DolbyA or Pre-emphasis, which is it or both?

Discussion in 'Music Corner' started by John Dyson, May 16, 2018.

  1. John Dyson

    John Dyson Well-Known Member Thread Starter

    Location:
    Fishers, Indiana
    One thing about 'pre-emphasis' -- on some CDs I have found that the HF boost is NOT just pre-empahsis but is left-over DolbyA encoding. De-emphasis will kind of help, but DolbyA encoding will give the typical hard edge to the highs and sometimes the 'compressed sound' on the highs. Not all such materials are DolbyA encoded -- I cannot guess what happened in the process, but I have several concrete examples with real CDs sitting in front of me rather than ripped copies on disk/memory stick media. For example, I have a Queen Greatest Hits, Hollywood Records that is DolbyA encoded. Yet I have a totally over compressed/messed-up copy of ABBA (GOLD, 40th, Polar) which is not DolbyA encoded. I just got a Carpenters from HDTracks -- DolbyA. It is sometimes difficult to tell, and de-empahsis can kind of help, but you won't fully get rid of the compression or get the spatial relationships unflattened without decoding. (Fast linear compressors tend to flatten worse than RMS or slower compressors, and DolbyA compression is about as fast as you can get without lots of troubles - in fact is faster than what you really want at low frequencies.)
     
    Mbe and c-eling like this.
  2. Y9771

    Y9771 time traveller

    Location:
    Lithuania
    Maybe CDs which I have mentioned previously that I think have PE but are not flagged are actually with Dolby A ? Titles in question would be a German Alsdorf pressing (Specialty SRC matrix) of ELP - Trilogy, Heart - S/T 1985 (standard mastering), REO Speedwagon - Wheels Are Turnin' (standard mastering).
     
    Steve Martin and c-eling like this.
  3. John Dyson

    John Dyson Well-Known Member Thread Starter

    Location:
    Fishers, Indiana
    Ill look into friends/associates to see if I can find a copy of at least a song or two from that CD. Decoding DolbyA is pretty easy -- there is a decoder on the repository that I have talked about over and over (all over the place), and it runs on recent Intel/Windows PCs (also have a linux version available.) Not terribly convenient for some users because I don't normally do GUI programming and the DSP code in the audio projects is already complex enough, so it is command line oriented. (No need to decode everytime that you listen -- decode once, and put on a memory stick.) Hopefully will be able to give you feedback soon.
     
    c-eling likes this.
  4. c-eling

    c-eling Forum Resident

    Thanks. Your posts get about as much attention as PE discs, it's like a forbidden topic or something :laugh:
     
  5. John Dyson

    John Dyson Well-Known Member Thread Starter

    Location:
    Fishers, Indiana
    Yea -- I know about the skepticism -- some minds are opening to the possibility though. One thing for sure, even though I know that some disks are REALLY DolbyA encoded, I can make mistakes in judgment from time to time also. So, when I make my proclamation about the copy of the Howard Jones "Dream Into Action" that my friend sent me, I would only give a 90% probability of being accurate. In some cases, I would stake my life on it, but in this case, I am pretty sure, but not 100% sure. One of the reasons is that i haven't exhaustively chased down the needed threshold level, but it appears that it needs to be a little on the high side (many albums need the slightly higher threshold.) I am only talking about +- 0.5dB. Most albums seem to need between -15.75 to -15.25, while some need as low as -16, and I have found a few that need -15 or so.

    After taking a quick listen to the original, it is electronic music and as such, some of the natural hints are difficult to discern. However, it passed my first 'filter' because of the edge on the HF -- different than typical HF emphasis.

    The Howard Jones, DIA album is most likely DolbyA encoded. The tested threshold was between -15.25 through -15.75, where -15.75 was the best. It might even need another 0.1dB or so more.

    The examples (just after -- not before), assuming that you already have the album is on this repository (different than the others):
    https://spaces.hightail.com/space/BO4VDF6GNA

    The decoded files are dia01.mp3 and dia02.mp3. Both were encoded at approx 200k. If I were to produce a real distributed disk with these effectively 'master tape' copies, I'd do a little EQ, but as is, the results are typical. DolbyA tends to boost the lowest bass a little (at lower levels) and boost the super highs more with an unnatural hard edge. For example, the lowest bass (below 80Hz) would lose a few dB on average, but the peaks fully pass through. The super highs would typically sit at -12 to -15dB, but when something appears in the super highs the gain jumps up to about -6 - 3dB. The way that the highs are handled on DolbyA have really wierd effects, so it is hard to judge the HF based upon the single channel gains at all. All of the gains between MF(80-3k), HF0(3k-9k) and HF1(9k-20k) have to be taken together, because in reality the 80-3k range has major effect up to 12k+, but if that level is low, it is more like the other channels make up for the lack of MF gain. The freq ranges are not really all that separate. There is a quality flaw with DolbyA that is related to the attack times not being the same and the frequency range choices, but all-in-all it is a wonderful design for probably being done in the early 1960's (first version in approx 1964-1965). Ray Dolby was really special.

    The results sound natural, and the running gain ranges appeared to be typical for DolbyA encoding. Even in the difficult sections -- it appears that the L+R image is stable. IF it wasn't DolbyA encoded (or compressed by an separate channel sidechain compressor, which it might have been), the image would bounce around quite noticeably.

    John
     
  6. c-eling

    c-eling Forum Resident

    Thanks for taking the time John, appreciated.
    Yes I have four different manufactures for this album, three that carry the same levels and a possible Diament US.
    I'll get them on the big system and give a listen :cheers:
     
  7. John Dyson

    John Dyson Well-Known Member Thread Starter

    Location:
    Fishers, Indiana
    Suggest a redownload if you have already done it. I just reran the 2nd one, and found that the threshold was MUCH better at -15.90. This is one of the troubles with this decoding thing -- they erase the DolbyA tone, so it has to be done by ear. (I am a perfectionist!!!) I have to be VERY picky, because I work with gain control stuff, and I am very aware of the effects. I wasn't 100% sure until I corrected the threshold, now I am 99% sure this is DolbyA. Cannot even think about listening to FM radio!!! It weirds me out thinking about all of the crazy processing that they do.... The posting for the 2nd song is ready, and I am doing the 1st one again also. One of my major flaws is that I sometimes prematurally release things (dont get personal!!! :)).

    John
     
  8. Dave

    Dave Esoteric Audio Research Specialist™

    Location:
    Greater Vancouver
    Regarding Heart - S/T 1985 I had a discussion with Steve H about it and he did not mention that Dolby A was used simply that it was recorded that way. After checking both the Japan CP32 and the MFSL I agree and using de-emphasis just made it sound worse. John, if you can confirm it doesn't use Dolby A it would be great. Were they still using it in 1984-85?
     
  9. John Dyson

    John Dyson Well-Known Member Thread Starter

    Location:
    Fishers, Indiana
    I'll check -- but I suspect that DolbyA can be used to 'protect' the quality (actually just SNR -- doesn't really help headroom) when making any tapes, not necessarily just during mixes. Of course, DolbyA (however tedious) would be better used on less complex material because of the various dynamic distortion sources when using such a scheme. The more complex the music, the more opportunity for producing intermodulation (which is the major bug-a-boo for any kind of dynamic gain control scheme -- esp digital.)
    I'll look around for the Heart CD (or a digital copy somewhere) and see what I can find out. Might take a while -- I lucked out on the other thing...

    John
     
    Dave likes this.
  10. John Dyson

    John Dyson Well-Known Member Thread Starter

    Location:
    Fishers, Indiana
    Okay -- I lucked out. I checked the self-named album 'Heart' and I don't like what I am hearing when decoding it. At first, the slghtly-too-crisp sound and the compressed HF sound at the beginning almost sounds DolbyA encoded. Without trying to decode, I would have guessed that it was. However, when decoding, I got some effects that tell me that it is trying to decode something that wasn't appropriately compressed for DolbyA. Either this is the first DolbyA encoded material that my program has knowingly failed or not worked well on, or it is not DolbyA encoded. Could be that the threshold is set way too high -- so I tried a threshold a full 0.5dB lower than anything else that I have tried in the past, and it only helps a little. This is telling me that pushing the threshold down is making the decoding less active -- generally trying to generally increase the gain towards 0dB. I cannot find that 'correct place' to fit the decoder curve at all.

    I also noticed some slight problems with L+R, and the original didn't sound flat like undecoded material sounds. When I try to decode it, even in places when it sounds okay in a general sense, I detect some distortion which would imply that the decoder is trying to work with something that isn't encoded. DolbyA pushes the envelope for attack/decay times, and expansion seems to sound uglier than compression when it isn't appropriate.

    Most importantly, the lead vocal seems unnaturally distorted with DolbyA.

    Id give it a 25% chance AT BEST of being DolbyA encoded -- I'd say NO.

    So, personally, I'd listen to the copy of the first two tracks on Heart (1985) without decoding, and my doubts of DolbyA encoding are pretty strong.

    John
     
  11. Dave

    Dave Esoteric Audio Research Specialist™

    Location:
    Greater Vancouver
    Thanks John, so it's like Steve H told me then. I tried de-emphasis with not good results.
     
  12. John Dyson

    John Dyson Well-Known Member Thread Starter

    Location:
    Fishers, Indiana

    Geesh -- I wish one or the other worked well. I am going to try to avoid upsetting peoples sensibilities, so will back off a little -- I would have been happy if the DolbyA was the solution for you, but the sound sucks when trying to decode. I guess the music just has that 'sheen' and is meant to be. Normally the sheen (at the beginning of song 1 or song 2) is a good indicator, but not in this case. It JUST MIGHT be that DolbyA or similar was used to produce the sheen before mixdown -- but the entire recording is not DolbyA.

    Frankly, it would have been nice if CD players had the DolbyA built-in, but truly it either seems like someone was not being honest, that perhaps the DolbyA decoding was missing from the digital recording process somewhere -- or some other reason that I don't know.
    The DolbyA encoding has definitely been a bane against the digital distribution mechanisms, and it was about at the time of CDs that I started losing interest. I wonder about that... The weird thing is that I am very much into computing/digital processing and even operating systems design (I have been involved with writing much of the code of a rather important player early in the Internet), but always felt something odd about many digital recordings.

    Now, with the focus on the matter, I am pretty sure that a large percentage of the releases on digital media (in the olden days) were left DolbyA encoded. I am NOT claiming all, and will very strongly suggest when they aren't (like the Heart CD above.)

    More and more of the minds 'not set in stone' are starting to realize that I am not a crank, and this is a real matter.

    John
     
  13. John Dyson

    John Dyson Well-Known Member Thread Starter

    Location:
    Fishers, Indiana
    One more thing on this matter -- I ran through some more of the Howard Jones songs -- and I noticed that my decoder made a minor mistake on the decoding the 4th song on the album!!! Nothing major, but it is good to try to catch all of the bugs before someone else does :). I rather like the music -- thanks for turning me on to it. The mistake was like a dB or so a little too slow at a critical time -- the decoder MUST be very precise.

    Also, the bug -- I have been SUPER careful to shape the attack/decay dynamics so that the further filtering to hack out the intermod needs to do less work. Well -- some of my attack time calculations were tuned to be a little too slow, and I noticed it during one of the quiet passages on that 4th song!!! THANK YOU SO MUCH FOR HELPING ME -- and you didn't know that you were going to do so :)!!!

    This software is much more complicated than a typical compressor/expander -- there are lots of nooks and crannys that have to be dealt with before it is pro-quality. You helped me find one...

    Thanks again.
    And let me know when you get the chance to listen...

    John
     
  14. John Dyson

    John Dyson Well-Known Member Thread Starter

    Location:
    Fishers, Indiana
    I have found numerous cases of HF emphasized music to be DolbyA encoded or at least compatible with DolbyA decoding. Of course, if you can choose convenience, then just do some simple de-emphasis, but if it is DolbyA, de-emphasis only gets a small part of the job done. DolbyA decoding REALLY HELPS if it is needed. HOWEVER -- de-emphasis might be everything that can be done in many cases also...

    Most people know what pre-emphasis or other simple HF emphasis generally sounds like, but DolbyA encoding can sometimes sound similar. However, DolbyA has other manifestations, and if the material really benefits from DolbyA decoding, then there are other advantages like improved SNR, corrected dynamics, less LF distortion, better spatial depth, and probably other things that I haven't thought of right now.

    A difficult thing to overcome is the common wisdom that 'oh, gee -- it is just HF emphasis of some kind', and in some cases it really is just too much HF emphasis. In other cases -- in my specific collection, about 75% of the time, it really is DolbyA encoding or compatible. (The reason why I say 'compatible' is that I accept this as only a technical matter and KNOW that a DolbyA decoder can sometimes seem to work with non-encoded material, but NOT VERY OFTEN.)

    Firstly, I am not making any money on this, but I am providing a DolbyA compatible decoder for free that is primarily targeted for Linux, but the normally distributed version is for Windows. Currently, it resides on a repository that I disclose below. This is not a toy, and is very sophisticated when compared with a typical limiter/compressor or expander. It has to meet very specific criteria, and the attack/decay is so very fast AND NOT DESIGNED FOR SOFTWARE EMULATION, that it is not a simple piece of software.

    One of the goals for the decoder is that it really sound like a DolbyA, and in fact, some people say it sounds better, with close to the same freq response balance (unlike some others), and much clearer rendition of complex mixes of material (voices tend to stay cleaner when in chorus, for example.) This is related to the anti-intermod, superior level detection techniques, and careful avoidance both of intermod and aliasing (nonlinear operations like mixing gain control with the signal cause intermod which can alias.)

    So -- if you find that applying de-emphasis is not satisfying (probably one reason why I quit listening to music a lot when CDs appeared -- no longer satisfying sound much of the time), the DolbyA compatible decoding MIGHT HELP. Just recently, two people asked me to check -- and I found one album to be DolbyA compatible, and another NOT to be compatible. Again when DolbyA decoding helps, it REALLY helps.

    The filename usually looks like da-win-16may2018A.zip, where the date is the date that it was produced. 'A,B,C' are meant to be re-releases when I screw-up. There are two binaries provided along with necessary DLLS. One binary works on Intel/AMD, recent Atom-likes processors or better, and the other binary (da-avx) works on the newer i3,i5,i7,i9 type processors of the 4770 (Haswell) genre or newer. It might work on the 3770, but I haven't checked.

    Here is the repo, with some examples:

    Hightail Spaces

    If you are interested, give it a try -- ask me questions, or simply discuss this matter. This is open and meant for everyone to give input. It is easy to make a mistake by guessing or assuming that previous 'common wisdom' is correct. Really -- LOOK INTO MY EXAMPLES!!! I don't know why this has happened (the DolbyA encoding), but I can guess that there was a process difference between producing vinyl vs. digital media -- I just don't know why.

    John
     
    Yost likes this.
  15. Yost

    Yost Always Wondered How Other People Did This

    Interesting stuff. Is there a list available of CDs that benefit from Dolby A decoding?
     
  16. John Dyson

    John Dyson Well-Known Member Thread Starter

    Location:
    Fishers, Indiana
    I know of no list, and up until now (well, about 2mos ago), I didn't believe in the DolbyA encoding thing being so common either. I had a very precious copy of ABBA Gold which was DolbyA encoded, and I thought that it was very special. After further research (and a fully functioning DolbyA decoder after about 3-4mos of work part time), I found a lot of material to be DolbyA encoded. I do not want to suggest how much material in general is DolbyA encoded, but I'd suggest that IF IT IS DolbyA encoded, mostly it should be material produced in the middle 1990's or earlier -- more than likely no later than 1990 in reality.

    I don't know the pattern, but I just checked a Heart CD and found that it very unlikely is DolbyA encoded. On the other hand many of my ABBA disks (which aren't terribly damaged by excess compression) are DolbyA encoded, Nat King Cole, Sergio Medes/Brasil'66, Olivia Newton John, Carpenters also appear to be. I have a Simon&Garfunkel that isn't listenable without decoding -- the natural room vibrations and noise cause all kinds of problems because of the fast LF compression, and that disk REALLY needs decoding. Materal being decoded has to be DolbyA encoded or something very similar, or it would be all screwed up because DolbyA decoding is not very gentle. That isn't to say that something might come close to be decodeable and might even be improved, but unless it is truly DolbyA encoded it won't somehow sound 'right.' (One item of note, or perhaps two :), I have two digital 'albums' even purchased from HDtracks which are 99% likely DolbyA encoded: The 'unlimited' Paul McCartney album, and the Carpenters album. I am not claiming that they aren't enjoyable without decoding, but the come ALIVE when decoded. One of the most subtle improvements when decoding properly encoded material is that the spatial depth is recovered. Too often, a fast/linear compressor will flatten the spatial depth. The expander built-in to the decoder resurrects a lot of that lost depth -- even IF there isn't a lot of freq response damage. When a recording has a high average level, the amount of DolbyA HF emphasis diminishes, so the specific Carpenters album does have a high average level , so doesn't sound as 'shrill' as other recordings might.

    I SHOULD make a list, but 95% of my material is now on computer media (replicated in several places for security), and my CDs are languishing somewhere in storage. So, I wouldn't be too good about producing an absolutely accurate list.

    However, maybe once this catches on a little more, a movement should be started to catalog the material that benefits from decoding and/or de-emphasis.

    On a small scale basis, I am willing to do tests, but it might be much easier for each individual to break the ice and try a decode for themselves!!! :).

    John
     
  17. WiWavelength

    WiWavelength Forum Resident

    Location:
    United States
    I am sorry, but this theorizing over undecoded Dolby A noise reduction on digital releases comes across as far fetched. So, please answer some pertinent questions.

    What signal analysis are you doing to detect extant Dolby A encoding after the fact? Share some examples.

    If your signal analysis definitively can detect Dolby A encoding, how does it discriminate between intent and mistake? Dolby A encoding could have been left undecoded by artistic design. Dolby A encoding could have been could have been left undecoded by mistake on parts of a mixdown -- correcting that mistake by applying decoding now would cause errors in the other parts of the mix.

    If for sake of argument these all are mistakes, how would that happen across presumably dozens to hundreds of releases that involved thousands of people? Dolby A was a standard studio tool in the 1970s and 1980s, so it was a familiar process, not an uncommon one. Both rampant incompetence and widespread conspiracy seem unlikely.

    Absent further documentation, hypothesizing and applying Dolby A decoding to various digital releases because someone thinks they sound better that way is little different from hypothesizing and applying de-emphasis to various digital releases because someone thinks they sound better that way. It is conjecture that lacks evidence or even flies in the face of evidence.

    AJ
     
    Dan Steely and Tim S like this.
  18. John Dyson

    John Dyson Well-Known Member Thread Starter

    Location:
    Fishers, Indiana
    Firstly -- I know what DolbyA sounds like in various circumstances. Signal analysis for that kind of damage is difficult, so the best is the existence proof that the decoder works. Before asking very skeptical questions, do some tests and listen to the evidence. (Listen to the examples first, then do the tests -- it is easier if you'll never be convinced, but that is NOT my job.) If (AFTER LISTENING) you can honestly claim that the compression artifacts aren't removed and/or the spatial depth isn't restored and/or the hiss isn't removed, then we can talk about it. For the most simple example, please listen to the 'Scarborough Fair' example. Even the most 'dead' ears should be able to hear that improvement (even removing LF distortion by canceling it out.)

    DolbyA can DEFINITELY be left intact for artistic design, but would not usually be used in the entire mix. There was one recent example where the music had the 'sheen' of DolbyA, but wouldn't decode properly. It was quite common (including documented by CBS records) to use DolbyA for effect, but mostly not on the entire recording. It would sometimes be used to enhance a voice, or produce the sheen. Just because a portion of the recording was made with DolbyA, doesn't mean that it can be decoded.

    Your comment about DolbyA being a familiar tool is correct -- but DIGITAL WAS NOT A COMMON TOOL. I suspect that archives got digitized with the DolbyA encoding intact -- that is one very simple and realistic possible example, probably other possibilites also. Note that once the material was digitized, then DolbyA decoding would be a pain in the butt (especially in the olden days), so you have to d/a convert again, run through DolbyA, and then a/d again. Decoders like mine simplify that, but doing a DolbyA decoder (which mine has really been tested to be such) would not have been cheap or easy in digital back in the 1980's (or even the early 2000's, where even then you'd get stuck with a primitive toy.) My decoder does a lot more than a simple compatible expander -- you can hear things with mine that you couldn't hear with primitive decoders.

    REGARDING EVIDENCE -- please refer to that which I have provided, you don't HAVE to listen to the evidence -- but then I suggest not claiming authority about it.

    The best proof is the existence proof -- the reasons why are not important once you review the evidence.

    Thanks!!!
    John
     
    Last edited: May 16, 2018
  19. Massproductions

    Massproductions Forum Resident

    Location:
    Boston, MA, USA
    Dolby A encoding used on CDs? No such thing! Dolby A noise reduction is an analog noise reduction system used frequently on master tapes. You need the proper external encoder/decoder box.
     
    Nostaljack likes this.
  20. Massproductions

    Massproductions Forum Resident

    Location:
    Boston, MA, USA
    NONE! There is no such thing! Although it's possible that someone transferred a DOLBY A master tape and forgot to decode it, how would you apply a decoder to the CD unless you played back the CD player through an outboard DOLBY A decoder and re-recorded? Dolby A requires very careful calibration to the Dolby A tones at the start of the master tape. There's no way this can be done with a CD, ever!
     
  21. John Dyson

    John Dyson Well-Known Member Thread Starter

    Location:
    Fishers, Indiana
    No explicit DolbyA encoding was done for CDs and I never claimed that (nor implied.) The encoding was left on the original tapes -- please refer to my archive of my few examples (like I wrote above, even the most dead ears can hear the improvement of Scarborough Fair), and then lets talk.

    If you can find where I said that material was encoded for a CD, then please show me. Otherwise, I suggest reading what is written (even though I do have poor writing skills.) I don't think that there is any way that I have ever implied (even by mistake) that CDs were explicitly encoded. Mistaken left encoded, yes... Purposefully encoded, no...

    One of the first defects caused by DolbyA encoding is an HF sheen or APPARENT pre-emphasis (usually, depending on average level it is better or worse.) There are numerous other modifications made by DolbyA, but whether or not one wants to philosophically discuss whether or not it has happened -- there is existence proof that it does. AND my decoder sucks when you try to decode something that isn't encoded by DolbyA. That is the easiest (but only perhaps 95% conclusive test) about DolbyA encoding.

    repo: Hightail Spaces
    another demo: Hightail Spaces


    John
     
    Last edited: May 16, 2018
  22. Mal

    Mal Phorum Physicist

    What did you do to "Luka"??

    I don't remember the original Solitude Standing LP and CD being drastically different from each other and neither sounded anything like your decoded clip...
     
  23. John Dyson

    John Dyson Well-Known Member Thread Starter

    Location:
    Fishers, Indiana
    The only thing that I did to Luka was decode it and normalization. My original -raw version is highly compressed -- as if DolbyA... The DAdecode version isn't so compressed. Even old vinyl got a dose of 'love' before producing the final version. This doesn't have any 'love' and is pretty much a raw copy (maybe I did a little to it, but not much -- it is all automatic now, I'll check.) Sometimes I add 1.5dB at 45Hz, but that is usually pretty close to it. (That is to make up for LF limitations that they use to assume for vinyl -- actually should be stronger.)
    Oh -- I just checked -- I ran the unexciter (anti-aphex exciter on a few examples also -- esp ABBA examples and Petula Clark -- they were especially egregious, so i had to fix them. I hate the sound of the exciter.)

    BTW -- I am fighting a DDOS attack right now, must have irritated someone :).
     
    Last edited: May 16, 2018
  24. Mal

    Mal Phorum Physicist

    [​IMG]
     
    John Dyson likes this.
  25. kevin5brown

    kevin5brown Forum Resident

    I think too many people might be getting too caught up in the details. I see this simply as: if you think a disc doesn't sound right, try to deemphasize it. If it still doesn't sound right, then go back and try Dolby A decoding of the original CD. It either sounds better or it doesn't. Done. :)

    We have found CDs that we think should be deemphasized in the long PE thread, but don't have any PE flags. Should they have been flagged? So now we can try deemphasis, and now we can also try Dolby A decoding.

    I can 100% believe however, that some Dolby A encoded master tapes accidentally got transferred to CD without the Dolby A being decoded 1st though.
     

Share This Page