Hz is your friend, +2 dB with about 1.5 Q
 
Want to bring that piano forward?
 
Middle C is 262 Hz
 
Want more punch out of the toms?
 
The 'punch' for the drum kit is about 131 Hz
 
etc. etc.

 

[dprimary]

The mastering engineer also gets blamed for recordings that are already over-processed and over compressed. Not much you can do when the guitar and bass players each have racks of processing till little more then mush is left. The tracking engineer instead of having the players bypass a ton of signal killing processing, decides to run it though his latest wonder box and a dozen plug-ins. The mix engineer runs it though more outboard gear and plug-ins.
 
The final mixes are sent to the poor mastering engineer who left the task of fixing something that is little more then distorted compressed pink noise.

 

[watchnerd]

  The mastering engineer also gets blamed for recordings that are already over-processed and over compressed. Not much you can do when the guitar and bass players each have racks of processing till little more then mush is left. The tracking engineer instead of having the players bypass a ton of signal killing processing, decides to run it though his latest wonder box and a dozen plug-ins. The mix engineer runs it though more outboard gear and plug-ins.
 
The final mixes are sent to the poor mastering engineer who left the task of fixing something that is little more then distorted compressed pink noise.

 
You left out the guy who added the big dubstep bass drop at +18 dB that gets clipped because it smashes into the ceiling.

 

[dprimary]

It doesn't help that some of the biggest selling albums in the last few years have pushed distortions to all new levels.

 

[watchnerd]

 
It doesn't help that some of the biggest selling albums in the last few years have pushed distortions to all new levels.

 
Speaking of which:
 
I listened to Daft Punk's "Random Access Memories" today (5 Grammy's, including engineering).  
 
The "high resolution" version had such crappy dynamic range, with audible compression and distortion. It was no better than the lossy version.
 
I then compared it to the vinyl rip and it was much better.  And it would appear that the loudness database concurs:
 

 

[RRod]

   
Speaking of which:
 
I listened to Daft Punk's "Random Access Memories" today (5 Grammy's, including engineering).  
 
The "high resolution" version had such crappy dynamic range, with audible compression and distortion. It was no better than the lossy version.
 
I then compared it to the vinyl rip and it was much better.  And it would appear that the loudness database concurs:
 

 
Obviously the people who care about sound are using vinyl, the superior medium.

 

[spruce music]

I have only done some non-pro recording.  Acoustical instruments.  I have done this with two mike minimalist setups that most audiophiles would like.  Plenty sense of air and space with a reasonably accurate portrayal of the real sound.  I also have done the multi-mike thing with no processing beyond mixing levels.  All the non-audiophiles who heard both, liked multi-miking.  Add bit of tasteful helpful compression stopping when it is just enough to sound okay in the car.  Everyone liked it better.  This in a direct comparison on a good system.  Do some more EQ on a per track basis to bring out this or that instrument, and tone this other one down everyone liked it better still. 
 
Now as an audiophile I find comparing the last version to the purist stereo miked version no contest.  The two mike version is so much better.  I would also do it in a compressed form for use in cars or other noisy environments and that would be it.  I am alone in this versus everyone else including the very musicians themselves.  So were I pro making a living it is obvious what I would need to do. 
 
I too like the suggestion above that playback gear have a couple or three compression levels that could be applied as needed.  But it hasn't happened and likely won't happen.  As people stream and or download music it would be nice to have a uncompressed version and a compressed version of every track when you buy it.  I can see however it would be like one or two percent of all customers that would want the bother or complication so it isn't going to happen.

 

[gregorio]

  so don't think everyone blames you.

 
It does often seem like that. However, I'm not of course defending all mastering engineers. These days anyone with a laptop and DAW software can call themselves a mastering engineer and there is far more very poor mastering today than there once was. I was just pointing out the fact that even good mastering engineers can produce mediocre/poor masters (for reasons beyond their control) and that just making a blanket statement that it's the fault of the mastering engineer can easily be a falsehood. Furthermore, much of my OP was aimed at the fact that what a critical listener might consider to be a poor master may in fact be a very good master but one which has simply not been specifically aimed at critical listeners.
 
Quote:

  It would seem trivial to have a world where portable players came set to a stock radio-like compression + loudness normalization setting that could be turned off if desired.

 
It might seem trivial in theory but it's not in practice, for a few reasons: 1. Getting a bunch of manufacturers to agree on anything is next to impossible, especially when we're talking about additional cost to them. Yes, it's relatively easy as far as DSP is concerned but it involves not only designing the DSP software but designing the hardware to run the software. Audio devices aren't like computers which have a bunch of CPU power sitting there waiting for something to do (the user to run programs), they're designed with the amount of processing power they need and no more. 2. There really is no "stock radio-like compression", they tend to each have their own multi-band compression/limiting depending (presumably) on what they think is right for their content and target demographic. 3. There is already something similar to what you are describing. AV Receivers have this built-in by default and actually the consumer doesn't have to do anything, the compression setting is controlled by metadata embedded in the digital audio stream. However, the situation with AVR's is different, they have to license a chip from Dolby which is the same chip supplied to all AVR manufacturers and of course to work, it requires the audio to be delivered in Dolby Digital format. 4. In practise it's not so easy to implement a loudness normalisation which works acceptably well with a wide range of material. The music world has no "specs" beyond the obvious no higher than 0dBFS, a decent loudness normalisation implementation would therefore have to work very well with material only lightly compressed, material which has been compressed up to (and sometimes beyond) the limits of what is possible and everything in between. Loudness normalisation works far better as an audio delivery specification than as a process applied to audio after it's been delivered. 5. Many manufacturers would be wary of introducing anything which could degrade perceived audio quality, which the units they manufacture could be blamed for.
 

  It's a lot more than what was mentioned above, as well.

 
In all fairness, I did state that: "The role of the mastering engineer is much more than just the application of compression...". I used compression as the example because many of the discussions here on head-fi ultimately come down to compression specifically; bit depths and Hi-rez for example and even SQ in general, although the latter of course involves all the aspects of mastering rather than just compression.
 

  Now as an audiophile I find comparing the last version to the purist stereo miked version no contest.  The two mike version is so much better.

 
That depends on what you mean by "purist"? If you mean "simplest", sure, I would agree. In some situations, particularly those which involve larger ensembles, a mic array and/or stereo pair, plus some spot mics typically produces a better end result, a result preferred even by most "purists". Of course, there are always going to be some extreme "purists", for whom "simplest" automatically means better, regardless of what it actually sounds like.
 

  Obviously the people who care about sound are using vinyl, the superior medium.

 
As bizarre as this may sound, it is the "superior medium"! In order not to be bizarre, this statement needs some qualification. As an audio format vinyl is significantly inferior to CD but as a mastering medium it's superior because: You can't have a portable turntable or one fitted in say a car and even LPs are of fairly short duration. In other words, from a mastering perspective, our target demographic are going to be listening in at least a moderately decent environment, probably with at least moderately decent speakers/cans and are more likely to be listening critically than casually (IE. While doing something else).
 
Of course, much of the time it's entirely possible to make two masters; a fairly dynamic, detailed one and a much more compressed version. Indeed, making two versions is common and by no means a new thing. For decades it was common to make a release version and a different (usually more compressed) version specifically designed for radio broadcast. These days it's not uncommon to create a "hi-res" version and a "standard-def" version. My complaint is that a hi-res version is commonly used as a cynical marketing ploy to push higher data density formats/equipment, instead of distributing it in the same 16/44.1 format as the standard version.
 
G

 

[watchnerd]

  My complaint is that a hi-res version is commonly used as a cynical marketing ploy to push higher data density formats/equipment, instead of distributing it in the same 16/44.1 format as the standard version.
 

 
My complaint in the case of "Random Access Memories" is that the high-resolution version didn't use the higher dynamic range master (i.e. the vinyl one).

 

[RRod]

   
It might seem trivial in theory but it's not in practice, for a few reasons: 1. Getting a bunch of manufacturers to agree on anything is next to impossible, especially when we're talking about additional cost to them. Yes, it's relatively easy as far as DSP is concerned but it involves not only designing the DSP software but designing the hardware to run the software. Audio devices aren't like computers which have a bunch of CPU power sitting there waiting for something to do (the user to run programs), they're designed with the amount of processing power they need and no more. 2. There really is no "stock radio-like compression", they tend to each have their own multi-band compression/limiting depending (presumably) on what they think is right for their content and target demographic. 3. There is already something similar to what you are describing. AV Receivers have this built-in by default and actually the consumer doesn't have to do anything, the compression setting is controlled by metadata embedded in the digital audio stream. However, the situation with AVR's is different, they have to license a chip from Dolby which is the same chip supplied to all AVR manufacturers and of course to work, it requires the audio to be delivered in Dolby Digital format. 4. In practise it's not so easy to implement a loudness normalisation which works acceptably well with a wide range of material. The music world has no "specs" beyond the obvious no higher than 0dBFS, a decent loudness normalisation implementation would therefore have to work very well with material only lightly compressed, material which has been compressed up to (and sometimes beyond) the limits of what is possible and everything in between. Loudness normalisation works far better as an audio delivery specification than as a process applied to audio after it's been delivered. 5. Many manufacturers would be wary of introducing anything which could degrade perceived audio quality, which the units they manufacture could be blamed for.
 
 
As bizarre as this may sound, it is the "superior medium"! In order not to be bizarre, this statement needs some qualification. As an audio format vinyl is significantly inferior to CD but as a mastering medium it's superior because: You can't have a portable turntable or one fitted in say a car and even LPs are of fairly short duration. In other words, from a mastering perspective, our target demographic are going to be listening in at least a moderately decent environment, probably with at least moderately decent speakers/cans and are more likely to be listening critically than casually (IE. While doing something else).
 
Of course, much of the time it's entirely possible to make two masters; a fairly dynamic, detailed one and a much more compressed version. Indeed, making two versions is common and by no means a new thing. For decades it was common to make a release version and a different (usually more compressed) version specifically designed for radio broadcast. These days it's not uncommon to create a "hi-res" version and a "standard-def" version. My complaint is that a hi-res version is commonly used as a cynical marketing ploy to push higher data density formats/equipment, instead of distributing it in the same 16/44.1 format as the standard version.
 
G

 
I felt colorful:
 
Yeah I can dig all that, but again it's frustrating to have audio at the mercy of sociology. Companies want their stuff to be proprietary, musicians and execs want tracks to be louder, regular folks could care less about any of it and just want their Adele albums, and audiophiles, instead of making a fuss about their digital content, are flying back to vinyl. 
 
It doesn't sound bizarre at all, at least for the pop world. As a classical and sound effect lover, though, I can't fathom going back to vinyl with its worries about correctly tracking cannons.
 
Totally agree that there was not and still isn't a reason to have foisted new formats onto the listening public. I wonder how many people enjoying hi-res have a decimator running anyway and don't even know it. It's really gauche to have people paying extra money for what, in my opinion, they should have access to simply by buying the album; that is, a non-compressed and compressed version to use as they need. If you want to charge for something, charge a bit extra for a surround or binaural version.

 

[watchnerd]

owww my eyes.

 

[gregorio]

  I felt colorful

 
If you ever get that feeling again, please try to suppress it.

 
Quote:

and audiophiles, instead of making a fuss about their digital content, are flying back to vinyl. ... I can't fathom going back to vinyl with its worries about correctly tracking cannons.

 
Most of the time, when I read of some new audiophile fad, I just smile and think "a fool and their money", although I do feel sorry for novices just looking for their first step up quality who are instant snake oil targets. But at least most of the time these fads have extremely little or no effect whatsoever on SQ. Not so with the vinyl fad, which relatively speaking represents a serious degradation in SQ. I thoroughly agree, audiophiles are doing themselves and the audio world in general a great disservice in not demanding better 16/44.1.
 

Totally agree that there was not and still isn't a reason to have foisted new formats onto the listening public.

 
There is a reason, I just happen to disagree with it. The reason is marketing! It allows punters to be fooled into not only paying more for new "hi-res" releases but also to repurchase their existing collection in "hi-res" and of course buy new hardware capable of playing those formats. It's a win-win scenario for everyone ... except the consumer.
 
Quote:

It's really gauche to have people paying extra money for what, in my opinion, they should have access to simply by buying the album; that is, a non-compressed and compressed version to use as they need. If you want to charge for something, charge a bit extra for a surround or binaural version.

 
To be honest, I don't have a problem with people being charged extra for two versions (an audiophile version and an everyday version). It does after all cost more to create two versions but if it were me I would only charge a relatively small premium. One (rather pedantic) point I'd like to reiterate, there are few commercial recordings (in fact none that I'm aware of) which are "non-compressed". In practice it's not really a question of whether or not to apply compression, it's a question of how much compression is applied, as some amount of compression is pretty much always desirable.
 
G

 

[RRod]

  To be honest, I don't have a problem with people being charged extra for two versions (an audiophile version and an everyday version). It does after all cost more to create two versions but if it were me I would only charge a relatively small premium. One (rather pedantic) point I'd like to reiterate, there are few commercial recordings (in fact none that I'm aware of) which are "non-compressed". In practice it's not really a question of whether or not to apply compression, it's a question of how much compression is applied, as some amount of compression is pretty much always desirable.

 
By non-compressed I more meant "not turned into a sound sausage". I guess what I'd really like to see is some thought at all put into the matter of how to use the interwebz to get people the content they want, rather than the current "pay extra for a hi-res version that maybe sounds better, maybe doesn't, we don't ask questions we just distribute" state of affairs.