The No.1 Website for Pro Audio
About DolbyA, and why it seems to lurk in the consumer realm
Old 21st May 2019
  #1
Gear Head
 

About DolbyA, and why it seems to lurk in the consumer realm

This is a difficult subject to breach -- and is very likely to open up a bit of controversy. I am speaking about the actual issue of DolbyA encoding ieaching the consumer fairly often (I do NOT have statistics, and a moderly small sample set of maybe 30-50CDs and downloads.)

There are some diffiiculties and cause for some skepticism with this observation, and those include: not everyone has a DolbyA to double-check the problem, no calibration tones, sometimes no reference material, also an automatic distrust of an unknown person with unknown motives. I acknowledge these and other caveats.
Also, I can present plausible scenarios that describe why the DolbyA leaks occur.

First, I'll try to describe the matter in the most clear way that I can (my language/composition skills are that of a 6th grader :-)). My brains are all focused elsewhere.
The story starts with a really shrill sounding copy of ABBA Gold/1992. It was really bad -- something was wrong. Sounded like compression, but not 'normal' compression. Really, really bad, and I am not a 'golden ears'. Since I already design compressors/expanders/etc, I started playing with the recording -- and after playing off and on for a few weeks, I thought -- could it POSSIBLY be DolbyA? For some reason, that popped into my mind... Long long story follows that... Basically I started finding some of the same defects -- but usually partially hidden with a bit of EQ (something like treble cut [email protected]/Q=0.707 or maybe -6dB.) Not all material had these defects -- on SOME recordings, something was really similar about the recording problems.

Over the years, I started a research project that turned into a pseudo-commercial project (the kind that was never meant to make money -- a nuisance fee type thing.) I spent a few years with a semi-known recording/restoration engineer to develop a DolbyA decoder. It was a VERY tedious project, a very compatible decoder not done very well until now. (NO-ONE would ever make money creating such a dinosaur, and it is not a weekend project.)

This decoding tool (and admittedly a hobby out-of-control) has been helpful in finding material that suspiciously had that 'flat stereo space', 'compressed high end', 'semi-harsh sound' -- often could be TOTALLY cleaned up by doing a DolbyA compatible decoding operation on it. This statement of fact leads to the very high probability that the material at least has some kind of residual DolbyA encoding character. (The material might have been subsequently compressed, etc. It might not be pure, pristine DolbyA encoded material... I am NOT advocating that consumer do their own DolbyA decoding -- it is not pleasant without the calibration tones.)

Please read the following, more complete message -- and refer to the links for existance proof. BTW, a correspondent of mine found that one of his recordings sounds almost exactly the same as my decoded Cars example!!! So, there is 100% proven that there is both a decoded and undecoded copy available for sale, but which version are you going to end up purchasing? -- no-one knows!!!

Here is a message that I sent elsewhere (this is cut and paste, so it might have some off topic info, but seems okay):

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

All of this information is meant to let you know that I really AM trying to do good things, I am not (usually) a trouble maker, and have legit qualifications (AT&T Bell Labs engineer for 15yrs, Technicolor/Thomson for 5-6yrs, wrote big part of FreeBSD kernel, DSP/Analog HW, real-time/OS software expertise.) No weekend hobbiest here -- I take this matter seriously AND I CARE.

Be patient when reading this -- it might be contrary to 'common wisdom' in some circles, but is demonstratively true and I have numerous cases of existence proof (300-500 recordings from perhaps 30 or so CDs or downloads.) Given my music selection, I'd estimate between 30-50% of my digital recordings/CDs are afflicted with the problem.

*Over the last few years -- I found a dirty little secret, and is slowly but surely being proven -- a lot of material is being distributed (esp stuff created between late 1960's through early 1990's) with the DolbyA encoding left intact. * DolbyA encoding doesn't sound horrible, but without decoding, it does flatten the spatial depth, compresses the high end (sometimes sounding a bit distorted), and just doesn't sound as good as it could. DolbyA without decoding is nowhere as bad as DBX NR, for example.

I am showing you some links to some examples -- just to demonstrate that I am not a crazy person... One pair of links is from a CD (directly) that has all of the characteristics of undecoded DolbyA... Then, I did a DolbyA equivalent decoding on it -- voila -- much much better.

Also supplying you a link to a directory of comparisons: Some are decoded/undecoded. Some are decoded-by-my-decoder, commercial version, vinyl version -- you got the idea.

I am trying to wake-up people (because of the history of people used to claim: 'digital sounds harsh', 'digital screws up the soundstage' or whatever like that.) This 'revelation' might be 1/2 of the problem. I have also found that a lot of mastering/recording engineers tend to discount this -- until the existence proof is accepted. But, no UFOs here. Most, relatively open minded people who know DolbyA eventually accept the fact (and the Cars example is just one of many.)

My DolbyA demo program is a RESEARCH project that turned into something that a few pros use and is being sold on the PRO side for a nusance fee. When a consumer is really interested, I give away a free copy (with a simple license manager) -- the goal has not been to profit, but to illuminate -- and start to fix -- a real problem. I want to see fewer DolbyA leaks, and more reliable good quality sound. SO, I use the demo software DolbyA decoder (much, much much cleaner than an actual DolbyA) for my comparisons, but a DolbyA HW unit (whether 301A, cat22/361 or 363) could be used just the same. (We -- my project partner and myself) have tested against DolbyA HW units with significant success and show the superiority of the DHNRDS decoded by far most of the time (btw -- my SW decoder IS NOT a weekend hobby project, and is not a 'quick fix', but TRULY emulates the operations for DolbyA -- down to the smallest little nit.)

Here is the straightforward link to the song from the the direct contents of the CD:
https://www.dropbox.com/s/y8czwibps2...coded.mp3?dl=0

Here is EXACTLY the same file as processed by a DolbyA equivalent decoder:
https://www.dropbox.com/s/bfertqaeab...coded.mp3?dl=0

Here is a link to multiple examples -- note that I don't always get the calibration right because of the lack of calibration tones -- so please be tolerant about calibration errors (sometimes causing a bit of gating or surging.) I think that I have fixed most of the problems as detected by some other correspondents -- but might have missed some also:

https://www.dropbox.com/sh/mduka8faq...sDN5Dv7Aa?dl=0

*IF really helpful, I can also provide 96k/24bit examples if my use of mp3 is insulting :-). The main reason for using mp3 is that the online player on Dropbox doesn't seem to do .flac. I have limited Dropbox space, but I have more than 300 songs from various recordings that are very likely DolbyA encoded, which is the largest part of my collection... All afflicted recordings are from the Simon&Garfunkel era to at least Cars timeframe. (I do have some Nat King Cole that SEEMS to be DolbyA encoded, so must be a mixdown to stereo -- probably done on 3 track? So, I am not sure about the Nat King Cole example.)

Thank you
Old 21st May 2019
  #2
Lives for gear
 
bitman's Avatar
Fascinating read. The decoded Cars song does indeed sound better.
I have a Jo Dee Messina CD that is bright as the morning sun. On cheap system it's like ice picks in the eye.
I know that the louder and brighter (like people) garner the greatest attention but it's not fun to listen to over time.

I would love to have a copy of your decoder just to see what this disc sounds like decoded.
Old 21st May 2019
  #3
Lives for gear
 
FabienTDR's Avatar
 

Verified Member
To be honest, I fail to follow.

Dolby A is a magnetic tape specific "companding" method. A multi-band compressor to encode, and an ("almost") inverse multi-band expander to decode.

It has absolutely nothing to do with CD, and only severe idiots would ever put a dolby A encoded signal onto a CD. If it happens, it was a mistake/defect. It only makes sense on nonlinear and noisy media, which CD simply isn't. Using dolby A on a CD produces irreversible aliasing and IMD, for absolutely no benefit. On top of this, no consumer would ever be able to decode it anyway.


Are you really sure this isn't prehistoric U-Law or A-law WAV encoding? Likely by production mistake. I'd guess it is one of those, and NOT Dolby A.

Last edited by FabienTDR; 21st May 2019 at 05:16 PM..
Old 21st May 2019
  #4
I have a 24 track SP-24 Dolby rack here for sale loaded with all Dolby A cards. Anyone want to buy it and test this stuff?
Old 21st May 2019
  #5
Gear Head
 
S_mask's Avatar
 

Peep the D-code. I found 8 gematriot matches between 'Dyson' and 'Dolby'. Dyson writes us that he's trying to reverse the Dolby process, et, voilà, each 'Dyson' hit is simply the reverse cipher to the 'Dolby' match? Auguri.

1)

"Dolby" = 23
(Reverse Full Reduction)
D o l b y
5 3 6 7 2

"Dyson" = 23
(Full Reduction)
D y s o n
4 7 1 6 5



2)

"Dolby" = 233
(Satanic)
D o l b y
39 50 47 37 60

"Dyson" = 233
(Reverse Satanic)
D y s o n
58 37 43 47 4



3)

"Dolby" = 77
(Reverse Ordinal)
D o l b y
23 12 15 25 2


"Dyson" = 77
(English Ordinal)
D y s o n
4 25 19 15 14




4)

"Dolby" = 58
(English Ordinal)
D o l b y
4 15 12 2 25


"Dyson" = 58
(Reverse Ordinal)
D y s o n
23 2 8 12 13



5)

"Dolby" = 84
(Francis Bacon)
D o l b y
30 15 12 2 25

"Dyson" = 84
(Reverse Francis Bacon)
D y s o n
49 2 8 12 13



6)

"Dolby" = 22
(Full Reduction)
D o l b y
4 6 3 2 7


"Dyson" = 22
(Reverse Full Reduction)
D y s o n
5 2 8 3 4



7)

"Dolby" = 103
(Reverse Francis Bacon)
D o l b y
49 12 15 25 2

"Dyson" = 103
(Francis Bacon)
D y s o n
30 25 19 15 14


8)

"Dolby" = 115
(Franc Baconis)
D o l b y
7 30 24 4 50


"Dyson" = 115
(Reverse Franc Baconis)
D y s o n
45 4 16 24 26



(9) freebie extra - Sumerian = ordinal values x 6)

"Dolby" = 462
(Reverse English Sumerian)
D o l b y
138 72 90 150 12

"Dyson" = 462
(English Sumerian)
D y s o n
24 150 114 90 84



Check it and see: http://www.gematrinator.com/calculator/index.php

(gotta click 'ciphers' and then 'all ciphers' to see all the accepted ways to pin an Arabic numeral to a Roman letter)
Old 21st May 2019
  #6
Lives for gear
 
Greg Reierson's Avatar
 

Verified Member
Quote:
Originally Posted by FabienTDR View Post
... only severe idiots would ever put a dolby A encoded signal onto a CD. If it happens, it was a mistake/defect.
Yes, exactly. It's when a reel was either not labeled or misidentified and not properly decoded when creating the master. Some label catalogs are notorious for missing this sort of information. I spent countless hours doing 'greatest hits' type compilations back in the '90s trying to figure out which of the 12 songs on a particular reel were NR encoded, what the encoding was, guessing as a ref level, etc. It happened a lot.
Old 21st May 2019
  #7
Gear Head
 

Quote:
Originally Posted by FabienTDR View Post
To be honest, I fail to follow.

Dolby A is a magnetic tape specific "companding" method. A multi-band compressor to encode, and an ("almost") inverse multi-band expander to decode.

It has absolutely nothing to do with CD, and only severe idiots would ever put a dolby A encoded signal onto a CD. If it happens, it was a mistake/defect. It only makes sense on nonlinear and noisy media, which CD simply isn't. Using dolby A on a CD produces irreversible aliasing and IMD, for absolutely no benefit. On top of this, no consumer would ever be able to decode it anyway.


Are you really sure this isn't prehistoric U-Law or A-law WAV encoding? Likely by production mistake. I'd guess it is one of those, and NOT Dolby A.
First, I tend to agree about the severe id'jits, but maybe not so strongly. Just listen to the demo versions that I have provided.

Here is a new anecdote -- just found out today. The Cars CD from where I generated the Cars example had material that is the same as a premium 192k/24bit release.. A corespondent on another forum provided me with a basis of comparsion. The decoded version that I produced from the 'rogue' CD (of which I have many from different groups) came out super-duper close to the fully (correctly) mastered version on the premium download.

The difference between my locally decoded version and the premium release ended up being that my local copy had more clean/distinct transients (e.g. on the song 'Im not the one"), mixed vocals were more distinct, and the underlying hash generated from the sythesizer being in the decoded material was much suppressed. My own decoder is NOT a DolbyA, but is completely interchangeable for decoding purposes. The purpose of this discussion is NOT my decoder, but rather that some material (how much? a significant amount) is being distributed with the DolbyA encoding (which is a kind of compression) intact.

In summary, I have the 'undecoded', 'rogue' CD, which after decoding produced a result which is super-similar to the 'premium 192k/24bit release', but each is distinct enough to demonstrate that they are indeed different.

This revelation from my friend about the 'premuim release' is new since I produced the before/after decode of the song 'Moving in Stereo' as pointed to on the previous message.


As I wrote before -- this is NOT a UFO.

John
Old 21st May 2019
  #8
Gear Head
 

Quote:
Originally Posted by Greg Reierson View Post
Yes, exactly. It's when a reel was either not labeled or misidentified and not properly decoded when creating the master. Some label catalogs are notorious for missing this sort of information. I spent countless hours doing 'greatest hits' type compilations back in the '90s trying to figure out which of the 12 songs on a particular reel were NR encoded, what the encoding was, guessing as a ref level, etc. It happened a lot.
So, IS THAT HOW IT HAPPENS? I had been going though all kinds of guesses about how the files might have been archived digitally, but it would be too expensive to do a decode passing through a DolbyA taking 1:1 realtime... Maybe I made it too complciated.

Could have just been labelling issues/errors? WOW!!! Sounds like Occams razor... Simplest cause of the problem...

Since I mostly like music from between 1960s through early 1990s (about when MTV quit being MTV), I get a lot of the legacy greatest hits or benign re-releases (not the loudness wars stuff), and that selection set seems to have a very substantial amount of DolbyA leaks (that what I call it.)

So the mechanism is at least partially that of labeling, and not necessarily a choice to avoid the cost of dealing with the DolbyA?

Cool -- I like it when I learn something new :-). This thing has been perturbing me for 3yrs, even motivating me to write a fully capable DolbyA decoder (eliminates all of the distortion mechanisms -- no more fuzz, vocals kept clean/separate/etc.) I spent three years because of labeling errors? :-).

John
Old 21st May 2019
  #9
Gear Head
 

Quote:
Originally Posted by Jim Williams View Post
I have a 24 track SP-24 Dolby rack here for sale loaded with all Dolby A cards. Anyone want to buy it and test this stuff?
Thank you, but my project partner has the DolbyA units (I think both the cat22/361 and the 363 -- I don't think that we have access to an A301), Telcom C4 (next project), DBX I/II, etc.

At one time, I would have loved to have had a 361, but all of the testing is now done, it has been carefully compared/etc. Not touching the program any more -- files are protected. It was too difficult to screw up.

My project partner would like for us to do a DolbySR, but I wont' be living another 50yrs to get it done :-). SR is crazy complicated when compared with a DolbyA. (DolbyA is the worst project that I have ever done, and I have done some amazing things.)

John
Old 21st May 2019
  #10
Lives for gear
 
FabienTDR's Avatar
 

Verified Member
Are you really sure it isn't U-Law/A-law PCM? This is a long abandoned "feature" (vs today's widely used "linear PCM").


Quote:
Logarithmic PCM

Rather than representing sample amplitudes on a linear scale as linear PCM coding does, logarithmic PCM coding plots the amplitudes on a logarithmic scale. Log PCM is more often used in telephony and communications applications than in entertainment multimedia applications.

There are two major variants of log PCM: mu-law (u-law) and A-law. Mu-law coding uses the format number 0x07 in Microsoft multimedia files (WAV/AVI/ASF) and the fourcc 'ulaw' in Apple Quicktime files. A-law coding uses the format number 0x06 is Microsoft multimedia files and the fourcc 'alaw' in Apple Quicktime files.
From here: https://wiki.multimedia.cx/index.php/PCM
Old 22nd May 2019
  #11
Lives for gear
 
FabienTDR's Avatar
 

Verified Member
You specifically mention the CD format. Afaik, Red-book doesn't allow the a-law or u-law I mentioned above. But it has this ominous pre-emphasis flag, also long forgotten:

Quote:
Some CDs are mastered with pre-emphasis, an artificial boost of high audio frequencies. The pre-emphasis improves the apparent signal-to-noise ratio by making better use of the channel's dynamic range. On playback, the player applies a de-emphasis filter to restore the frequency response curve to an overall flat one. Pre-emphasis time constants are 50µs and 15µs (9.49 dB boost at 20 kHz), and a binary flag in the disc subcode instructs the player to apply de-emphasis filtering if appropriate. Playback of such discs in a computer or 'ripping' to wave files typically does not take into account the pre-emphasis, so such files play back with a distorted frequency response.
Of course, only very few players support this.

IMHO the above comes closest to your observation. I really don't assume that you are hallucinating, don't get me wrong pls. I just think there are many other potential causes.
Old 22nd May 2019
  #12
Gear Head
 

Talking about the pre-emphasis issue -- the DolbyA and pre-emphasis issues are different problems.

I have a Dropbox repository of numerous examples -- some Polar music decoded DolbyA material, some material decoded by me, some undecoded, etc, etc, etc.

Below is a pointer to a Dropbox repository. SORRY ABOUT THE MP3 -- I can supply .flac for anything in the repository.

One of the examples is '99 Red Balloons' from Nena. This recording had BOTH the CD pre-emphasis issue and DolbyA. I produced a step-by-step dissection:

Direct from CD:
01-99 Red Balloons-undecodednoDEEMPH-60.mp3

CD deemphasis added:
01-99 Red Balloons-undecodedyesDEEMPH-60.mp3

DolbyA decoding added:
01-99 Red balloons-decoded-60.mp3

That specific CD had BOTH the CD pre-emphasis and DolbyA encoding. Before suggesting that the material is not DolbyA encoded -- listen to the before and after.

There is also a very obvious improvement on the undecoded vs decoded version of ABBA's 'The Day Before You Came'.

DolbyA encoding doesn't just cause a shrill sound (handled by a bit of EQ), or a compressed high end -- it also flattens the stereo spatial imaging. Proper decoding restores the stereo imaging.

Example repository:

https://www.dropbox.com/sh/mduka8faq...sDN5Dv7Aa?dl=0
Old 22nd May 2019
  #13
Gear Head
 
S_mask's Avatar
 

Be advised that tape restoration guru, Richard Hess, is working with John http://dhnrds.com/, so, he probably is on the right path.

Though not as auspiciously as 'Dyson', 'Hess', also, synchs right up to 'Dolby'...

"Hess" = 77 (Francis Bacon)
"Dolby" = 77 (Reverse Ordinal)

"Hess" = 23 (Septenary)
"Dolby" = 23 (Reverse Full Reduction)

"DHNRDS" = 223 (Francis Bacon)
"Dolby A" = 2203 (Reverse Squares)
(0 doesn't 'count')

Also, here's another thread on this phenom from 2011: https://forums.stevehoffman.tv/threa...olby-a.248968/
Old 22nd May 2019
  #14
Gear Head
 

Quote:
Originally Posted by S_mask View Post
Be advised that tape restoration guru, Richard Hess, is working with John http://dhnrds.com/, so, he probably is on the right path.

Though not as auspiciously as 'Dyson', 'Hess', also, synchs right up to 'Dolby'...

"Hess" = 77 (Francis Bacon)
"Dolby" = 77 (Reverse Ordinal)

"Hess" = 23 (Septenary)
"Dolby" = 23 (Reverse Full Reduction)

"DHNRDS" = 223 (Francis Bacon)
"Dolby A" = 2203 (Reverse Squares)
(0 doesn't 'count')

Also, here's another thread on this phenom from 2011: https://forums.stevehoffman.tv/threa...olby-a.248968/
My DolbyA compatible (not sanctioned by Dolby) decoder project started as a side effect of the search for the answer -- so happy that I just got a specific answer after so many years.

People working with me on the DHNRDS DA decoder are working on the *decoder* not on the 'windmill tilting' search for the reason for DolbyA leaking into consumer hahds.

PS: the DolbyA compatible decoder, when it comes to distortion (stuff like IMD) and clarity of transients, etc -- is fantasitc -- recovering almost the original input to the DolbyA at the beginning. On my tests, even of material not used for testing during the development -- sounds very similar to (but cleaner than) the original hardware.

Bad news about the decoder: it is NOT easy for consumers to use. I have always vehemently hated writing GUI software, so the program is command line, and requires tedious calibration if the 'DolbyA' tones are not available.

But -- THANK YOU SO MUCH for the pointer on where the leaked DolbyA material comes from. We are starting a Telcom C4 decoder project -- it has gotten a slow start, but when it starts moving more quickly -- it will probably start working in about 1 month (I have learned A LOT and developed a LOT of technology during the DA development.)


John
Old 23rd May 2019
  #15
Deleted 691ca21
Guest
Be great to have it as a plugin eventually, to test out any releases that sound a bit off.
Old 23rd May 2019
  #16
Gear Head
 

Quote:
Originally Posted by Deleted 691ca21 View Post
Be great to have it as a plugin eventually, to test out any releases that sound a bit off.
It is truly my fault that there isn't a plugin or GUI version. If someone who already knows plugin programming (for a given package) or do GUI programming, I am definitely willing to support it with the IP from the decoder.

The biggest problem for doing the decoder as a plugin is that the decoder is very multithreaded. On might ask 'why is something so very simple multithread?', and the answer is -- it isn't simple in the more complex modes. In the highest quality modes, the decoder will 100% occupy a 4 core Haswell Intel box for realtime play. If the program would not be multi-threaded, it would run very slowly. When fully enabled, the program does a lot to dig-out all of the modulation distortions associated with the kind of gain control needed.

Lower quality modes can probably be done easily as a normal plugin -- the lower quality modes ARE very close to DolbyA HW quality -- the 'hghest' low quality mode is a little better -- it does all of the straightforward nonlinear control-signal pre-filtering that help with the modulation distortions, but just is not mathematically possible to do the really big clean-up that the higher modes can do. As a single thread, the decoder would run approx 5X faster than realtime on an up-to-date (within last 5yrs, AVX2 capable) CPU.

The GUI iissue is another story -- I am really good at doing the kind of programming in the 'meat' of the decoder, but totally deficient in the ability to visualize GUI type stuff in a way that it can be expressed. Basically -- and I mean this literally, I have a kind of disability in representing things in a drawing -- so doing GUI is as confounding to me as a non-programmer would have doing the actual decoder program.

When adding all of these things up -- and recognizing that the market for the decoder is micro-small, the only benefit for someone joining the team and doing a GUI version is the small amount of 'boost' that they'd get for their name being associated with the project.

If I could focus 100% of the time when doing a GUI version -- it might take me a year to do some kind of simple GUI. I was originally going to try to do something simple in Java, but my project partner (rightfully so) said that there is a bit of a negative aspect to Java -- so I had to give up on that. (Java would be the only technology that I know that could give me a chance to do something primitive on the screen.)

I would really like for the decoder to be easier to use. It is REALLY REALLY REALLY good. After running a leaked DolbyA Cars CD -- and comparing it with another friend's 192k/24bit professionally mastered (correctly) download -- the decoded CD version came out very favorably. It was NOT a skills issue with the professional mastering person, but the improvement on my version was purely the decoding technology. I cant get a full mastering & care and feeding to work correctly -- so all I do is a direct decode -- I make WAY too many mistakes.

Old music WOULD be much more clean using the DHNRDS decoder -- IT IS NOT PERFECT, but I worked on it so intensively and with absoute intolerance for mistakes -- it really frustrated my project partners. They put up witih a lot of over-perfectionism -- but I wanted the results to be as perfect as I could do... And -- right now the source code is protected for no modifications -- it has hit the point that it cant be any better for apparent sound quality -- we might still benefit from a few calibration tweaks, but we have exhausted all references in our posession...

John
Old 23rd May 2019
  #17
Deleted 691ca21
Guest
Quote:
Originally Posted by John Dyson View Post
It is truly my fault that there isn't a plugin or GUI version. If someone who already knows plugin programming (for a given package) or do GUI programming, I am definitely willing to support it with the IP from the decoder.
@ chrisj from AirWindows might be interested? Although he does GUIless plugins...
Old 23rd May 2019
  #18
Gear Head
 

Quote:
Originally Posted by Deleted 691ca21 View Post
@ chrisj from AirWindows might be interested? Although he does GUIless plugins...
If it would help convince someone who was REALLY considering doing an interface for the decoder, I can provide many fantastic examples (imagine ABBA without much audible IMD produced during decoding -- material from ABBA was some of my worst case test material because of the mixed/choral vocals -- almost perfectly hitting the ltricky places for DolbyA.)

I have a lot of (legally obtained) leaked 'pop' material from (mostly) the 1970s, but some from the 1980s -- and just a little bit from the 1960s for quick demos. Some of the results are incredible, some are 'okay'.

The DHNRDS truly sounds a lot like a DolbyA -- with less modulation distortion.

(caveat about the highest quality modes -- a lot like a soft focus on a camera, the slgiht fuzz of a the true DolbyA HW -- or the DHNRDS running in one of the lower quality modes, can sometimes be a good thing -- also, the higher quality modes on the DHNRDS do motivate towards using more precise calibration than can be attained by matching calibration levels.)

The DHNRDS was really written (sometimes with some pain) to be somewhat timing accurate also. The input/output timing error (file start/file size) is in the several sample range (a little worse at 192k, but still AFAIR, in the 10 sample range -- maybe a bit more, maybe a bit less.) The vast amount of math and transforms/etc, made the program a bit tedious in regards to precise input/output timing alignment. WRT the actual internal timing WRT the gain control/audio -- it is dead-nuts perfect. We tried to make it as easy as humanly possible to deal with multi-track issues -- and it was a very tedious endeavor to get the timing alignment as close as it is --actually it might be in the several sample range at 96k -- but I truly do not remember right now.



John
Old 23rd May 2019
  #19
Gear Guru
 
Thomas W. Bethe's Avatar
 

Verified Member
There is already a decoder for Dolby A that works GREAT. Here is the website. https://u-he.com/products/satin/ Pro tools just released a similar plugin. FWIW
Old 23rd May 2019
  #20
Lives for gear
 
bitman's Avatar
Quote:
Originally Posted by Greg Reierson View Post
Yes, exactly. It's when a reel was either not labeled or misidentified and not properly decoded when creating the master. Some label catalogs are notorious for missing this sort of information. I spent countless hours doing 'greatest hits' type compilations back in the '90s trying to figure out which of the 12 songs on a particular reel were NR encoded, what the encoding was, guessing as a ref level, etc. It happened a lot.
Well, that gives me something to take with me as I continually ponder why the Monkeys greatest hits CD I have is unlistenable junk. I always just figured the records I wore out as a child sounded better that they did.
Old 23rd May 2019
  #21
Gear Head
 

Quote:
Originally Posted by Thomas W. Bethe View Post
There is already a decoder for Dolby A that works GREAT. Here is the website. https://u-he.com/products/satin/ Pro tools just released a similar plugin. FWIW
Satin isn't nearly as accurate... We did some comparisons -- doesn't really sound the same as a DolbyA -- but it is all what people like/perceives themselves. (Like my somewhat 'earthy' grandma used to say -- whatever blows your skirt up.)

Sounding good doesn't count IMO -- sounding JUST LIKE a DolbyA sans distortion is important. Refer to the demos. The most recent Cars comparisons (compared with a 192k/24bit professionally done CORRECT release, vs a leaked CD being decoded by DHNRDS -- they sound literally almost the same -- the DHNRDS doesn't have the veil of distortion.)

John
Old 23rd May 2019
  #22
Gear Head
 

Sorry (REALLY) about the previous emotional response. I just listened to a test Satin decode vs. the DHNRDS vs. the DolbyA (the 361/cat22 version) -- I also have tests on the 363, but too many things to compare. Note that the DHNRDS is a true 3.5 band decoder (that is, it does the full 0-80, 80-3k, 3k-20k, 9k-20k decode -- kind of strange merging of the 3k-20k and 9k-20k bands to get the extra 5dB of NR.)

It has been a LONG time since I listened to the Satin decode, and I wanted to refresh my memory (and double-check what I wrote earlier.)

Here are the differences per my perception (not an exacting measurement):
All relative to the DolbyA: (ANY difference from a DolbyA is a negative -- less his than aa DolbyA is NOT a good thing -- accuracy is the only good thing.)

Satin -- more hiss (seems like 3-6dB more), a bit more 'thin' sounding, ALMOST as if Satin was doing a single band expansion? Not sure what they did. The low end room rumble seems VERY thin when compared with the DolbyA.

DHNRDS -- about the same/maybe a bit less hiss than the DolbyA (the MF filter is a bit more narrow on the DHNRDS than the DolbyA -- so that can account for the hiss difference -- it is NOT a plus that the DHNRDS has a little less hiss -- IT SHOULD BE THE SAME, but isn't quite.)

Also, the DHNRDS has a bit stronger low-end vs DolbyA on the test. That is also a mismatch, and I can actually quantify that, because I have measured it. There slghtly greater low end (on material sitting at the approx -15 to -20dB peak level (about -25 to -30dB RMS), with lots of room rumble.) The reason for that is two-fold -- the gain curve for the DHNRDS is about the same a bit stronger in the audible low-end (maybe about 0.5dB lower at 20Hz and maybe about 0.25dB higher at 30Hz), but also there is a 'hump' in the desired decode response in the approx 100Hz region. The 'hump' is a bit larger on the DHNRDS than a true DolbyA. The hump doesn't exist at the very highest and very lowest levels, but there is a point in between where there is a several dB hump that appears in the DESIRED response. Both of these differences account for the very slightly greater (but very clean) room rumble on the DHNRDS. I found that on the specific test, that a very slight HPF at about 70Hz (like 1dB/Q=0.707) corrects the difference. (During our design choice discussions, it appeared that a 'bit more' low-end might be more desirable than a 'little less' because it was felt that the DolbyA was a little thin.) Another, more important difference between the DHNRDS and DolbyA is that the highs & chorus are more clean and more distinct.

It is more difficult for me to directly compare the Satin results with the DHNRDS results because of really strong bias on my part. But, I can separate myself enough when comparing the Satin vs DolbyA or the DHNRDS vs DolbyA.

But, in a casual comparison -- on material that isn't full of room rumble, the similarities between DolbyA and DHNRDS are somewhat astounding -- even to me.

Here is a pretty severe example of the difference (unforutnately truncated) between DHNRDS and DolbyA (note much less IMD in the synthesizer on the DHNRDS):
DolbyA:
https://www.dropbox.com/s/4kd1vfi9rk...lbyA.flac?dl=0
DHNRDS:
https://www.dropbox.com/s/ajj53hlvv4...NRDS.flac?dl=0

Perhaps most important thing -- I am intending on releasing the technology (not necessarily the decoder), because there is some 'holy grail' code in there (the IMD suppression.) I think that the technique should be available to everyone. Right now, I am searching for someone technically adept enough at DSP to help me demonstrate the math...



John

Last edited by John Dyson; 23rd May 2019 at 04:34 PM.. Reason: Add a comment...
Old 23rd May 2019
  #23
Mastering Moderator
 
Riccardo's Avatar
 

Verified Member
Quote:
Originally Posted by Thomas W. Bethe View Post
There is already a decoder for Dolby A that works GREAT. Here is the website. https://u-he.com/products/satin/ Pro tools just released a similar plugin. FWIW
What's the name of the second one? (the Pro Tools one you mentioned)
Old 23rd May 2019
  #24
Gear Head
 

Quote:
Originally Posted by Riccardo View Post
What's the name of the second one? (the Pro Tools one you mentioned)
Yea -- I'd actually like to hear a full comparison of all three -- might be really educational. :-).

If a different approach from the DA that I wrote really works well (sounds similar, and also provides the full NR) -- I'd actually like to know it and a little bit about what they did (I don't want to know any secrets.) One thing for sure -- the method as described in the Sony patent for DolbyA is WRONG -- there is a lot missing, and trying to actually emulate a DolbyA with a direct converison of the HW design will just not work (I got about 6dB of gain control -- the problem is that the HW audio feedback scheme cannot work in the software world.)

Actually ferreting out the front-end attack/release mechanism was the most difficult of the entire project. I wasted 2/3 of the time on that -- the attack/release is very complex and dependent on numerous factors -- an RC time constant (or even several RC time constants) will not work correctly. I finally had to 'design' it, trather than to use normal kinds of cookbook design or design from experience.

If someone else TRULY emulates a DolbyA for decoding (3-4 bands, real NR, etc), I'd really like to communicate with them... There are lots and lots of odd stumbling blocks, and just because someone has designed a 'really good compressor/expander' doesn't prepare them for the DolbyA thing. That would be a very special person to talk to, and I'd be very interested to hear their war stories.

John
Old 23rd May 2019
  #25
Gear Guru
 
Thomas W. Bethe's Avatar
 

Verified Member
Quote:
Originally Posted by Riccardo View Post
What's the name of the second one? (the Pro Tools one you mentioned)
This is what I was referring to but it is only an Encoder not a Decoder

https://www.audiothing.net/effects/type-a/
Old 23rd May 2019
  #26
Gear Head
 

Quote:
Originally Posted by Thomas W. Bethe View Post
This is what I was referring to but it is only an Encoder not a Decoder

https://www.audiothing.net/effects/type-a/
Oh yea -- I knew about that... Very interesting -- I wonder if they actually implemented the DolbyA detector design -- that is the *big secret* that keeps DolbyA from creating distortion worse than it already does. R Dolby was an absolute genius.

Back in the day, I have heard (from reliable sources) that they used a DolbyA for vocal enhancement (like on Karen Carpenter.)

Just putting together a fast FET compressor (or direct software equivalent) without the 'special touch' -- or something equivalent -- in the DolbyA would not be a pretty thing. So, using the precise design in the DolbyA would be a good start for a vocal enhancer -- and maybe do a few things different to optimize. An A301 would probably be great, because maybe disabling some of the bands might be a good thing -- it seems like 20-80 and 9-20k bands might hurt more than help?

My 'decoder' cannot encode, so I cant test the hypothesis, but it would seem that compression in the 3k-20k band (or 3k-9k band) would be sufficient. The 9k-20k seems like it MIGHT be undesirable? Oh well, I am sure that the guys/gals selling the product have made good decisions.

The 'encoding' side does interest me -- but I have a personal (emotional) reason for NOT actually doing an encoder. (The old tapes should be recovered, but seldom should more be created.)

John
Old 25th May 2019
  #27
Gear Guru
 
Thomas W. Bethe's Avatar
 

Verified Member
Quote:
Originally Posted by John Dyson View Post
Satin isn't nearly as accurate... We did some comparisons -- doesn't really sound the same as a DolbyA -- but it is all what people like/perceives themselves. (Like my somewhat 'earthy' grandma used to say -- whatever blows your skirt up.)

Sounding good doesn't count IMO -- sounding JUST LIKE a DolbyA sans distortion is important. Refer to the demos. The most recent Cars comparisons (compared with a 192k/24bit professionally done CORRECT release, vs a leaked CD being decoded by DHNRDS -- they sound literally almost the same -- the DHNRDS doesn't have the veil of distortion.)

John
I know in the "good old days" certain Dolby A decoding units sounded "different" I guess changes in manufacturing, aging components OR Dolby A encoders that were not properly set up when the original encoding was done. I personally like the way Satin handles Dolby A decoding and also how it handles DBX decoding.

We sold two of our Dolby 361 units to a recording studio, they were going to use the as vocal processors. I guess they liked the Dolby A encoded signal.

FWIW
Old 25th May 2019
  #28
Gear Guru
 
Brent Hahn's Avatar
 

Quote:
Originally Posted by Greg Reierson View Post
Yes, exactly. It's when a reel was either not labeled or misidentified and not properly decoded when creating the master. Some label catalogs are notorious for missing this sort of information.
Yup. I recently heard an artist compilation that had maybe 6 of my mixes on it. They were A-encoded and obviously dubbed without decoding. The shriekiness aside, the compression aspect sounded surprisingly, well, contemporary.
Old 25th May 2019
  #29
Gear Guru
 
Brent Hahn's Avatar
 

Quote:
Originally Posted by Thomas W. Bethe View Post
I know in the "good old days" certain Dolby A decoding units sounded "different" I guess changes in manufacturing, aging components OR Dolby A encoders that were not properly set up when the original encoding was done.
It took quite a bit of time to do a careful setup on a 24-track. And then, even if you did, you might find yourself "de-calibrating" to deal with some weird issue. Like when the Dolbys would treat bow rosin or snare brushes as "hiss."

Last edited by Brent Hahn; 25th May 2019 at 05:24 PM..
Old 26th May 2019
  #30
Lives for gear
 
Adam Dempsey's Avatar
 

Verified Member
"Never put Dolby on a snake detector." – Bill Bailey.
Topic:
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Forum Jump
Forum Jump