Jump to content
IGNORED

Lies about vinyl vs digital


Recommended Posts

I couldn't read the entire subject -- but there is one area that I know about the odd history of the quality of CD sound (and sometimes compared with vinyl.)  I remember early in the days -- and with evidence just gathered in the last few years -- that CD's sometimes sounded 'wrong', sometimes 'harsh'.  That led to an industry associated with 'keeping pure analog/vinyl sound".   There has been a problem with digital mastering, with material even being sold today with the flaw (material mastered/created before the 1990s'.)   This is a bit part of the solution about the 'harshness', and I do have existence proof if anyone is interested anymore.

 

This 'solution' to recovering the sound from early digital masters is NOT de-emphasis, even though that IS a partial/incomplete solution.  Very often, the problem has been left-over/undecoded DolbyA encoding.  Contrary to many popular rumors -- much undecoded (DolbyA) material doesn't sound crazy bad, but just a little bad.  The ears can be made to tolerate the material a little better once a bit of de-emphasis is applied (maybe minus 3-6dB at 3k-9k, depending on material.)   The real solution is doing the proper DolbyA decoding that the music distributor didn't do.

I didn't plan a big set of demos, but per a discussion elsewhere, I have some short snippets of Howard Jones which came to me DolbyA encoded (alas -- only mp3, but still is tolerable), and decoded versions.  The encoded versions have 'orig' in the filenames, while the decoded versions have 'DHDA' in the filenames.

In a little while, I'll upload a few more examples that might be more insightful.  However, the Howard Jones examples (and you might have to do a quick A/B on the corresponding 45second segments -- I provided 3 for two songs.)  The 'orig' versions might sound brighter, but they are fairly aggressively compressed at HF, which is the main activity of DolbyA at normal signal levels.  The 'DHDA' versions are decoded, and actually sound more 'natural'.

I do have a massive amount of DolbyA encoded ABBA (and some others) material as ripped from numerous CDS (In fact, about 1/2 of my CDs have latent DolbyA encoding on them.)  I'll upload a few more examples with the clear 'ORIG' and 'DHDA' designators in the filenames for your own comparison.

Right now (also) I have some snippets of a vinyl copy of an ABBA song 'Me And I' exactly as digitally distributed, a vinyl copy, and some DHDA decoded versions (carefully done.)

 

Again -- a big part of the digital 'sound' issue had been caused by improper EQ and ALSO improper handling of the masters by skipping the DolbyA decoding step.  The only reasons why the decoding didn't happen is that the process is cumbersome, and the DISTRIBUTORS COULD GET AWAY WITH IT!!!   In the end, the bean counters win.

 

The WWW site with short examples is: https://spaces.hightail.com/space/yDG3L339Rn

 

(Again, I'll add some more substantial examples later on today.)

John

 

Link to comment

They did not come from the video.  The clarity comes from using DolbyA decoding on good quality and appropriate commercially distributed digital and CD material..

I have several digital recording sources for ABBA recordings (all available commercially, none proprietary.)  My ABBA sources include the first Japanese CD releases (which are leaked DolbyA encoded.)  Also I have some ABBA Gold and more ABBA Gold (Japanese again), which are DolbyA encoded.  I alos have lots of other material like Eric Clapton, Anne Murray, Petula Clark, Carpenters, Simon & Garfunkel, etc.  All DolbyA encoded.  One item that can still easily be purchased (with a high probability of it still being DolbyA encoded) is available from HDtracks, the Carpenters album.   (Cannot tell before purchase if material is DolbyA encoded or not.)   Some material is 192k/96k at 24bits, but most is typical CD 44.1k/16bits.

When DolbyA decoding, the SNR is improved, and so there are more bits of music that can be extracted.  So, there IS a benefit when starting with 16bits to go ahead and expand the result into floating point or 24bits.  (My DolbyA decoder does the math correctly so the noise reduction is very effective.)

The reason why the decoder output is 48k, 96k or even 192k is that 1) the decoder works much better at 48k or even 96k especially because the distortion sidebands are easier to deal with.  2) At 44.1k, the distortion sidebands come too close to the audio range, so the results become a little metallic.  Since also at 44.1k there is no frequency detail above 22k, there is little sense in producing results above 48k sample rate.  However, if I plan to use the results for subsequent processing, then I leave the material at 96k.

The DHDA (or more formally, DHNRDS -- I am working with a moderately well known audio engineer) is a gain control device, and as such, produces distortion sidebands (many fewer than a real DolbyA by virtue of the wonders of DSP).  These distortion sidebands are a simple mathematical necessity (almost), and it is important that those sidebands never encroach into the audio band, and that is why I almost never use the decoder at 44.1k.  There are distortion sidebands that remain in the audio range, and those are handled by two seperate proprietary techniques.  (There ARE specific distortion cancelling mechanisms  inside the DHDA -- very very advanced algorithms -- there are patents on similar techniques -- but at 44.1k, the damage is done too severely to recover...  Still sounds better than a cat22/361 type DolbyA though.)

This is one of the few times that it IS mathematically beneficial to use sample rates well above 44.1k in order to reproduce audio up to 20kHz.  The DHNRDS actually processes audio up to about 40kHz if being used at higher sample rates.  At lower sample rates, it is limited to about 21.5kHz.  (I implemented decoding up to 40kHz so that HD audio distributors would be happy.)

 

John

 

Link to comment
2 hours ago, fas42 said:

Such encoding may be on the CD, but a sorted out rig won't present problems in the presentation of such. I listened regularly to a Howard Jones CD back in the 80's, and it came across very well. Also have an original release CD of the first ABBA album, and this is superb to listen to - demonstration quality in terms of the sound scape thrown up.

Some of the original release CD's of ABBA (and other material) are DolbyA encoded.  If THAT sounds good to anyone, then I'd worry.   I simply find vinyl to be distant and muddy sounding -- not enough transient capability to provide good time relationships.  Digital (with the use of linear phase filters -- unless necessary to meet a spec) maintains precise timing and structure of the signal.   (Simple analog filter equivalents (IIR) need not apply unless they are used for meeting specifications, sound effects or 'bass boost/cut' or 'treble boost/cut.')  'Normal' filters play with phase too much to maintain any kind of accuracy.  Nowadays, there is no excuse for using processing that modifies phase/timing relationships unless being used for artistic, tweaking, or simply for sounding the way that one wants.  Users need to know that when a signal is modified -- it just isn't an accurate representation anymore -- BUT THAT IS OKAY.

I don't argue against tweaking a signal in a certain way because it sounds nice to someone.  It just isn't very accurate to go around tweaking a signal unless you are doing it for a very precise technical reason.  Nothing wrong with being inaccurate.

Link to comment

Many  CDs of older material  (including US versions) are either hyper compressed or in DolbyA form.  I used the Japanese versions because I had gone through enough verification to know that they were DolbyA.   The recent CD/digital 'defects' are not just DolbyA encoding, but also sometimes incredibly aggressive compression.   (I was going to write a long story as to how I ended up with all of the various CDs, but suffice to say -- it was a long and boring story about the destruction of my interest in listening to music on expensive audio systems back in the early 1990s.)

I have a big collection of CDs (fancy, imported ones)  in storage, and recently found  some nice Japanese CDs -- noticing that they sounded better than the 'new' one, much less compressed but too 'tinny' sounding.

Making a very long story short -- I figured out that DolbyA encoding was part of the problem with the 'tinny' sounding CDs (part of the reason for my loss of the listening hobby.)   Lots of CDs seemed to have a similar defect -- DolbyA encoding.

 

So -- the defects that you might hear are 1) DolbyA encoding, 2) hyper compression, 3) just poor mastering.

Here is an example of hyper compression:  On the Complete Studio Recordings, Waterloo/'Sitting In The Palmtree' has a peak-RMS ratio of approx 15dB and a crest factor of about 6.  That indicates very, very processed/compressed/limited.  My cleanest copy of 'Sitting In The Palmtree' still DolbyA encoded has peak-RMS of about 22dB and crest factor of about 11.  When decoded (where DolbyA doesn't really affect the indicators of waveshape that much): peak-RMS of 21dB and crest factorcloser to 11.5.  A vinyl version of 'Palmtree' has a peak-RMS of 20.6dB and crest factor of about 11.5.

 

The numbers above show that the 'vinyl' version was likely processed a little bit.  Simple compression doesn't affect the peak-RMS and crest factors all that much, so even a small difference in those numbers does show some kind of processing or EQ.   I suspect that my best copy of 'Waterloo' (where Palmtree resides) is probably close to a master tape in quality, but is also DolbyA encoded as a bonus (for me), but a frustrating matter for most other people who like ABBA and want to hear the best.  (By the time that you read this -- I'll put 30seconds of the beginning of 'My Mama Said' from the compressed "Complete Studio' and a part of the very raw version with little compression. 

for 'My Mama Said' on TCSR:  Peak-RMS=13dB, Crest=4.5, the 'good' DHNRDS  (DolbyA compatible) copy: Peak-RMS=21dB, Crest=10.   I also added a more 'responsibly' compressed version, using the DHNRDS version as a start and my 3 band compressor (source/binary on the same site) to tighten up the sound a little bit.

 

Filenames: ABBA-MyMamaSaid-TCSR.flac, ABBA-MyMamaSaid-DHDA.flac, ABBA-MyMamaSaid-DHDAcompressed.flac

 

(The raw DHDA/DHNRDS version is NOT 'sweetened', but is a raw decode meant to represent original sound character.)

IN IS AMAZING THAT BOTH VERSIONS HAVE THE SAME PEAK LEVEL!!!   Compression can make material sound A LOT louder.

 

Repository: https://spaces.hightail.com/space/yDG3L339Rn

 

Frankly, the heavly compressed version might sound better to some people!!

 

 

Link to comment
36 minutes ago, The_K-Man said:

 

That seems to suggest it's of a remaster, not an early or initial CD release.

Oh yes -- I didn't mean to imply that it (TCSRs) wasnt a recent extreme example of compressing an original master.  It was done in the early 2000's timeframe.

Much of the early material (of many bands) has been released (leaked) as DolbyA.   I don't think that TCSR is DolbyA encoded, but a DolbyA decoder can process it because of the extreme compression.  (I mean, really extreme compression -- maybe by one of the processors used in radio stations, rather than an artful, simple compressor.)

 

I was simply giving it as an example/answer about how bad recent CDs can sound (or be over-processed.)  My comments about DolbyA encoding of earlier material (I know -- even recently purchased CDs -- I just got a Queen greatest hits a few years  ago, DolbyA encoded) do still stand :-).

 

John

Link to comment
33 minutes ago, DuckToller said:


John, I feel with you, having had similar experience after I changed into CD's (around the millenium), even I am not into Abba and Howard Jones. Thank you for the detailed information on the subject!
Is it, in your opinion, safe to assume, that music from digital downloads, if processed correctly and with care from the best available master, may sound better than most of the versions available on CD?
And more generally, that we won't get any additional comfort from the majority of digitized music, apart of not moving the CD tray anymore (formerly: the stylus drop down ;-))?
As the major music industry may not intend to to review their production process of their back catalogs (apart from the infamous MQA machinery  ...), only some audiophile Labels like Analogue Productions/Acoustic Sound will remain as the Gallic villages (Asterix)!
I just looked for some records of my favourite artists issued in the period 2016-2018 with DR >12,  and found that most of them are just slightly better than Metallica (DR3) , having more than halved the DR of their output since the 90ties.
OTOH, just having an excellent DR does not make good and interesting music/recordings,  as well as an impressive live performances ...
Cheers, Tom

I do think that a good digital download or a mistakenly released DolbyA CD can often be processed into something superior to what is legally available  *anywhere* else.  A simple remaster can be almost the same as taking a DolbyA CD (properly dithered -- not screwed with, of course) and DolbyA decoding it, and doing some subsequent sweetening/EQ to make it more listenable.  A DolbyA master tape vs. a CD produced by playing the DolbyA master tape without manipulation will produce essentially the same results.  (The biggest problem with the CD is the 22kHz frequency cutoff.)   Not all masters are all that 'listenable' without further TLC.  I do have an idea that some of the ABBA stuff (I know you might not be a fan, but they are my example audio 'basket case' -- needs fixing :-)) isn't really meant to be relased with SIMPLE DolbyA decode, but a small amount of compression or EQ seems to be in order.  (I find thatraw DolbyA decoded  ABBA often sounds significantly better with a slight amount of multi-band compression -- they seemed to have planned for the raw, decoded material from master tape to be AGC/compressed.)

 

I'd suspect that more than a few NEW vinyl releases would have been mastered on digital platforms, and even the original tape might be digital (I mean at least 48k/24bit digital, not 44.1k/16bit.)  I doubt that the record companies are interested in doing a run to Iron Mountain (or wherever they keep the family jewels) just to do an audiophile release of something.  So, they likely grab the best (easily available) digital or analog master, do the appropriate TLC (if they are being honest) -- then produce the audiophile CD, or a bit more processing so that it will fit on vinyl.  (Appropriate manipulation to avoid mistracking, keeping grooves a reasonable size, etc.)  Of course, the necessary processing for digital includes an HF rolloff leading to a 22kHz brickwall for CD, or the appropriate processing for other sample rates.

 

I agree about the dynamic range not being the sole determining factor of 'quality'.  There are many aspects to recording quality -- even some subjective ones that are best not argued about -- that the experience actually becomes fragile if too much 'screwing around' is done to the recording.  When I am chatting with one of my RESPONSIBLE 'recording' friends, I find that they tend to be almost paranoid about 'following the rules' because of the troubles that 'too much misguided TLC' can cause.

 

My guess (somewhat educated from talking to people in the industry) is that the distributors make decisions based upon profits and cost (bean counting), and it costs money to take a digital tape which had been directly copied from a master tape and go through a time consuming process of a DolbyA decode.   it is very easy to take the master tape (digital or analog) and apply a broadcasting processor to the material -- easy to do, as the broadcasting processors can be pretty much automatic.  After such processing, the subjective value mostly decreases because of the extreme abuse.  (Again IMO -- not making ANY absolute claims.)  Some people DO like the sound of compression!!!

 

John

 

 

 

 

Link to comment
  • 2 weeks later...
On 1/7/2019 at 6:00 PM, fas42 said:

 

It's a shame, that you're depriving yourself of the chance of experiencing what CDs can deliver. "Litanies of high end components" mean nothing, unless the person who assembled the rig really understood what care and attention is needed to get Redbook sounding as it should.

 

Over 30 years ago I heard probably the best vinyl setup in Sydney - and his CD player. The LPs at their best hit a very good peak, but his CD replay fell well short - loss of low level detail was shocking. But I had not the slightest interest after that in pursing LP ... :).

 

The "limits" of RBCD you hear are the 'standard' distortion characteristics of the medium when playback is not well implemented - they are the "snap, crackle, pop" of that method of music storage - and are just as hard to completely eliminate, as is vinyl "noise".

 

I have DolbyA decoded material that will TOTALLY knock anyones socks off.  A lot of the problem is the inprecision of the DolbyA decodes (and other NR systems -- I'd suspect that SR is even more critical.)  I just made several major improvements just in the last few days -- and the decoder was already LIGHTYEARS beyond a real DolbyA to begin with. Gibbs has NOT been my friend (that was part of the problems that I had resolved.)

The problems that you are hearing are not generally CD problems or even 16bit digital problems (because the actual resolution is finer than vinyl because of the total lack of even stats on the vinyl noise.)  44.1k can be a bit of a problem due to Gibbs, but even then -- the amount of energy above 12-15k is pretty small, so the liklihood of significant Gibbs is small most of the time.

Material mastered for digital material is often severely mismastered, and every little flaw shows up.  Vinyl has a nice imprecision -- so that can be nice.

Link to comment
1 hour ago, mansr said:

That is, in fact, entirely wrong.

 

I don't know if I agree/disagree with the OP's post -- but I do have/know and trying to herein explain the facts about digital resolution:

 

It is truly amazing about how misguided the opinions are about the 'stairstepping' or similar issues in digital audio.  I sometimes try to give an analogy that might sometimes help insightful people, as below:...

You know that cell system uses digital receivers.  The needed dynamic range is tremendous, and because of sharing the front-ends, the usage of AGC is necessarily limited.  But, in order to receive the incredibly wide dynamic range, and be able to utilize signals deeply buried in the input to the A/D on the input of the receiver, the amount of resolution has to be MUCH GREATER than the implied resolution of the 12bit/14bit and starting-to-be 16bit A/Ds.

So, how do they do this magic trick?  The answer is effectively similar to 'dithering' as used in audio.  It is specifically a little different, but in the same way as your hearing being able to hear material at/below the implied minimal signal of '16bit stair stepping', the cell receiver can extract very small signal from a bunch of noise.  (Peoples hearing is not quite as aggressive as the very special purpose receivers -- but this makes the point):

Summary:  There is much more resolution available than the 16bit (or even 24bit, if dithered) audio.  (24bits might adequately be effectively 'dithered' because of system noise -- but it is probably best to dither anyway.)

John

Link to comment
18 minutes ago, hifitommy said:

it is amusing to me that the chorus here is all in lock step with the glorification of digits.  the gang is banded together to exclude any variation from the mantra.  regardless, analog sounds more natural than most digital .  

i have incredibly picky hears -- having to listen all of the time -- but one thing for sure, analog is everywhere.  If you are speaking of recorindgs -- most of your recent analog recordings are done digitally.

 

J

Link to comment
8 minutes ago, The_K-Man said:

 

More natural than a low-bitrate mp3, for sure!  

 

But Red Book, and higher bit depth and sampling rates will kick analog's a$$.

I cannot even consider using mp3 in my work.  There is a severe severe quantifiable problem with it -- when you have things like quickly repeated vocals, then mp3 can totally miss them.  The loss of detail is astounding -- even at 320k full out.  Opus is a little better - but detail  still a little bit fading in and out.  Since I am working on recording studio software, I have to catch/perceive every little detail.  However, since I have no hearing about 14k, I have been fooled from time to time.

Link to comment
31 minutes ago, The_K-Man said:

 

By "least mastering" I meant least abusive mastering - as represented by the example in my avatar.  If that means mastering as we knew it before the mid-'90s, then I concur.

 

As far as 'flat transfers' are concerned, I prefer those in a reissue context.  I'm generally not a fan of remastering, considering the evident abuse that were sold to the public as 'remastered'.

I often use ABBA for my test material, because the recordings are so very difficult for gain devices to process without splatting IMD all over the place.  That said,  the original ABBA recordings  on DolbyA masters have  a crest factor of between 7 and 13, and a peak to RMS ratio of approx 16-22dB.  This 'complete studio recordings'  nonsense or somesuch result in a crest factor of between 4.5 and 5.5 anda a peak to RMS of approx 13.5 to 14.5dB.  THAT is an example of abuse, only enabled with broadcasting type audio processors.  Superficially the processed audio sounds 'okay', but then when I turn my 'analysis hearing' on -- the distortion is horrendous.  Doing that much processing cannot happen without significant distortion.

Point being -- a lot of recent material is heinously processed, but such evil isn' t just recent -- because even back sometimes when Karen Carpenter was being recorded, they actually purposefully used a DolbyA to enhance her voice.  That is so very sad, and I hear the effects on my DolbyA copies of her music ALL OF THE TIME for my testing.  I have (DolbyA digital) copies of every one of their albums betwen 1969 and 1981, and only the 1970/1971 wasn't severely over enhanced.  The 1977/1981 albums were pretty severe, with 1981 being the worst. (didnt mention 1969, because it was very amateurish.)

Of course, the ABBA wall of sound is another story...

 

John

Link to comment
14 minutes ago, PeterSt said:

 

Hey John, I have been reading your posts with interest but actually can't place much of what you are saying, except that nothing in me tells me that you're making up any of it. But small problem: maybe I was that guy who is able to squeeze so much out of any 16/44.1 that it is very hard for me to imagine what could actually be so destructive that it would be close to hell itself. Understand ?

 

In-other-words ...

 

By now I am super eager to experience what you mean. Also, you talk about Dolby as the worst of all evils while you yourself have (made ??) the best of all best, and ...

what would I myself possess of that ? I select on "non-masters" and generally the highest DR -  maybe that is not the way to go. But then I never ever heard of "Dolby" in the realm you talk about it ever (of course I am blind and thick and such in your eyes, but I am serious anyway).

Many albums I have in several versions (54K or so in total), so what can I do to perceive what you are talking about ? or ... can I obtain a few examples from you ?

 

Really, day in day out I am improving on Redbook. This now seems a whole new opportunity ...

 

Thank you, John,

Peter

I am posting a quick response here right now. After I produce some examples from the most recent DolbyA compatible decoder, I'll post an URL with some undecoded and decoded examples.  They will likely be an original Carpenters snippet, ABBA snippet and a Louis Armstrong or maybe Simon & Garfunkel (not sure exactly yet.)

 

The reason for multiple examples is the style  (technical attributes and genre.

 

The slower response will be in about 1-2Hrs while I set it up (it isn't that much work, but just a matter of pulling it together.  The computer does all of the work.)

 

I am currently restricted from distributing a demo release of the decoder without permission from my team partner (he is the marketing/sales/recording expert, I am the software nerd and audio processing person.)  I do own the software, but must keep the project coherent because of the interface with the recording industry.  (I have been considering a limited release, but need to discuss with my partner.)

 

Expect a response again soon (very few hours, not days and certainly not minutes :-)).

 

John

Link to comment
3 hours ago, PeterSt said:

 

 

By now I am super eager to experience what you mean. Also, you talk about Dolby as the worst of all evils while you yourself have (made ??) the best of all best, and ...

what would I myself possess of that ? I select on "non-masters" and generally the highest DR -  maybe that is not the way to go. But then I never ever heard of "Dolby" in the realm you talk about it ever (of course I am blind and thick and such in your eyes, but I am serious anyway).

Many albums I have in several versions (54K or so in total), so what can I do to perceive what you are talking about ? or ... can I obtain a few examples from you ?

 

 

Right now, I am doing my best!!!  (There is info about the DEMO pieces below -- PLEASE read the section about what to expect)

 

The DHNRDS decoder (there is a www site by the same name, but everything is super preliminary) is still being fine tuned, but the vast majority of the development is complete.  Sometimes there are bugs, and sometimes we run into material that causes the decoder to do the wrong thing.  Most of the recent problems are associated with the filtering and attack/decay shaping (it is complex -- requiring an emulation of a rather tricky diode network -- not just a detector or something simple), but I just knocked off a few more bugs.  I'd suspect that the version that I have just finished & just used to do the example decodes will be going out as demo versions (1 month timeout) in a few days.  if you are REALLY interested, get in touch with the www site 'www.DHNRDS.com', and/or I'll see if I can strategically be allowed to distribute time-out versions of the decoder.  (I am the owner of the software, but not sole owner of the project, and the project&contributors are necessary for the success of the project.)  We don't expect to make lots of money on the decoder, but a real version will cost a little bit of money (IMO, CHEAP considering the complexity and unique technology.)  Please don't think that we are wanting to 'jab' anyone for money -- I have spent approx $200k/$300k of billable time on the project, of course not doing it for money, but for the TRUE interest in being able to help optimally recover a lot of old recordings.  BTW -- I don't mention the name of my project partner out of respect for him -- I tend to be a loose canon, and don't want to negatively affect his reputation -- however, I am REALLY trying to be a 'good boy.' :-). (AFAIR -- and my memory sucks -- the planned price is in the $350-$500 range more or less, but I am not a sales or marketing person, so I have no authority on the subject of price.)

 

The DEMO web site is different from the project web site.  You are allowed to download or listen to anything on the repository (including the nice, 3band RMS compressor software with source code.)  This repository is solely controlled by me, so I can upload demos or whatever.  (There might be a potential project of changing the compressor program into a plugin -- truly it is likely one of the better SIMPLE compressors available.  It is not a MAXIMIZER, and only meant to tighten up recordings a little bit -- it will self-limit on damage, so it is hard to make it totally destroy material.)

 

I did provide some  demos for decoding DolbyA material on the DEMO site, and LET ME EXPLAIN WHAT TO EXPECT -- or maybe your expectations JUST MIGHT NOT be met.

 

DolbyA encoded material is weird.  It can sometimes have harsh HF, it can sometimes be hissy, it can sometimes have very apparent HF compression, but almost always -- encoded material has a very, very thin perceptive spatial depth.  DolbyA compression is not like a normal compressor as the compression curve is mostly active below -20dB and squeezes the music in the range of approx -20 to -42dB, and changes that range into approx 10dB range (10dB compression for 0-74, 74-3k, 3k-9kHz, and 15dB for 9k-20+kHz.)  DolbyA encoded material that is already compressed by the studio REALLY doesn't appear to have a horribly large HF emphasis, but uncompressed material tends to manifest the entire 10-15dB of gain-up (esp at high frequencies.)  There IS LF distortion in non-decoded material because of the approx 60-180msec decay time constant for LF/MF.  DolbyA is weird also in that it doesn't really have a fixed attack/decay time -- which makes it *interesting* to emulate!!!

 

Bottom line DO NOT EXPECT THE DOLBYA ENCODED (nondecode) MATERIAL TO BE UNLISTENABLE, but like I wrote above, it tends to be very 'thin' sounding.  The encoded/decoded state is obvious in the filenames, and I did have to limit the examples to reasonable  length (I might have made them a bit longer than they should be, but they are certainly not complete recordings.)

 

One more thing -- the recordings are imperfect, and many (not all) of the imperfections come from the source.  Also one of the Carpenter's examples comes from a Carpenters album from HDTRACKS.  So, DolbyA encoded material is somewhat common, it is just that one needs to know how to recognize it. (Please dont' hurt my feelings if you don't like something, but CONSTRUCTIVE criticism is super welcome!!!)

 

Repo: https://spaces.hightail.com/space/yDG3L339Rn

 

Link to comment
1 hour ago, One and a half said:

I

 

Right, for ABBA, and quite a few CDs from that era, the bass was wrong, does it exist? There’s the debate on phase polarity being wrong, wrong RIAA curve applied. Now this  kind of audio engineering accounting driven mediocrity makes a lot of sense.

 

Very keen to listen to the samples with the Dolby A decoding at the end of this flight. Even the consumer Dolby C was too bright to listen to without flicking the switch on cassettes, Dolby A decoding should bring this back and jerk the accountants. 

 

 

 

Right now, I am decoding ALL of ABBA's mainstream albums, but only uploaded one song (which wasn't really an extreme example.).  In  a short while, I'll upload a few 'just decoded' examples, and maybe one or two more undecoded examples.

One really frustrating issue with ABBA is that I do have the DolbyAs, but the decode results down't always sound like the digital albums (and a few songs don't sound like even the vinyl, even though most of the time they do.)  SOMETIMES, the ABBA stuff seems to be sweetened -- even like the better stuff from Polar.

On the other hand, the personally decoded versions REALLY sound good, and are often very, very close to the original vinyl (but cleaner.)

It is 21:59EST/USA time, 20Jan, and it will probably take me about 1-2Hrs to uploade more than just 'Bobbys Brother.'  The "Brother' example DolbyA doesn't really sound bad, but I kind of like the song (and it iisn't so common.)

I'll upload 45-60 seconds of a few more very soon.  If someone proves to me that they have original copies, I might (if desired) decode a few songs completely and return them to them.  (The copyright rules are frustrating, and don't even know if that is legal.)

Link to comment
1 hour ago, PeterSt said:

 

Hi John,

 

At this moment it is unclear to me what it takes to provide this proof. Also, a layman regarding this like me, wouldn't even know what to look for regarding Dolby(A). Would it "definitely" be printed on the back cover somewhere, for example ? Would, for example, a remaster never comprise of Dolby compression ? (but it wouldn't be regarded an original to begin with)

 

I must learn a few things ... :) (I guess I am not the only one)

 

Would there be a way of you referring to a concrete example (artist doesn't matter) of two production versions, one properly encoded, the other not at all or badly or whatever it takes to audibly LEARN what this is all about ?

 

And now I am even browsing through an AES paper. You guys are serious ...

Peter

 

Tomorrow morning, I'll put up several copies of the same song, from the same master tape -- DolbyA encoded, Vinyl version, Normal digital version from consumer source, and my decoded version.  For funzies, you might want to compare.

DolbyA is NEVER written on the jacket of the consumer material because it is  'leak' caused by beancounters telling the mastering people that it is 'too expensive' to run the digital copy through a slow realtime process on inconvenient hardware.  (That anecdote has been told to me on the QT.)

As you can tell, DolbyA encoded material (not decoded) is good enough for the 'little people' like me and you, so we get the nonsense messed-up stuff.  Places like Mobile Fidelity will spend the time and money to get the old DolbyA unit out, and spend the extra $1000 that it takes in mastering engineer time to process the material.   Normal distributors will often just playout a master tape (perhpas with EQ) and call it 'mastered.'

A part of this DolbyA encoding problem is that there are ZERO normally available *SOFTWARE* DolbyA decoders available that sound properly like a real DolbyA (except this new decoder that I have been developing.)  Even Dolby hasn't bothered selling a software DolbyA decoder, even though they have a primitive one (I think) that is part of the cinema series of equipment (the quality requirements of cinema aren't quite as stringent as high quality audio -- e.g. 10-12kHz max.)   There is another DolbyA decoder being sold, but it truly doesn't sound like a DolbyA, but is PROBABLY a '100hours worth of work' approximation.

My guess as to why there is no other software DolbyA decoder that works close to a real DolbyA (and I'd suggest that my decoder has MUCH MUCH less distortion/etc) is that (for example) no-one will ever make the time investment back in profits. The development has taken (literally and quite honestly) about 1-1.5yrs of almost full time work by a 30yr software /Analog HW and somewhat DSP expert (if I might say so myself), and there is NO WAY a competent company will spend this kind of effort on a DolbyA project.  (Actually, I haven't been programming/doing Analog EE hardware for 30yrs, but actually 40yrs.)

Doing the decoder project SUCKS -- but once I started it, it is in my nature to finish it...   This is the singular most difficult project that I have ever done (and I have done some dueseys.)

I sure hope that people do have a chance to use it, and it WILL be available some how/some where soon!!!

 

John

 

Link to comment
8 hours ago, John Dyson said:

Tomorrow morning, I'll put up several copies of the same song, from the same master tape -- DolbyA encoded, Vinyl version, Normal digital version from consumer source, and my decoded version.  For funzies, you might want to compare.

 

 

I uploaded several versions of 'Take A Chance On Me'' from ABBA.  Enumerated here (also with my opinion/comments)

ABBA demo repository: https://spaces.hightail.com/space/zqrnA5nqnV

 

Take a chance on me short demos:

ORIGeq -- exactly as decoded

Probably the least interesting sound --  it is all there though.

EarlyVinyl -- playback of an early vinyl copy

Sounds like vinyl -- HF emphasis

EarlyVinylEQ -- locally decoded, EQ similar to the vinyl

Roughly equiv to vinyl emphasis -- note deeper dynamics

PolarDigital -- Polar music digital copy

Usual fairly clean sound.

PolarDigitalEQ -- locally decoded, EQ similar to Polar Digital version

Even though roughly same long term dynamics, has more distinct sound

(due partially to less IMD and also better controlled attack/decay dynamics)

undecoded -- DolbyA encoded original

Almost similar sound, but thin spatially. Weaker midrange/bass.

 

(The above are my opinons, and I do hear a flaw in my local decode -- I still have too much HF limitation in the attack/decay (gain control spectrum.  This causes an odd inconsistent gain-between-bands effect, and I am certainly going to fix that poste haste :-)).  The various choices of tuning in the decoder are very tricky to decide -- because if they are too aggressive, then we get artifacts, and if they aren't aggressive enough, then we get small increases in distortion.  Very, very tricky and time consuming to make it correct.

Link to comment
47 minutes ago, John Dyson said:

I uploaded several versions of 'Take A Chance On Me'' from ABBA.  Enumerated here (also with my opinion/comments)

ABBA demo repository: https://spaces.hightail.com/space/zqrnA5nqnV

.....

EarlyVinyl -- playback of an early vinyl copy

Sounds like vinyl -- HF emphasis

....

My note in the previous post was unclear.  I didn't mean to imply that all vinyl had HF emphasis, but rather it 'sounded like vinyl', and there was some HF emphasis.  More than likely the HF emphasis was meant to overcome typical vinyl playback, some characteristic of some equiopment elsewhere, or simply the attempt to be 'brighter' sounding.

I'd never claim that vinyl sounds 'more bright', unless it was created to be 'more bright'.

 

Sorry if I made an implication that might be very confusing...

John

Link to comment
9 minutes ago, Axial said:

 https://www.audioholics.com/audio-technologies/dynamic-comparison-of-lps-vs-cds-part-4

 

Make yourself a nice cappuccino, sit down comfortably in your favorite sofa with your laptop (tablet, iPad), light a big Cuban cigar (for non-smokers have a couple chocolate mints), relax in taking your time reading and assessing (assimilation and adaption; graphical and from mental memories) ... all that plus jazz ...

 

 

I took a look a the review -- interesting.  One little tidbit that might upset those who love a lot of content above 20kHz...  Do you know what happens when material limited to approx 20kHz is applied to a DolbyA and then played back (DolbyA being the ubiquitous NR system for before 1990?)   Magically, after the DolbyA processing (encode only or full encode/decode), lots of the beloved >20kHz 'music'.  Guess what that 'music' is -- distortion synthesized from the music.  Maybe we can call it a high frequency enhancer?  Kind of like the old DBX boom box, right?

Point being -- there is a lot of nonsense in the audio signal -- esp at/near 20kHz or above.  Interestingly (at least to me), there are sometimes spectral bands at 'interesting' frequencies in 'wideband' audio also...  Let see, sometimes around 20kHz (accurately 19.2kHz), sometimes around 30kHz (accurately, 28.8kHz..)  Hmmm...  I have even seen spectral bands at about 9.6kHz (not a good thing), or even 14.4kHz (also not a good thing.)   Those frequencies remind me of serial port bitrates (not always using the standard rates, but those ARE the more accurate numbers for the bands that I have seen.)   More often than not, the bands that I usually see are at 19.2k and 28.8k.

 

With the realization that some of the 'wonderful' HF material that extends from high frequency percussion isn't really from the instrument itself, but often comes from signal processing equipment (or some bands come from serial interfaces) -- why does that stuff need to be carried along with the hearable audio?

 

There are cases where carrying the super high frequencies are useful -- for example, for that really genius time base correction scheme (but those frequencies tend to be much higher), or for making sure that if there really is significant energy in the 20kHz range, that the transient behavior due to Mr Gibbs doesn't cause other problems in the circutry or clipping in poorly written software (or when using +-1 based encodings.)

 

Just commenting on the reality that I have seen.  As they say, your mileage may vary.

 

John

Link to comment
39 minutes ago, The_K-Man said:

I would rather have CDs of vinyl master transfers than over-processed, loudified CRAP,  to be honest.

 

As for HD tracks and High res, all one has to do is read this:

https://www.audiostream.com/content/high-resolution-downloads-nevermind

 

to understand why they will never occupy my digital collection!

I wanna re-iterate one of my comments made previously here (or maybe another group)...  Paraphrased:  Please don't trust the content about 20k as being content from the performance!!!   If you see spikes that transcend the 20k or 22.05k barrier, don't necessarily believe that content to be useful music information.  Noise Reduction (NR) devices can very happily synthesize those nice intense percussive spectral blasts (slaps, impacts, etc) from a nice true audible burst (remmebering that not many microphones do accurate reproduction much above 20k -- except maybe a few esoteric or maybe Sennheiser RF condenser mikes.)  So, in a 1990s or before recording, if that nice impulse has spectrum well into 30-40kHz, a big part of can besynthesized distortion from NR devices and below 20kHz energy.  (I have seen it by my own eyes.)  Informationally,  my DHNRDS doesn't do that evil blast thing, but almost all HW devices do it.

(I am not claming that all content above 20k comes from audio processing, but I do know that it happens.)

Link to comment
3 hours ago, PeterSt said:

Hi John @John Dyson, two quick questions:

 

1. Why is everything resampled to 16/48 ? (except for one of the two Carpenters for its decoded version)

 

2. How can it ever be that your original versions normalize to something like 16dB more dynamic (over the 16 bits range) than the normal Redbook versions ? Please notice: "as far as I can tell at a few quick glances" (but this is for the ABBA and the others won't be far off).

 

Ad 2.

These are denoted the "originals" but I suppose come down to me earlier (maybe implied) questions like: what is to be considered original ?

I mean, 16dB ... that can't be in normal circumstances. So or you already processed the originals (OK, you did because 16/48 now and not 16/44.1) or you possess sources "we" can't reach anyway. In that latter case: how to compare without comparing (for us) real life apples with desired oranges ?

 

Of course the general hint is: would this be useful to a few of us at judging, we should not compare with the mere master tapes than we could ever dreamed of. And this is what could be happening ? (this relates to your question of a few days back "if someone can prove he has an original" ... etc.).

 

No hurries with an answer. I just try to understand what could be new here for us.

Thanks,

Peter

 

Why resampled?  Space reasons.  Process works best at 96k, and 48k is a nice submultple, and there is no information above 20k anyway.

 

It is okay to create more dynamic range -- same as tape hiss.  The recordings are probably something liek a signal on -60dB of tape hiss. 

 

By doing the NR, that extends the dynamic range (effectively)  by up to 10dB (pushing much of the hiss, and some of the very low level mid freqs down.)  16bits is similar to -90dB tape hiss, so the predominant source of hiss is from the tape, and the addtional tape hiss (over dithering) just makes the dithering more effective.  The reason why I often use 96k/24bits is that I sometimes do more processing -- it is best not to willy-nilly up/down convert,and REALLY there can be problems with a 22.05k brickwall if there is any energy in that region.  Best just to keep full bandwidth after one bandwidth limit (best to do a soft limit like a Q=0.500 from 40k, for example.  Brickwall where the energy is softly diminished to a very low level.)

 

Originals are versions that are leaked DolbyA available commercially. I can tell accurately maybe 90% of the time if something is DolbyA encoded, and the big issue is undoing any damaging EQ on the signal.  Much of the time, there is surprisingly no HF EQ(that is good)  But  it hurts(or my decoder starts sounding more grainy) when HF is cut.   It seems like about 50% (more or less) of the pre-1990 material is available DolbyA in one form or another somewhere.  Of the examples, I didn't buy anything 'DolbyA', I just know what DolbyA sounds like, and have a decoder that does bad thins when given non DolbyA (actually, it sometimes isn't really bad, but things mess up.)

 

I do have  real originals with tones in some cases, but I cannot let even samples out of my grasp.  I have very used to full/complete albums on real digital master.  I run them for tests quite regularly.

 

You can make your own 'master tapes' (effectively) if you find the CD or download that is DolbyA encoded  (e.g. the Carpenters Album from HDtracks is one example -- Also, I think one of the Paul McCartney albums, the 'unlimited' one or somesuch.)  Master tapes are often full albums, but you know what I mean.

 

The result of decoding is often as good or better than what you can get elsewhere (competing with MFSL for quality.)

 

John

Link to comment
1 hour ago, Rexp said:

Can anyone explain why in this comparison of vinyl v cd (Anne Bisson, 3rd track demo) the vinyl is vastly superior to the cd? Even on youtube the difference is clear. 

 

Most of the time, when you think that vinyl is better than CD, it is either mastering is faulty on digital, or it is something that you happen to like (the way that the vinyl was mastered), you prefer the kind of distortion in the transducer (e.g. the cartridge), the kinds of distortion in the  mastering process, other things llike that.  Notice that I didn't talk about defects in the digital system -- I am not trying to sell more magazines, build up a following or preferentially sell one thing over another.   This is after 40+ yrs as an engineer (analog HW, digital signal process and OSes.)  So, if you happen to like vinyl over digital -- it is something about how you perceive (and many others will agree.)  But for accuracy in almost every way -- digital is the way to go.  CDS in general have vastly superior dynamic range, less distortion (much less), and generally easier/more efficient.   Some of those 'hi tech' turntales simply look sexy also -- that can be emotionally very convincing.  More money nowadays definitely doesn't mean better.  (BTW, it is possible under certain circumstances -- incompetent mastering -- that the 22.05kHz limit can be relatively defective for various technical reasons -- but it is really seldom a problem..)  The 16bit limit is meaningless for hearing (not for the mastering process) as cell systems with much greater & accurate dynamic range requirements often get by with 12/14bits converters and careful dithering, and they need about 1uv (plus or minus) through 100mv ability to get at the sngal (much bigger than 90dB, or even the 70-80dB required for anyone over 20-30yrs old.)

Link to comment
1 hour ago, The_K-Man said:

 

What "digital noise" are you talking about?  NONE of my CDs exhibits "snap, crackle, or pop" except the one with 'RADIOACTIVE' by Imagine Dragons on it - but that's how I.D. wanted that track to sound, and they got it.

I think that fas42 is joking. :-).  Look at the smilely faces with the tongue :-).

 

John

Link to comment
4 hours ago, fas42 said:

 

Older, 'straight' recordings suffer just as much as overly compressed stuff - pop recordings of the 70's and 80's are spectacularly good to listen to - but you wouldn't know this if you listen on a typical audiophile rig.

The older vinyl were good for their day. Often, the major problem with the older recordings on digital is the DolbyA encoding left on during the transfer process...  There IS a problem with 'depth' why playing something that is DolbyA encoded.  That is one reason, after 30yrs a reallly capable decoder of DolbyA material is being created.  It has been needed for 30yrs, getting worse and worse.  Since CPUS are powerful enough to do the job now (probably starting when the Pentium4 came out -- but no marketing interest), and a capable developer has finally become availabe to do the work essentially for free (2000+ person-hours has gone into because of the need for testing and lack of specification), then the decoder has finally come-to-pass.

However, to proclaim that digital itself is at fault is so wrong in so many ways.  Many of the recordings, especially from the era before 1990, have been left unpleasntly encoded whn distributed in digital form, it has caused alot of people to deny 'digital' itself its proper place -- which is the best possible way to transfer audio to the masses.  Rumble (from many sources), transducer distortion (from many sources), the need for special EQ (because of cutter/vinyl groove issues), or special dynamic limiting (because of tracking ability issues), do not make vinyl a very good way of transferring recordings nowadays.

The recording distributors have not help the matter, and the idea of a class action (by someone who is actually interested in such things -- I AM NOT) might be in the offing.   It is in the interest ot the industry to keep the secret about the lazy lack of decoding secret.

 

Link to comment
2 hours ago, kelvinwsy said:

I Contend that Digital Files with a superior Music Playing Software and with even a mildly improved Digital Playback system - Audiophile PC, Software/ USB streamer/ USB Decrapifier etc makes the Sound Quality of Playback far superior to the CD.

 

(

I dont' think that you can find a vinyl copy of this material that comes close to the clarity (stuff like HF cut/compression would be needed), and will never be this clear on this material. This is RULER FLAT from most recent version of my decoder, and probably the same DolbyA as produced the record.  No-way that vinyl can be as time coherent as the digital example that I am showing.   If you want album type HF response, do something like 'treble -6 17.5k 0.500q sinc -t 5k -a 100  -21.5k'. (and no, this is not a severe rolloff,  not really -6dB in the audible range, should probably be stronger for vinyl, with some additional  time distortion built-in, purposefully the freq/level is diminished before the brickwall.)  You probably never have heard the Carpenters this transparently clean.. (The 'Q' that I used is close to linear phase -- unlike 0.707, or worse 0.840 or 1.0, often used in audio processing -- esp power amplifiers like cutters.)

Example:  https://spaces.hightail.com/space/LIZxyP0oxf

Link to comment

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now



×
×
  • Create New...