Misleading Measurements

[John Dyson]

Thought that I'd chime in about the confusion between 'harmonic distortion' and 'intermodulation distortion'.   The term 'distortion' is prejudicial, because not all such nonlinearities are undesirable, even though they (harmonic/intermod) do distort waveshapes INCLUDING that of mixing freq of sine waves. 

 

In common audio terms, it appears that 'distortion' without a modifier is generally some kind of nonlinear amplitude/gain distortion, which can have the effect of 'intermodulatoin' or 'creation of harmonics.'    Both 'intermodulation' and 'creation of harmonics' are kin, where intermodulation can be used to create harmonics.   Perhaps, intermod is the mother/father (maybe aunt/uncle)  of harmonic distortion...  (for example, an extreme noninear gain curve of f(x) = x^2 (which is a distorted gain curve), is pretty much the same as m(x,y) = x * y (which is modulation of two waveforms, where x^2 is directly implented when x=y.)   The modulation example simply gives an additional degree of freedom where two signals are used, but COULD be the same.

 

No need to read much further unless interested in 'how many angels are dancing on the head of a pin' type argument -- because the two kinds of nonllinear signal operations are so related that arguing about differences is almost specious...  Intermod tends to consider mulltiple frequencies/sources, while harmonic looks at only one frequency at a time.

 

------------------------

 

Intermod is mostly thought of as between two signals, while harmonic can be manifested by similar effects as intermod.   In electronics, these effects mostly come from a bend (either constant or dynamic vs time or other parameters) in the gain curve.  Such bends can directly create harmonics (or subharmonics) on a single tone (usually thought of as harmonic distortion), or multiple tones (usually thought of as intermod.)   "modulation" or 'intermodulation" comes from a dynamic bending of a gain curve between different  signal components.

 

Usually, these distortions are fairly well-behaved for simple combinations of signal frequencies and small integer power functions comprising the nonlinear gain curve.

 

In the physical world, there are always nonlinearities in the generating devices (behavior of a plucked string), or resonating components, or doppler effects.   These create harmonics and can intermodulate, and as long as constrained by 'artful intent', aren't really 'distortions'.  These characteritics can come from the same places as 'distortion', but  aren't distorting the results, unless they create an undesirable artifact in the sound.   (Distorting a sine wave might actually be a desirable result, even though the nonlinear process is still sometimes called distortion.)

 

Things get more 'interesting' when the power functions are non-integer, where  fractional power functions can easily create subharmonics (various kinds of parametric schems/similar to modulation can also be interesting in varying ways.)   Because the non-integral power functions tend to create more random-looking frequencies, the distortion becomes less and less correlated (or correlated in an audibly confusing way.)

 

When these intermod distortions become complex (lots of fractional aspects, but greater than 1), the sound devolves into a 'blur' rather than fuzz.  When the distortions are mostly integral, then the power is more strongly concentrated at discrete frequencies, so the sound is more coherent, perhaps more grainy/gritty.   When random looking generated components (from non-integral powers in a transfer function)  below the fundamental frequency  tend to sound uglier, while the higher frequency components tend to blur more.   Basically, the lower sideband of audio intermodulation tends to be grainier/grittier, while the higher sidebands tend to be more smooth fuzz.  This 'fuzz' is one of the causes of 'DolbyA fog'.

 

One generally gets more or less intermod along with harnomic distortion, but depends on the shapes and gain parameteric effects along with the levels of each frequency components.  Also, more 'pure' forms of intermodulation can be created if the various signal components are multiplied together in more controlled ways (e.g. true modulation, gain control, etc.)

 

The 'sound' of electronic and mathematical nonlinear signal modulation can emulate real world effects -- but it is probably tricky to precisely emulate the real-world sound effects.   Whether the intermod happens in 'air' or in a 'string', the underlying math is essentially the same as software & electronics emulations of the effects.  I am NOT claiming that precise emulation of the real world is simple, but I don't think that is what we are worrying about.

 

Software used in testing might use a precise emulation of real world signal components, but might also be a simpler implementation for testing purposes.

 

Bottom line:  intermodulation is intermodulation -- there is nothing special about real-world vs simulation, other than the specific impllementation and parameters used.   Same goes for creation of harmnoics based upon nonlinear gain curves.   'modulation' and 'harmonic distortion' are closely related.  There ARE some fine grained differences, but we even with all of this detail, we necessarily are speaking in generalities here.

 

John

 

 

 

[John Dyson]

6 hours ago, pkane2001 said:

 

So you too hear amazing sound coming out of the built-in laptop speakers? ;)

 

I believe that some people mistake the language 'sound quality' for 'enjoying the music'.   They are NOT the same thing.

I do lots of test playbacks, listening for various kinds of subjective problems (sorry about that, but no objective measurements available yet), and even though the processing might mitigate some technical problems, I might not enjoy the music experience.

 

Sometimes certain music sounds ugly to me, even though the sound is clean.

And -- vice versa.   Sometimes imperfect quality recordings can be enjoyable.

 

Emotions might confuse the thinking about enjoyment vs sound quality???

 

John

 

[John Dyson]

3 hours ago, kumakuma said:

 

Like the fact that the files names Alex used telegraphed the expected results? 

 

🦄

Heh -- I just ran a test where I was getting SUPER-inconsistent feedback on some tests.   So, to control the experiment, I added some metadata to the audio files from a major organization who it is known that I am working  with.

 

Alas -- the initial results were something like:  I'd love to have a copy of this recording -- until I made it known that the metadata was meaningless.   Then, got all kinds of specious criticism about the recording...  

 

The whole thing about 'expectation bias' or simple dishonesty/manipulation can always be operative.   Double blind tests can be trusted, if done correctly.   Even results from people with integrity and some discipline cannot be trusted -- because it is easy for the mind/opinion to interfere.

 

Bottom line:  JUST DO THE TESTS CORRECTLY  -- it eliminates all kinds of troubles.   There are just too many statistical noise sources to shortcut a test.   This is why I quit trusting 'feedback' on tests...   Some things that enable errors is a lack of discipline, but also cannot overlook the matter of manipulation and dishonesty.  Or, lets re-interpret the manipulation issue to 'extreme expectation bias' :-).

 

John

 

[John Dyson]

1 hour ago, Confused said:

Reading the recent posts here, and mindful that this is an "objective" thread, it occurs to me that the bit perfect files sound different idea could be subject of a controlled blind test, similar to the ones performed recently by @Archimago (although I suspect he might not be interested in this one) and Mark Waldrep.

 

Maybe something along the lines of a number of sets of four music files, all bit identical, one pair saved by method A, another pair by method B.  The test would be to match the pairs.

 

Just one of my idle thoughts.  I might be wrong, but I suspect that nobody would be motivated enough to set this up.

 

It would be fun if somebody did though, and who knows, it might just provide some interesting results.

The results might be interesting, but the laws of physics&math&computing limit my interest.   I have done too many things in my engineering life where bit perfect means bit perfect, with all of the associated ramifications.  Stuff really works that way.

 

There is an 'out' for the metaphysics advocates about real-time play of a flac vs. .wav file with exactly the same content, but that is all about analog noise sources and not-so-good analog layouts.   Those kinds of comparisons have nothing to do with the payload content though.

 

I don't think that many people who are competent enough to do such an experiment would be interested in (wasting time) doing it.  If the stats would be done correctly, physics and math show that the results will be the same.   Any 'signal' that shows a difference will be deep into the statistical noise.  Perhaps, though, maybe someone competent enough to do it, and  had time to totally waste, or they were paid to do it.....?

 

John

 

 

 

[John Dyson]

52 minutes ago, manueljenkin said:

1. Audibility thresholds and weights is still and active area of research. So nothing can be analysed conclusively at this point of time. So if the claimers don't have proof, the same way the naysayers don't have proof either (until you solve cognition).

 

2. One can form any correlation they want based on experience, since the measurements donot seem to show the whole picture. Proving it, is unlikely, agreed. Doesn't mean it exists, doesn't mean it doesn't exist. It is uncertain, it's up to the audience to try and come to a conclusion. No one has the right to opinion bully them to "not try/explore".

 

3. I don't really know what you're trying to say here. You're saying we shouldn't doubt anything, and should have stayed with the 14th century thought that earth is flat? See point 1 again, has the answer to "prove every claim"? The claim that they cannot make any changes to sound quality is not proven either.

Any good engineer that I know (and there are some on this list) will suggest double-blind or other such variant of tests that mitigate the effects of personal & external bias in measurements.   Some more specific cases, we have laws of physics involved, and such problems are often not very interesting to do measurements.  (If the laws of physics&math aren't obeyed, then why do measurements or tests anyway?)   When it superficially *appears* that the laws of physics are violated and tests appear to be properly done, then there are engineering/technology problems that biased the results.   Technology can be a problem, but math about bits isn't.

 

You should never get an argument from a true engineer or scientist about 'sounds better' -- that cannot be measured except by someone who is honestly listening.  However, there are objective measures&methods and most often there are good ways of testing for differing levels of technical quality.   Without the techniques needed for real engineering solutions, then we get  much too many long random walks of design (which I have recently been involved in) or misleading claims.   Sometimes 'real enginering' in specific situations  is difficult or impractical, but very little discussed here is resistant to good engineering and good science (except for sounds-good).    Stuff is already difficult enough to do  without adding the additional variable of prejudice or feeling.

 

Of anyone who posts here, and those of us with a  generally have a good engineering mind, I have been mixed up the worst with a design that is very difficult to measure -- and it is hell.  Now, I even more strongly regret that EVERYONE doesn't try to adopt scientific methods than before my project-from-hell.   Strongly held non-objective beliefs causes people to hurt themselves with faulty decision and research methods.   Emotions & feelings need not apply -- even though they will always be involved SOMEHOW.

 

The test about 'bits not being bits' (for example) is of no interest to a scientist, is because EVERYTHING will fall down and fail if 'bits aren't bits'.  

 

There is space for subjective evaluation (e.g. sounds better), and objective techniques (design improvement, testing, verification, etc.)   Since I am an engineer (almost to the core) and I know that my own listening is not a very reliable measurement method, I try to use scientific techniques when possible.   It would also be a good thing if more people would realize that their own hearing isn't very reliable for measurement purposes.*  IMO, avoiding scientific methods on purpose is masochistic - or perhaps wish for a forever hobby of tweaking.

 

* I AM convinced that there can be individuals who have trained their hearing to be accurate, but such training and testing also needs to be based upon scientific method.  Just using experience for a basis of such training is probably not sufficient.  It seems that there must to be a scientific basis somewhere , even if the technique appers to  have elements of being subjective.

 

John

 

[John Dyson]

8 hours ago, opus101 said:

 

As another engineer here I disagree that measurement is the foundation of science. This would appear to be a dogma held by 'objectivists'.

 

I completely agree that 'my own listening is not a very reliable measurement method'. When I listen, I am making an observation, not a measurement. An observation is fundamentally a qualitative thing, not a quantitative thing.

Measurement isn't the 'foundation' of science, but is certainly needed to do science more reliably and more efficiently.

 

Measurement can take all kinds of forms -- some more reliable than others.   When a measurement system has poor accuracy/reliably, it will almost guaranteed lead one down rabbit holes.   I have been chasing rabbit down rabbit holes for 9yrs on my project, and eventually I have found a good solution.  If I had (or could find) a reliable measurement mechanism and/or a good design spec, then there would have been much less pain.   Of course, someone else would have figured out the solution if the problem was easy to solve.   Few engineers/individuals are as persistent as I am -- but that is also why I got the 'good/difficult' projects at the labs.

 

I am not espousing dogma -- just the opposite, my opinions come from and have been re-enforced by hard earned results and lots of learning experiences.

 

John

 

[John Dyson]

7 hours ago, opus101 said:

 

I'm not saying measurements don't play an important role in science, of course they do. But I am saying they're not foundational, observation is foundational.

 

'Hard science' is used here because you believe other (non-hard) kinds of science aren't science? I'd say some kinds of science are characterized as 'hard' because of their heavy reliance on measurement so the intersection of 'hard science' with 'measurements being not a key method...' is indeed the empty set.

Measurement is a critical subset of observation.   Biases/imprecision creep in with the typically less disciplined, relatively ad-hoc  'observation', therefore it is best/easiest to depend on more reliable, intrinsically more accurate objective measurement when at all possible.   If one really reads what I write, my recent experience includes both measurement and the more general observation.   Voluntarily choosing a more general, less solid form of 'observation' over and above accurate and reliable objective measurement can (will) be wasteful of time and effort.   I do suggest that *probably* because of a lack of expertise/knowledge, many people will choose the looser 'subjective' observation instead of the more clean, accurate 'objective' measurement.   In fact, I have used my own experience as an example, where there was NO OTHER choice but to use 'subjective' observations because the 'objective' was simply not available at all, to ANYONE.

 

If easier, better measurement methods were available, my project would have been done/completed almost 20yrs before, by someone else.   Using tools beyond that of our native senses requires more thinking and analysis that is often much less intuitive.  It is sometimes INITIALLY more difficult to be disciplined, but in the longer term the results can often come more quickly.  (Believe me -- a lot of math is very UN-intuitive, and some of that non-intuitive math is just under the hood of a lot of electronics design - but also overlooked by hobby level and even some pro developers.   Some of those non-intuitive aspects can make problem-solution much easier in the long term, and that is why I was sometimes used as an answer-person back at the labs.)

 

I give my input as a result of painful experience, even though as a very effective engineer, I already knew that it is best to avoid time wasting exercises.  An engineer doesn't get good feedback in their job if they dawdle too much.   Just because it looks like someone is doing a 'science project' doesn't mean that it really needs to be one.  ('Science project' here is used in a prejudicial way.)

 

 

[John Dyson]

8 minutes ago, opus101 said:

 

Yes, I understand that but I disagree - substituting measurement for (admittedly subjective) observation leads to reductionism. The solution to biasses in observation is more impartial observation, not measurement. 

Huh?  Did I write that?   I am for both, but you'll waste lots of time (per experience) if you live in the world of the subjective.   Maybe some people like tweaking -- I don't.

Observation INCLUDES measurement, but observation using normal human perception is naturally VERY susceptable to bias.   If you have a choice, you'll generally waste a lot less time with objective measurement.    I have visited WAY TOO MANY rabbit holes because of work on the edge of solid technical knowledge, and it is a good idea to avoid those visits.

Therefore -- when you have a choice, be careful about your human observation when it doesn't match a measurement.   We see this problem all of the time in the mostly hobby audio world -- too many 'observations' that don't match mathematical/physical reality IN ANY WAY -- EVER.

Sometimes, once in a long while, a human observation might be a good hint that there MIGHT be a problem with the measurement system, but that VERY SELDOM happens in this realm of 'very established' science.  (again, we aren't talking about 'sounds good', but more like objective equipment evaluation.)   Most often, the biases in observation overwhelm the perception of reality.   Do your best to avoid biases.   In this hobby audio realm, we are so far away from the edges of scientific knowledge that, if there is a choice at all,  measurements will be the winner.

 

I don't think that we are talking abstract theory on the edges of science or truly difficult problem solving, are we?  Aren't we talking about estabilished science with good measurement techniques?   Also, we are not talking about 'sounds good' -- that is purely subjective and outside of the scope of objectivism.

 

John

 

 

[John Dyson]

6 minutes ago, opus101 said:

 

The voltmeter doesn't produce an 'observation', its a measurement. I would of course agree, using a tongue to 'observe' the state of a battery is inapt.

The act of reading an accurate, good quality voltmeter used competently is also an observation.   A tongue makes a very inaccurate voltmeter, and if you try to use it for measurement of something that needs accuracy, the you'll be visiting a lot of rabbits.

 

[John Dyson]

8 minutes ago, opus101 said:

I certainly have spent (I wouldn't say 'wasted' myself) a lot of time in the world of the subjective. There aren't any short cuts to putting in the hours in the world of the subjective. I conjecture (but have no evidence for) that substituting measurement for subjective observation is an attempt to find a short cut, an attempt to avoid 'wasting' time. 

When you use a well considered 'measurement', it often does require more of an intellectual understanding of what is going on.   A quickie human observation is easy, but not always so accurate.   A subjective observation can be a double-check, but not much else, that is-  otherwise can be useful, if there is no other resource.   Measurements are not just on the exterior/outside of a design, but also in the interior.   Using subjective listening (for example) limits the domain where you can observe.   Not all designs are linear througout -- a measurment will give much more information.   Processing that greater amount of info does require more intellect,  and IS often an important shortcut to finding an answer.   (I am not using the term 'SHORTCUT' in a prejudicial way, but instead using it as a synonym to efficiency.)

 

[John Dyson]

2 minutes ago, opus101 said:

 

I think it makes sense also to turn this around. To wit :

 

When you use a well considered observation it often does require more of an intellectual understanding of what's going on. A quickie human measurement is easy, but not always so accurate.

I agree --in our discussion a quicky human 'measurement' is the 'tongue' technique.

 

So, it all depends on how you define measurement -- I tend to see that (measurement) as with some clear 'measure'.   The using the 'tongue' as a measurement device, 'tongues' measure isn't very precise or stable.

 

Likewise, an observation of a meter isn't very useful, without competent use of the meter.  (Using the term 'observation' as the result of human perception instead of a more general term that includes measurement as a verb.)   It is this general inability to competently use the meter, understand the measurement context/circuit where missing knowledge can give troubles.   Just reading a meter seldom needs much training though, even understanding what is going on 'underneath the needle' isn't very intellectually challenging.   It is this lack of understanding (not just reading a meter) that pushes a non-technical (or even inexperienced techie type) into an expedient subjective 'listen'.  I am all for helping people to understand what is really going-on under the hood, but that kind of learning can be time consuming (or even all-consuming.)  Electronics has so many layers, and it takes a few layers down to really understand what is going on in a most general and adaptable way.

 

It is all about trying to find the best reliaiblity or resulting stability & quality.   In the equipment realm, there are few places where a well considered measure/observation of an aid (device/meter/data-processor) isn't more accurate and reliable than just using the subjective hearing/listening mechanism.   This is especially true because FEW people who even consider themselves experts have actually calibrated their hearing and trained it based on hard core physical references.   Those who have, and trained themselves carefully to avoid biases, which would also include testing, could POSSIBLY attain some level of reliability, but feelings will almost always creep in.

 

John

 

[John Dyson]

21 minutes ago, opus101 said:

 

I define a measurement as a subset of observation that produces a purely quantitative result. I.e. a number (or series of numbers). If the result is qualitative then the operation wasn't a measurement it was an observation.

 

The use of a tongue on a PP3 doesn't produce a number hence can't be classed as a measurement, at least to my way of seeing things.

I try to make sure that people who depend on their observations or measurements truly understand the quality of their results.   Low quality results can result in many rabbit chases.

 

Just depending on the human body's stability doesn't bode well for the results.   Subjective evaluations result from many variables, therefore making it difficult for any kind of serious precision.   Subjective is okay for verifcation, but not precision.   Geesh, subjective viewpoints are even somewhat dependent on metaphysics.

 

[John Dyson]

3 minutes ago, opus101 said:

It looks like we are using 'reductionist' (or 'reductionism') in rather different ways.

 

I would agree theories do simplify, or reduce things to simpler things but they're not 'reductionist' in so doing.

Reductionist seems prejudicial...   There are many times that one cannot 'divide and conquer' in a design -- but that often becomes MORE of an intellectual challenge rather than over simplifying into subjectivism...   Being subjective when not absolutely necessary  is like 'giving up' on actual engineering.

[John Dyson]

5 hours ago, John Dyson said:

Reductionist seems prejudicial...   There are many times that one cannot 'divide and conquer' in a design -- but that often becomes MORE of an intellectual challenge rather than over simplifying into subjectivism...   Being subjective when not absolutely necessary  is like 'giving up' on actual engineering.

My comment about 'divide and conquer' not being operative is very important.   Some designs are easily divided and conqured, but make a minor architectural change or a structural change, then the result can be a morass of interdependencies.  Sometimes these kinds of changes are more of an intellectual challenge, but there is great benefit in  'solving the problem' from a totally rigorous standpoint.   Absolute rigor is sometimes not practical, so I claim that the answer is NOT to give up and devolve into a subjective 'tweak until it works', because this becomes the 'roomful of monkeys' scenario.   There are often hybrid approaches, where the structure can be predicted, but the math is impossible (or nearly so.) 

 

With the subjective design using the roomful of monkeys, then the project will never be completed.

With the requirement of total rigor, then the project will not be completed for different reasons.

 

Sometimes a hybrid approach is a good thing, but in my own necessarily hybrid approach, avoiding the subjective has been wise.   When trying to use other people as data inputs, using their subjective abilities, then I often ended up with more noise than data.    Dunning-Kruger effect is very operative in the audio world, probably other psuedo-engineering and psuedo-science activities also.

 

My resistance to ad-hoc, tweakabily approach is well considered and comes from REAL experience.   Resistance doesn't mean total rejection, I tend more to wisdom instead of religion* in my approaches.   My only 'unwise' decision was to do my project at all, but the result has been well worth it.   Sometimes doing an impractical thing is worthwhile, because in the end-game, it wasn't really impractical -- it was invention.

 

* There are better applications to 'religion' other than psuedo-science and psuedo-engineering!!!   That is, I am not anti-religion in general!!!

 

John

 

[John Dyson]

1 hour ago, fas42 said:

As being discussed, observation is the start of achieving understanding - unfortunately, philosophical as well as ego factors cause some to downplay, or ridicule other people's observations ... and science is always the loser here. Insistence on measurement is used as a weapon by one side, in the hope that 'uncomfortable' observations can be made to go away ...

The problem that I have recently seen -- ego is often messed up with the subjective observations.   Getting away from the subjective also helps to mitigate the expectation biases and the Dunning-Kruger effect that seems to be so associated with  subjective observations.   The only real answer is proper scientific and statistical methods, sometimes even simple applications of scientific method can expose the personal and ego biases that are oh so associated with being subjective.   I have recently done some simple experimental controls  to expose a severe problem with certain subjective reviews.    Subjective must not be the 'default', but instead objective -- then use subjective as a cross check.   Unchecked subjective review is almost the same as relying on a high priest, and is even less valuable than a random choice -- because there are sometimes negative biases.   In actual development and invention, ubjective evaluations really need a competent statistical wrapper, or they are just opinion -- and we all know about opinions.

 

If someone else wants to chase rabbits and play whack-a-mole, then so be it.   I strongly counsel that using subjectivism with double checks and controls should be secondary.   Objective methods with double checks and controls (e.g. controlled experiments) should be the first choice to avoid wasting time.  Subjectivism without controls (something like scientific method) should only be used in the realm of 'sounds good' and that is it.

 

Some people, however enjoy tweaking as a hobby -- being a tweakabilly isn't my goal, but my goals have usually been pragmatic results.   I won't take anything away from people who enjoy the endless 'tweak', but I have useful things that I prefer doing.

 

 

[John Dyson]

7 hours ago, sandyk said:

I agree, and John may not be aware that you are also well qualified technically, not just anotherSubjective type.

Sometimes, there are people who are technically oriented and enjoy tweaking -- I don't.

In order to move technology forward (rather than copying schematics), there isn't much room for wasting time -- therefore use objective as the driver, and subjective as the cross check at best.  (Subjective does need the stats wrapper.)

[John Dyson]

9 hours ago, sandyk said:

 It hasn't always worked out that way in your PM group though, has it ? 😉

I will leave it at that, and not reply further, as we are so far off topic already. with virtually no poster  giving OBJECTIVE illustrations , whether by later refuting measurements or the results of DBT sessions that showed the measurements in the example given were misleading instead of the philosophical discussions  currently  .

 

 

  

My focus is on informing about the benefits of avoiding short term laziness, and spend some time doing what is needed for the longer term benefit of proper stats procedure and objective measurement.   In the longer term, the investment in more structured approaches will most often give a great payback.

 

Answering comments about my very necessary usage of the subjective (but I imposed controls that might not have been obvious):

 

My opinion about the weakness of subjective input is re-enforced from the PM group.   Sometimes subjective input is the only way to do things as I had noted before, but not very efficient and VERY noisy.   I counsel people on the best approach, but have been greatly frustrated by the statistical quality of subjective opinions.   It isn't to say that the subjective is to be ignored, but the data signal is weak and the noise is too strong to depend solely on unprocessed subjective inputs.   After some simple experimental controls, my opinion of objective data has recently been further diminished.  Unfortunately, I have necessarily spent lots of time with chasing rabbits -- and I suggest NOT to do that.   Relying too much on the subjective, esp without proper experimental controls, is tantamount to technological masochism.   I have sometimes needed to use the subjective on my project that had a similar effect as an act of self-flagellation and gotten the same results -- misleading.   All I can say -- only use inferior data sources like the subjective, if you really must.

 

If I wasn't pushing the bounds of what had been done before, then less subjective input would be needed.   We have learned a lot through the years about what needs to be measured on simple amplification and transducer equipment (well -- transducers aren't always simple, but we understand them very well.)   Dynamics processing is trickier to understand WRT impairments.*   Things can be very different and  subjective, noisy sources might be needed for verification, esp when chasing something WITH NO SPECS AT ALL.

Even  a very complex control system -- there are often parameters where the measurement of typical complexity needed for a linear system would not produce adequate input.  A well defined complex system could be measured (certainly avoiding the dependency on the subjective) by proper data processing and reduction.  I am afraid that some people might choose the subjective for a very near-sighted shortcut, but long-term costly decision.

 

* the linear processing is such a small part of the project, but still the HW variant can benefit from the well known techniques just the same.   If working in the HW domain, would definitely use the more reliable objective than just being a tweakabilly designer.

 

For the dynamics processing, unlike most compressor/expanders, the attack/release is multiple continuous nonlinear functions, and combinations of complex dynamics controls.  The dimensionality of any quality measurement on my project is huge.   Typical linear systems like preamps, amplifiers, even tuners have much less complexity than my project just wrapping up.   One thing for sure, I trust the quality of any attainable objective and controlled measurement much more than any subjective results that I have measured or have  been provided.   There is NO-WAY that the subjective could have given accurate response curves as created from my objective test materials.

 

For most anything done that IS NOT pushing the boundariess of technology, then take the very wise short cut, and use the objective technological means and data reduction methods that you have available, then go to the subjective as a cross check (make sure there isn't something obvious that you have over-looked.)   Subjective review is great at finding obvious errors, but I have even found that to have real troubles with errors and bias.

 

John

[John Dyson]

3 hours ago, pkane2001 said:

 

If you want to use your ears as a measuring instrument, then you should use it properly, as any other scientific device. Calibrated, tested, with known error margins.

I agree so very much.   Many of us cannot maintain a good, accurate comparision longer than a few seconds -- and drift off into 'opinion' very quickly.   Even then, hearing isn't going to have very good absolute accuracy in general.   I don't know how many people here have sat down and spent several hours per day while training their hearing based upon objective measures....   I  doubt very many of us has done that.

 

Just having 'experience' isn't good enough to claim accuracy, it really DOES require calibration -- I recently have dealt with LOTS of listening opinon, and even though helpful, it has definitely not been definitive.

 

On the other hand, I believe that subjective double checks are definitely okay, and in fact very beneficial -- but also not adequate in any way.   This caveat about hearing stability includes my own.   When in a casual situation, I have not been as careful as I should have been either, and I do KNOW that my hearing isn't reliable.   Sometimes, that is all that we have to test with -- and that isn't good.

 

John