Can Bad Recordings sound Good?

[John Dyson]

28 minutes ago, Audiophile Neuroscience said:

Can bad recordings sound good?

 

This seems to be a recurrent theme across some threads eg

  

 

 

 

 

My view is that over 90% of "Hi-Fi" is in the quality of the recording. GIGO.

 

The better the reproducing system the more transparently it will reveal what sounds like real (unamplified) sounds in a real acoustic space.

 

The better the reproduction system the more it will reveal good and bad bits of a recording.Good bits of bad recordings can make the presentation sound less unpleasant, indeed pleasant if one is able to mentally tune out the bad bits.One can rediscover many old recordings in this way. The bad bits however are still revealed and no amount of system tweaking will overcome this if transparency is maintained.

 

Tweaking a HiFi system to make the bad bits of recordings sound "good" (less bad) = coloration. All recordings start to sound of the signature of the color chosen and one may gravitate to certain recordings that suit the color. In essence you convert a HiFi system into a mid or more likely Lo-Fi system. Radios and car stereos can sound 'good' with bad recordings because of the information discarded - you end up with a truncated, compressed, music-in-a-tin sound. This is fine to get the gist of the melody and rhythm especially for familiar tunes.

 

The other biggy with tweaking of course is the possibility of confirmation bias. But the emperor has no clothes if nobody else perceives it.

When you go beyond making a consumer system as technically accurate as possible, then you get into the realm of re-working and re-mastering.  A little shelving to correct a recording, or more serious stuff to correct the speakers -- that is still in the realm of proper reproduction.  Some recordings also have a repeatable kind of damage that can be at least partially corrected.  However, when a consumer does things with tools that are normally available to the audiophile, using a locally corrected system -- then coloration is about all that can be done.   A consumer who does more than EQ for a recording, and perhaps a small amount of 5BX-type expansion, then we are talking about correction and remastering.  Without proper tools, the correction and remastering is futile.

 

In my project, I have found that 'rework' can be harder than hell, because of the extreme variability of recordings.  Almost the only thing that you can generally rely on -- a given album TENDS to be similarly damaged.  Series of albums from a given artist tends to be more similarly damaged than between different artists.  However, sometimes you do find cases where individual selections have also been tweaked.

 

I have respect for those who clean up a recording, that might be beautiful, but dependent on ancient technology - there will be very-difficult-to-correct damage, but mostly well correct.  Small defects in media, transducers,etc can hide a lot of these sins. 

 

A consumer without specialized hardware and software cannot undo the major kind of damage on pop recordings, however there is LOTS of properly mastered material also -- but the label won't be 'Atlantic' or whatever major labels that are available today.  The label would more likely be a specialty one, and some super picky person will do the mastering, and not allow the distributors to put the bad-touch on their recordings.

 

Unless using very speciality software, correcting a damaged recording might be fun to try, but the results will not be as good as possible.  Speciality software of various kinds CAN do a lot to fix old recordings.  Trying to fix the recordings without specialty HW or SW - just coloration & tweaking.

 

John

 

[John Dyson]

12 hours ago, Audiophile Neuroscience said:

I agree John. So, in a nutshell, forgive me if an oversimplification, some bad recordings can be improved by fixing aspects of the recording itself. This is difficult and not for the inexperienced working with poor tools.

 

The only thing i don't get is " or more serious stuff to correct the speakers". Do you mean DRC?

What I meant about speakers is serious kinds of dsp EQ.  Also, a super serious hobbyist might design his/her own speakers with negative feedback schemes/etc.   All of these aggressive schemes are intended to reproduce the electrical/digital signal accurately, but not MODIFYING the signal.

 

Modifying the signal other than simple EQ starts becoming mega complex.   It is like my project -- few EE/DSP people could even start it without a lot of mistakes.  (Like me, I made LOTS AND LOTS AND LOTS of mistakes at the beginning. I am *still* correcting my mistakes, and it is NOT a financially rewarding thing to do.)  'Correcting' the recording is infinitely more difficult and specialized than making a reproduction system work super well.

 

John

 

[John Dyson]

3 hours ago, Allan F said:

 

And I respectfully disagree with you, at least with respect to the degree of certainty you appear to proclaim. You are expressing the point of view of a practicing musician with a "tuned ear". While many aspire to that level, whether they can achieve it is an open question. You assume that most audiophiles have heard their favourite artists live which, desirable as it may be, is also questionable. Moreover, even if they have, the nature of the venue and its unique acoustics may play a more important role in the characterization of the sound than the identification of the instruments themselves, especially if any amplification is used.

 

"It is the difference in opinion that make horse races."

(Not really answering the person I quoted, but showing general frustration about these kinds of discussions seldom coming to consensus)

 

Trying to discuss deviations from accuracy does require a commonly agreed upon language to discuss -- other than just creating a per-discussion blather (I use that term often -- for when the intent is sometimes not communications, but just making statements or unclear claims.)  Participating in the discussions can be frustrating.

 

Being totally accurate, and not pendatic: If anyone even talks about most pop/country and even some classical recordings providing anywhere near 'pure' sound -- stop there, because most of those recordings cannot even attempt an accurate representation of a performance or even not reproduce the original mixdown, without special clean-up.  I get so frustrated when someone is worried about even the truly BEAUTIFUL possibility of 1% actual distortion in a speaker system, yet doesn't consider the effective 20% distortion (or worse, relative to actual relative levels) in most available recordings.  I am not just speaking of THD or IMD, but there are other kinds of distortions which are other deviations from relative accuracy -- even 'distorted' relative to the artistically created mixdown at the original recording.  (These are technically demostrable facts.)    If the distortion manifest isn't the more common one started at the beginning of CDs and digitial distributions, there were still issues with making changes during mastering -- but not quite as heinous, it was more of a media issue before CDs.

 

I want people to have better access to clean audio, but I get frustrated over the futilityof these kinds of discussions -- there ARE very important issues that can/should be discussed about music reproduction -- and indeed 'sounding like' the real performance is an important issue.  There are SO MANY steps inbetween, where esp nowadays with moderately high quality equipment, the consumers sound system a the SMALL part of all kinds of distortion (excluding the transducers/speakers/listening space.)  Even then, getting an accurate copy of the performance requires a media source that has the correct information -- but most do not.  There are some esoteric recordings available, that haven't been twisted badly -- those aren't the music that I normally listen to (even though I can respect and appreciate the results.)

 

There ARE real problems with the sound getting to the ear from the performance -- but the most common audio problem isn't the consumers system (unless the system is something like a boombox.)

 

John

 

[John Dyson]

10 hours ago, gmgraves said:

Your ear/brain might adjust quite unconsciously to this 'imbalance', but mine rejects it with a “Life’s too short to put-up with lousy recordings” and simply will not accept such recordings. And your insistence to the contrary, no system can “fix” a lousy recording. Nothing can properly “uncompress” a badly compressed recording, not even a fully adjustable DBX unit - I’ve been down that rabbit hole before. Nothing can “undistort” a distorted recording. No system can fix a recording made with a forest of microphones and a myriad of tracks to allow it to sound like a real stereo recording with a real soundstage and actual imaging. Nothing one can do will remove the non-linear distortion in the treble region that people call “overly bright”. Nothing can remove layers of veils of distortion and phase anomalies that plague so many poor recordings. And no system can fix (or any ear “listen around”) a Colpix soundtrack recordings (“The Long Ships“ or the soundtrack to “Phaedra”) for instance. These records sound like they were recorded acoustically with “Little Nipper’s” horn feeding a nail being dragged through a wax cylinder! Why you keep insisting that a “properly set up playback system” can overcome these abominations of studio incompetences, is beyond me. Everybody knows you’re full of crap, here Frank, and that you and your “method” have no credibility (especially on this subject) and it makes me wonder why you continue to foist this ridiculous charade on the rest of us???!!!!!

 
 

 

About not being able to 'undistort' a recording....  Once a signal is compressed by a 'loudness wars' DRC compressor, it is impractical to remove the modulation distortion, and matching the compression curves is impractical also.   As a compressor gets closer and closer to inf:1 (whether dB linear or not), matching the compression curve with expansion really does become impossible.

 

I did listen to a copy of the Adele 21 thing, and the compression that I heard and measured was on the order as strong--pe9k=4,-1.5 as the ABBA "The Complete Studio Recordings), but actually sounded worse.   I cannot even guess there there would have been 'artists intent' with Adele 21, and that is very sad.   ABBA music was actually planned to be playable through an AM radio recognizing that it would be further damaged -- the TCSR actually does sound 'okay', it is just that I personally don't like a lot of compression.  (Tretow seemed to know what to do and what not to do to make the ABBA stuff sound good on AM radio.)   Adele doesn't make that kind of music, and some dynamic range seem to benefit her works.

 

John

 

[John Dyson]

--pe9k=4,-1.5

43 minutes ago, sandyk said:

 

My system is very transparent and quite low distortion, and like Frank I am left wondering if something in your system may have been emphasising this problem further ?  Yes, it is a bright and dynamic sounding recording with a  wide stereo image.

 Perhaps John Dyson can verify if it has an improperly decoded Dolby-A problem.

 

The brightness added by DolbyA (even when it is used in its unequalized form) isn't created by strong amounts of compression.  DolbyA is damned fast compression, but not HARD compression.   After a signal gets above -10dB, then DolbyA is done with its job.  It does nothing above -10dB at high frequencies, and nothing above -20dB at 100-3kHz.  (Below 80Hz, it is somewhere in between.)   So, if you hear a really hard, peak-limited style compression, that is most likely not DolbyA by itself.

 

They CAN use DolbyA to brighten material artistically, and it was used on Karen Carpenter's vocals, but it has to be done intelligently with some EQ during mixing, because DolbyA will tend to overly brighten the sound.

 

All I know is that I tried to 'decode' Adele21, and nothing that I could do would repair the recording...   The music might have been enjoyable, but the compression and even compression distortion was distracting to me -- very distracting.  I'd have to have already been an Adele listener to have appreciated the music.

 

John

 

[John Dyson]

Okay, if the subject changed to the 'Locomotive' thing, when listening, I heard the woody midrange that is my now strongest indicator of FeralA.  I just did a quick online/realtime deocde, and it shure sounds FeralA.   Below is  a 30 second snippet of a decoded copy.  BTW, the basic compression used before FeralA DOES sound like artists intent.  The stats coming out of the decoder do tell me that there ARE dynamics in the sound.  My guess is that there is compression before mixing, and some careful mixing going on.  I don't think that the material is 'loudness wars' compression of the whole thing.

Outr-demo.flac

[John Dyson]

27 minutes ago, Confused said:

 

No it does not.  Plus for the record (accidental pun), I have listened to plenty of live music, both acoustic and otherwise amplified.

 

So, I am never bothered by treble energy in live music, but I am on occasion bothered by overly bright recordings.  I am pretty sure I am not unique in this regard either.  Plus I guess treble energy in a live performance could be annoying if it was purposely designed to be so.  maybe I could start a band, the fingernails on a blackboard trio, or something?  But no, I cannot ever recall being bothered by treble in a live performance.

 

As for if it is the linear distortions in my rig or whatever, we will never know.  Obviously systems can be optimised in this regard, but this fact does not logically mean that it can never a recording issue.  What I can say is that when I do find a recording that sounds too bright, it tends to sound too bright in my car, on my headphone rig, my main system etc., whereas recordings I do not find bright tend to sound OK on everything, in this regard at least.  Using the power of logic, it would appear most likely that the irritating thing is in the recording itself.

 

In terms of the track I posted earlier, one person found it bright, another found it too bright, and I found it too bright.  There really is nothing controversial here.

Yeah -- about live music...  It is really hard to describe the intensity of a live instrument in person, instead of even GOOD recordings.   Most complaints about 'too bright' (unless the material really is too bright) come from people who haven't been in person or heard raw/live audio through headphones (which can be a close second to real life.)  By the time all of the dispersal in most all normal speaker systems, some of the liveness is lost.  I am NOT claiming that a good speaker system cannot come close -- but normal speakers and normal power levels -- hard to get that real, live intensity.  It is even difficult to capture the liveness on a recording, but some people specialize in that.

 

My own headphones have an upper midrange peakiness, and I do not like the intensity of material that has lost its time coherence -- it can indeed sound ugly.  That 'time coherence' and stealth modulation distortion are important drivers for the DHNRDS decoder project....  It is really important, and very difficult stuff.  I cannot say that I have achieved perfection, but is a lot better than the alternative.

 

John

 

[John Dyson]

8 minutes ago, Confused said:

Thanks for posting this John, interesting.  
 

So, as an open question, how does anyone find this version versus the one posted by myself or @sandyk in post #205?

I listened to each one, remember that I took exactly the opus version of what you posted, converted to flac and decoded it as if it was FeralA, which it is.

 

The differences are, just listening, no prejudice (my opinion):  The original has compression, stronger midrange, slightly brighter all the way across, but is missing the clarity of the vocalization approx 10seconds.

 

The original has a slightly smeary 's', in 'thats right', exactly as expected from EQed DolbyA compression.   Doesn't sound bad.

 

Alex's version does sound different, almost like a 1/2 way FA decode...  Doesn't sound bad.

 

The decoded version has stronger dynamics, the vocalization sounds fairly natural (as natural as the recording is)...  You can hear down to the ambient background (almost). Doesn't sound bad.

 

I tend to prefer the stronger dynamics, but many people are accomodated to the ubiquitous compression (even if audiophile,)  I can not nor will judge which 'sounds better', but which one is closer to what was mixed down?  probably the decoded version.  Probably not a perfect rendition of the original mix, because there are variables, including I don't know the target sound, so I guessed with default decoding parameters.

 

John

 

[John Dyson]

16 minutes ago, Allan F said:

 

Since you NEVER listen to them, your generalized hyperbole about pop music recordings would appear to be nothing more than ignorance based opinion formed primarily from your well known extreme distaste for the sound of electric guitars. As with every genre, there are recordings that sound good and others that do not. Having said that, I would certainly agree that the incidence of poor and/or overly compressed recordings from the major pop music labels is more prevalent than for either classical or jazz recordings.

I'd rather not judge anyones music taste.  I just decoded some Oasis, and it sure sounds dissonant to me, but it is music to others.

I understand the frustration about a genre that I enjoy being looked down upon, but I'd rather think that is is a matter of individual taste and interest, not a value judgement.

I have heard SUPER-WELL-RECORDED hellacious Classical, but not all classical sounds bad to me -- again, it amounts to personal taste, and NOT relative value or values at all.

I do believe that we should respect each others enjoyment.  On the other hand, if someone imposes dissonant, chalk-scraping noise against my hearing, I WILL complain :-).

 

John

[John Dyson]

Just now, sandyk said:

 Hi John

 The only difference with  my version is that it was directly sourced from the YouTube .mkv container as a 529kbps .aac file ,which most of their videos now appear to have, although in recent days I have seen some .mp4 videos where 529kbps audio can also be demultiplexed from using Video S/W.

It would appear that if people want the higher quality audio inside YouTube Videos that they may have to pay for it.

Thanks for you input.

 Kind Regards

Alexf

And decode it, if they want the original dynamics.  Once in a while, I have found material that isn't FA.  This makes me wonder if the 'fake DolbyA' compressors being sold today are intended to sythesize the ubquitious FeralA sound character instead of for vocal enhancement alone.   That is, is there a market for the sound qualities that were originally deemed (the ugly digital sound) back in the middle 1980s'?

 

John

 

[John Dyson]

1 hour ago, sandyk said:

 John

 Originally you were finding numerous examples of  tracks from different albums used in their compilation albums where it appears that with an occasional track they may not have had access to the original master, so used an internal Dolby-A copy.

However, it now apears that there are numerous recordings that must have been deliberately encoded this way, not just due to negligence, perhaps to make them sound more acceptable on AM radio as well as perhaps FM stations that used auto programming using .mp3 carts.

 

Alex

 

 P.S. 

 Has anybody tried my suggestion of using the freeware 4K Video Downloader (Windows) ,  saving the .mkv version and seeing if their existing gear can play them directly ? JRiver 26 is able to do this.

4K Video Downloader is also capable of downloading a complete play list in a fraction of the time that it takes to play them.

 

Actually, they almost always have access to the original masters, or something similar (unless destroyed by fire), but don't know the calibration for the tapes, or other EQ that was lost.

There is still another issue about old material, burnt up in fires, and recovering a high quality copy from FA.

 

There is ZERO benefit for doing DA with EQ encoding for loudness, it is a ham handed method, and frankly, the processors used in AM/FM stations since the middle '70s are FAR more sophisticated than what DA can do.

 

(My intense language is not intended to insult anyone, but to be vehement and strong in my statements)

 

If anything, there is a negative benefit, because DA doesn't really do what is needed to increase loudness AT ALL.  DA only increases the density a little bit, and creates an ugly (woody) sound.  IT DOES LITTLE FOR AVERAGE MODULATION.  Also, AM and FM need significantly different optimizations -- DA is a break-even for broadcasting.  Broadcasting isn't just a poor reason to use FA, it is a STUPID reason.

 

Reason for FA:

 

However, I think that it has become clear that the major reason for FeralA is actually more insidious than I originally thought, because I couldn't believe it.  That is, the owners of the recordings DO NOT want people to have access to the full quality family jewels, so the FA process is a convenient way to 'encrypt' or 'distort' the signal such that the original quality is not available.  But, this distorted signal is still plausible to listen to.

DA sucks as a GP compressor, DA doesn't significantly increase average modulation, DA doesn't make the signal much louder (except in limited cases) -- mostly DA is intended to be reversable, and was originally designed for that reason.  Broadcast processors need not be reversable.  A reversable compressor has a lot more limitations than a non-reversable WRT loudness.

 

Read this, and read it again:
The only SANE reason for the form of FA that I have seen -- that is to keep the ultimate quality away from the unwashed masses, so that the owners can maintain original versions in digital form as proprietary.  The way that FA is encoded DOES NOT come directly from normal DolbyA usage -- FA is in M+S form, and DA is generally used in L+R form.

Keeping full quality away from the consumer is probably the reason why I started getting political pushback, that is, when the FA decoder started working SUPER WELL.

 

Isn't it interesting that 'full quality' is  easy to come up with, why hasn't it?

 

John

 

[John Dyson]

5 hours ago, Audiophile Neuroscience said:

Frank, why not give it a break on promoting the "method". It must be exhausting for you to keep on quarreling with one then the next and then next person.

 

It is a futile exercise for you as all evidence is that nobody hears it like you do (to the extent you do)  in this forum.

 

So, why not just enjoy the music and be satisfied in your beliefs without needing to convince others on a daily basis, which by now must be spanning years.You twist just about every thread into "your topic" and yet you have obviously so much more to offer.

 

Just a thought

I pretty much agree...

 

My own belief is this:  there are at least two kinds of audiophile people:  1) those who should simply try to enjoy for themselves, or 2) those who try to improve for others , while enjoying the experience (whether commercial or not.)   Type#2, of which I am a  non-commercial example is either technically much more difficult than one might initially guess, or financial success happens because the results are being over-sold.   Most people, including my early years were in type #1 category, just try to enjoy the experience.   When the experience was no longer fun in about 1990, (bad sound from CDs), then I left the hobby and went on to writing OS code.

 

Type#2 is hard as hell, and I don't suggest that people embark on it, unless they have something very special in mind...  Even then, don't waste your time unless there is a lot of time to waste!!!!   It is much better to be 'Type#1'.  :-).

 

Just tweaking don't cut it in the type#2 category...  Lots of unfocused tweaks are marginally useful when trying to help others. Gotta either have a super-major improvement, or a really good, perhaps at least mildly dishonest marketing person over-selling an improved product.  Nowadays, REAL improvements don't come easy, as there have been 25+yrs of current DSP techniques, 50+yrs of brialliant people with almost current analog technology, and 70+yrs of modern audio techniques in general, there are many thousands of brilliant well educated engineering people to compete with.  Actual innovation and improvement, after these at least 70yrs of development in current audio techniques, do NOT come easy.

 

'sounds better' to a single person is very easy, as almost every Type#1 audiophile does it.   A few nice specific techniques might help others, but 'sounds better' with a bunch of random suggestions doesn't materially help anyone else.  It is always good to listen to other peoples hints though -- keep ones mind open.  Brilliant improvements are not likely to happen though.

 

John

 

[John Dyson]

About digital-circuitry sourced noise getting into an analog signal, and perhaps analog sourced noise leaking into 'analog' cicutiry past a digital intermediary, this happens often because of (in this order) 1) conducted noise (most often ground reference problems), 2) coupled noise, 3) radiated noise.   What is sometimes misunderstood as CD clock noise leaking into an analog output after a digital intermediary -- that almost always happens because of an inadequate ground reference or other conduction problem.  DIGITAL BITS DO NOT LEAK NOISE.  Digital bits can be in error or not, digital binary bits don't have an 'inbetween' or secret side-channel noise path.

 

Digital circuitry and  mixed digital/analog circuitry can easily leak the current surges from a digital transistion into the analog portion of the circuitry.   The path for noise leakage into analog IS analog, but can result from poor grounding associated with a digital/analog interface.

 

Most all engineers/pseudo engineers know about the voltage 'spike' being generated from a current change through an inductance (or piece of wire), along with another component of the voltage being an image of the current flowing through resistance.   These 'current surges' that invariably happen from digital circuitry transitions, then map directly to ground/power current surges, then into voltage surges in grounds/power, power supplies and perhaps even coupled into signal lines.   This 'mapping' of the current surges into noise can be blocked (well, magnificently mitigated)  by using engineering knowledge, know-how and design.

 

So, a lot of analog background noise can come from ground/power/maybe-coupled noise generated by nearby digital circuitry, but can most often be mitigated or remedied by proper design and LAYOUT techniques.  For the case of very high quality audio being mixed with digital circuitry, the needed layout discipline as at the level required by greater than 100MHz signals.  If proper discipline and actual layout DESIGN isn't done, then the result will certainly seem to be more of an 'art', which doesn't bode well for consistency in the apparent product quality..   When analog/digital design in the current technology starts being an 'ART', instead of 'ENGINEERING', then there is a competency gap somewhere.

 

Nowadays, a mixed analog/digital layout is tantamout to an RF layout, and needs to be taken seriously as a part of the base design.

 

John

[John Dyson]

3 hours ago, shtf said:

 

 

En contraire.  I engaged in a bit of shall we say "meaningful dialogue" with John Curl and Mark Levinson in another forum some years back.  Eventually, Curl admitted that all of his designs and all others' designs contained at least one serious unknown flaw for which even their professionally calibrated measuring instruments were of no value to them.  I informed Curl that his designs contained no such serious unknown flaws but rather his designs (and others') were simply incomplete.  Curl didn't take too kindly to that.  Not saying Curl specifically, but some people would rather remain ignernt rather than admit somebody else may know something they don't.  Especially about their own designs.

 

Bad recordings?  Do they even exist?

 

As usual, and I'm unsure how or why, but Frank seems right.  Again.

Is this an example of sounding 'good' or sounding 'bad'?   By the time that it goes through the mic setup, I cannot tell how good it is at all.  I listen for precision and errors in processing -- the ambient environment confuses (time wise) the signal.

 

This is definitely a matter of personal preference.

 

John

 

[John Dyson]

10 minutes ago, sandyk said:

 

I don't have the measurements, but John Dyson checked this recording to see if it had improperly decoded Dolby-A and made that remark.

 This double CD was originally sent to me by a Music Teacher in the U.K. who has heard the Master tapes and knows the hall it was recorded in. Ian originally sent me the CDs as a gift as he wasn't happy with his own rip of it.

Yea - I have very seldom/never seen/heard a CD with this dynamic range.  I forget exactly, but it has wider dynamic range than what is comfortable for me to listen to in my quiet bedroom.  (Of course, I am tinnitus limited.)

 

I mistakenly convinced myself that it was FA - I guess dynamic range greed kicked in, but realize that I wasted my time and others because it is NOT FA.   Another London Philharmonic CD I have IS FA, and silly me, I went into automatic decode mode.  Of course, with so much of the disk at such a low level, there was a pronounced increase in dynamics when doing an errant decode -- and, just could never make it sound right after decoding :-).

 

John

 

[John Dyson]

1 minute ago, Kal Rubinson said:

It measures as 19 with JRMC and, at that, ties for the highest among the many recordings I have.  There are a few multichannel ones that measure 18.  I will get a chance to listen to it in the morning.

What it FA?

FA is my name for the ubiquitous DRC scheme used on most POP CDs, many jazz and at least some classical CDs.   It is probably the major cause for the 'digital sound' complaints that happened when CDs came out, but for some reason industry persists in doing it.  This is a primary reason why old vinyl doesn't sound like the CDs, or kind of, almost sound similar -- but not quite.  About the only place that doesn't persist in selling POP CDs with FA is MFSL, where they do sell material that has been closer-to-correctly mastered.

 

John

 

[John Dyson]

1 minute ago, The Computer Audiophile said:

I have some albums from Reference Recordings with a DR score in the twenties. They can be hard to listen to. 

With this kind of material, with so much dynamic range, it can be beneficial for optional versions available  with a little bit of carefully applied compression, or perhaps the user having access to a nicely designed compressor available in their equipment.   It isn't too hard to design a good, quiet compressor with limited effect...  Of course, how many recordings really do have such extreme dynamic range?  Not all that many.  I doubt that the number of uncompressed recordings and the number of listeners would make a market for such a compressor.

 

I really don't generally like compression -- but compressors are a tool, and sometimes might be useful for one reason or other.

 

John

 

[John Dyson]

4 hours ago, Confused said:

I actually mentioned in an earlier post that I have some recordings that actually sound better in the car.  (and many that do not, of course)  I quite like the sound I get in my car, although my car system does lack headroom, so very dynamic material can distort on louder sections.  99% of the time this is not an issue, because what I tend to listen to in the car does not have a high DR.  I guess at a audiophile level, we complain that music is compressed to suit radio / earbuds / and in car listening, so should it be a surprise that listening to a lot of modern material in the car is satisfactory?  

There should be a knob, with a well designed compressor on ALL appropriate equipment, normally disabled.  This will allow listening to music when the kids are playing, driving down the road in the car, or optimally in a quiet listening room or good headphones.  It is critical that the compressor NOT be in the signal path when disabled, and it MUST be of a kind of technology that mitigates distortion. (More sophisticated than a FET, THATcorp or even LDR type compressor -- LDR comes closest to mitigating distortion though -- just less stable than a solid design.)   Doing a really good compressor should use AT LEAST the very subtle techniques in a DolbyA, and that isn't even good enough -- because it doesn't mitigate the modulation distortion adequately.

 

John

 

[John Dyson]

13 minutes ago, gmgraves said:

I’ve often thought that car audio systems need to employ a variable compressor feature. That way, low level detail doesn’t get lost in the road/engine noise that is inherent in all* automobiles. 
 

* An uncle of mine had a Bentley when I was a teen. He used to complain that as quiet as the Bentley was supposed to be, he still couldn’t hear the soft passages in the classical music he perennially listened to on his Becker Mexico FM radio!

When I worked at Delphi, I offered them some technology, but they didn't seem to be interested in anything beyond their own  (I was working at their research center.)   I think that some vehicles have variable levels based on speed or estimated noise levels, but *properly designed* compression along with some level contorl would be better.   Increasing loudness only stresses the signal handling while marginally helps the lower signal levels.  Nothing outlandish would need to be done -- but only some increase in loudness, with compression when the signal is below -10dB -- something like that, but not much more.  Perhaps do something preferentially in certain frequency ranges like DA does (certainly, certainly not DA though -- just an example.)

 

Frankly, going down the road in a typical $25k vehicle, a small amount of HF compression with some slopping up the higs on pop or cymbals isn't as bad as not hearing the material.  I used to drive A LOT (>>3k miles per month) when I commuted between Indy and Kokomo or Indy and Detroit, and there were many times that I couldn't reliably hear music cleanly.  (I won't drive a car that I own l that much, so I rent vehicles when I have to commute or go out of town.  I used to get fairly good deals on a weekly/monthly basis, like $120-$160/wk, better than a torn up personal car in a year or two.)  So, I had little control over the noise isolation of the cars that I drove on the road.   ( had a short life in the automobile industry once Bell Labs left Indy, and I quit commuting between Indy and California.  Then, I quit workign because of medical reasons.)  Had LOTS of experience with variable (sometimes okay) road noise issues.

 

[John Dyson]

28 minutes ago, opus101 said:

 

I'm out - Frank didn't answer my question. Now @fas42 if you don't know the answer to a question just say so. That way a conversation is kept going. Ignoring a question makes it boring for the asker.

 

As to whether a masters degree in EE has anything to do with optimizing power supplies I'd be fairly sure the answer is 'no'. I am also a degreed EE and we learned nothing about power supplies that I can recall. So you gotta dive in and play with them (and sim them) if you want to learn. Several years ago I improved a Xindak integrated amp a large amount by lots and lots of extra caps and inductors between the rectifier and the amp circuitry. I still have that amp but its bettered by an optimized single-chip classD which goes for $3 nowadays so it stays in the cupboard.

 

By the way George I didn't take any of your words as 'harmful' so didn't feel at all hurt. Apology accepted nevertheless.

About being an 'EE' -- it is such a wide field as to be not-descriptive of expertise.  I am pretty sure that there are EEs who don't know how to design an FIR filter without several days of research, just like many EEs wouldn't know immediately how to design an RF amplifier or mixer.  All an EE degree does is to help start the person with a technical background.   Having an EE degree doesn't even guarantee an engineering work discipline of any kind...

For young engineers, an EE degree is a foot in the doo for employmentr.  An MS or PhD simply means that the person has demonstrated the ability to learn or research & document research to some extent.

 

For people like me -- my degree is so ancient as to be meaningless wrt knowledge.   I would guess that 5% of my electronics background and knowledge is represented by my degree.  (And my degree covers very few CompSci classes also.)

 

From a 'technologists' standpoint, an EE degree is a toolkit.  I don't have a PhD for example, but perhaps 10-20% of my co-workers were PhDs, and I was helping them get started with design and programming much more than I ever got help from them.  However, the general PhD is a proof of a learning ability and the ability to 'follow the rules' and work within a system, and that can be VERY important.

In some cases, a PhD's research will be in the exact field where the engineer/scientist works, and that is really a nice thing, but much of the time, a PhD will do the same thing as we lowly non-PhDs do.  (Only being slightly more upwardly mobile, and able to leverage credentials from time to time.)

 

So, a degree is a statement of learning discipline -- but isn't everything that it takes to be a good engineer (or designer.)

One thing for sure though -- I am glad that I have a degree and did what was necessary to get the career credential.

 

John

 

[John Dyson]

22 hours ago, Confused said:

I have often wondered the same thing.  I have a suspicion that the techniques used for mastering were "fine tuned" over many years for release on vinyl, and it took a while for the industry to get things right for CD.  This does seem a bit tenuous though, we may never know.

Much material is FeralA -- I just cleaned up some Classical stuff that was double encoded.  Remember that when CDs came out, the latent DolbyA compression was added.  One of the side-defects of the compression is a dynamic boost of the highs.

My previous decoding examples often removed only ONE layer of sometimes two layers.  This problem was pernicious.

 

Very often, after decoding, there is little need for substantive EQ (maybe a slight rolloff on highs.)  The HF compression being totally gone.

Here, I have attached a 'harsh/over-enhanced' recording as normally sold on CD (and even downloads), and the corrected version.  This required NO EQ, and normal recordings only require the correct choice of three modes.  (A kind of mode1, mode2 or mode3.)  There are outliers also, but those are not usually the more large scale pop groups...

 

 

 

John

 

02. Nobody Does It Better-ORIGCD.mp3 02. Nobody Does It Better-fulldecode-undistorted.mp3

[John Dyson]

5 hours ago, Audiophile Neuroscience said:

I think the real rot set in with compression and I think the use of compression was in a way previewed by the use of peak limiting. Barry Diament has commented that peak limiting started "especially during the cutting of lacquer disks for vinyl production.  The purpose was to avoid an "overcut" where a peak too large would cause the groove to overrun a previously cut groove as the spiral is carved into the master disk - which would make the disk unplayable as the stylus would then skip out of the groove when it reached this point."

Peak limiting is not an evil unless misused...   I have looked at many pictures of music waveforms, and so sometimes there are just a few errant peaks that might extend a dB or so above all others.   Even clipping those peaks might not be very audible, but peak limiting certainly appears to be reasonable for the case where tradeoffs are allowed.  That is pretty much the same as the peak limiting for cutting vinyl, where there are different characteristics for each frequency range.   There can be unexpected peaks in the needed power or needle displacement that a juducious peak limiting isn't all that evil.  (Again, when the tradeoff is reasonable.)

Of the ways available to improve average loudness, the modification of errant peaks seems to be the kindest way to do it.  There ARE ways of mitigating the peakiness in certain parts of the spectrum without actually limiting peaks.  The phase can be spread, keeping the peaks from being as coherent, yet maintaining the same energy level.   This wouldn't work below 500Hz, but above 2kHz or so, that technique might work.  It *might* be better than limiting in some cases.

 

John

 

[John Dyson]

3 hours ago, Confused said:

 

Is there any way of telling if a recording has FeralA other than listening?  In other words, how could you tell if a recording simply has some inherent distortion picked up during the recording and / or some strange EQ during mastering, over something that might be definitively lacking Dolby A encoding?

1) Unreal stereo image

2) Squishy cymbals/high hats

3) Artificial high end (too electronic sounding.)

4) Fumbled background -- vocal chorus are confused, cannot hear much ambiance

5) On older material, using traditional tape technology without DolbyA, you'll often hear WAY too much hiss (e.g. Nat King Cole Story.)

6) Look on the spectogram for generally rising (slightly) hiss above about 12kHz.  It isnt' a huge amount, but is one of the 'tells' contributing to the decision.

7) For not loudness wars:  excessive/meaningless compression.

(Note that I didn't say EXCESSIVE high end, because that depends on the FA variant -- for single pass there are two or three, for two passes there are about six.)

 

Most of the time, I have found that POP material is FeralA or some variant.  In fact, I have ZERO pop, except for MFSL that isn't at least one pass of DolbyA compression (FA.)   Any confusion in the past was because my code isnt dead-on accurate until recently..  I have even some historical pop that is also definitely FeralA (Petula Clark 4 disk set recording IS two passes.)  Nat King Cole Story might be one or two passes.   Al Stewart is two passes.  Almost all of ABBA's recordings are two passes.  The old Carpenters albums might be one or two passes (depending on album).   The Beatles 2009 remasters are one or two passes (I believe two -- but have to double check.)The London Philharmonic '50 greatest hits' or whatever, were two passes.   I HAVE BEEN HUNTING for non FA -- some MFSL recordings are the only ones so far.

 

About the thing that FeralA does is to make an unreal stereo image -- we (including me) have all forgotten what stereo sounds like, considering the very high probability of recordings being FeralA.   I just found about 1wk ago that a lot of material was run through two passes -- I wasn't getting a proper stereo image, and also the decoder wasn't accurate enough to withstand two passes (I mean, must be super accurate.)     Yep -- those ba*t*rds have been running their music through in two passes.      I am still not sure about the really good Brothers In Arms, so ran it through only one pass.

 

I have found one recording that is DEFINITELY NOT FA, Alex showed me a 'perfect' recording, that I tried to decode, and just ended up with cr*p, but it was NOT FeralA.  I tried hard, because everything that I find IS...    Even a Sheffield Labs CD was an especially nasty variant of FA.

I would LOVE to find some pop that isn't FA -- So, I challenge ANYONE reading this -- pass me a 50 second snippet (yes, I know its long) with substantive material, making sure that NO normalization was done on the recording -- normalization makes my decoding effort much more tricky...   Most likely I can cleanly decode it after three tries for one pass decoding (which is often not enough) and approx 6 tries for two pass decoding.

 

This is NOT the result of a delusion, even though it would be rational to be skeptical.

To me, this FA travesty is almost unbelievable, but keeps getting proven to be true.

 

John

[John Dyson]

Here is a very short and non-extreme example of the typical compressed CD sound (evilcd.flac) and the more correct sound with bette/stabler stereo image, smoother vocal, but less bright (clean.flac.)   I am not claiming that the more natural sound soundsbetter for all people, but the 'clean.flac' is a more gentle sound, with less dynamic/compressed boost of the highs.

 

Also, if you listen carefully -- you'll notice a better stereo image on 'clean.flac' -- I notice that the left hand side has a more natural sound to it.  There significant level/spatial 'wandering' on the piano on the far left on the 'evilcd.flac' version.

 

I do promise that the raw cd (evilcd.flac) version IS brighter, and it has ALL of the information that is in the clean version, ihe 'clean' version is algorithmically (not tweaked with EQ) from the 'evilcd' version.   The 'evilcd' sound is on almost all pop materials.  I used a standard method to obtain 'clean.flac', and it works directly on perhaps 2/3 of my pop CDs, and a variant works on by far most of the rest.

(the difference in data formats comes from the decoder expanding CDs to 88.2k/floating point -- doing the FA decode probably adds 2-3 bits to the 16bits of actual CD quality.)   There is NO fakery in the 'clean.flac' having fewer bits...

 

*BTW, this is a 2 pass decode.

 

John

 

evilcd.flac clean.flac

[John Dyson]

1 hour ago, Confused said:

I am genuinely fascinated by the potential of this John, and I do regret that currently I do not have enough free time to try it for myself.

 

So, I do not want this question to be perceived as someone just posting to be negative, I have a genuine positive interest, and have been impressed by some of your "snippets".

 

Looking at your points 1 to 7 above, it seams to me that points 1, 2, 3, 4, and 7 are fairly subjective, and could be the result of poor recording or clumsy mastering.  For this reason, I struggle to see how these points confirm the existence of un-decoded dolby A.  They may very well be indicators of the possibly of un-decoded dolby A, but nothing definitive.  Is this correct, or are their "tells" within the sound that could differentiate between say poor recording / mastering and genuine dolby A?

 

Point 5 - I think this could simply indicate that dolby was not used, tape does produce hiss if no noise reduction system is used.  Of course you mention WAY too much hiss, this I understand.  So, trying to be objective, could this be measured? That is up to a certain level of hiss, it might just tape hiss with no noise reduction, above a certain level it is un-decoded dolby?   For this point, I suspect not, because the HF is boosted prior to recording with a dolby system, so the inherent tape hiss is presumably unchanged?  So the treble in the recording would be boosted, but tape hiss itself would be at normal levels for whatever tape machine was used.

 

Point 6 - this does look like it could be objectively verified.  Out of interest, what sort of rise is typical in dB terms?

 

Regarding the hiss issue...  I guess that might be easily misunderstood.   When DolbyA is ENCODING, it COMPRESSES.  Esp because it is a fast compressor, and compresses about 10dB in the 3k-9kHz range and about 15dB in the 9kHz-20+kHz range, it basically BOOSTS the hiss on already hissy recordings.  Since a lot of the FA encoding is done in TWO passes, the amount of hiss increase could be as much as 30dB!!!

 

* So, using DolbyA in the FA configuration actually increases the hiss in the more audible range by up to 20dB, and 30dB on the very high frequencies.  (There is filtering that doesn't make it actually 20dB bad, but it is a VERY GOOD 10dB more hiss, or worse.)

 

YOU DO NOT GET NR with DolbyA in the FA config!!!!

DolbyA gives NR upon DECODING, which is what the DHNRDS FA mode actually does -- otherwise, without using the DHNRDS FA, you get that old tape hiss from the three track Nat King Cole recordings, boosted by at least 10-15dB or more!!!

 

A lot of older recordings DID NOT use DolbyA for tape noise reduction, but that is NOT what FA is.  FA is the *mis* use of DolbyA HW boxes for compressing consumer materials.  When you take a DolbyA HW box, then add some EQ on the output, then you end up with what is effectively a very fast multi-band compressor.   Now, FA is most often TWO applications of the compression at HF, and volia, YOU GET MORE HISS!!!

 

Regarding measurements -- with 20dB to 30dB of super fast compression (not just 1sec release times, but more like 40msec!!!), it IS audible, but does require some learning to hear it.

 

Also, I think that I found a metric that I can instrument the decoder (simply because it already has all of the file I/O and infrastructure built in) to measure the narrow dynamic range of FA materials.  The kind of dynamic range measurement is not over the very long term, but looking at the short-term very severe compressio.

 

ADD-ON:  wrt objective measure -- that would be nice, but there is no specification to base an measurement, but the latest versions of the decoder produce PROFOUNDLY improved results.  Maybe the best way (for now) to detect FA is to prospectively decode the material.  (There is NO longer a multitude of settings needed -- it is EASY to use, if used to command line programs.)

 

John