|
Post by johndyson on Dec 23, 2019 21:34:04 GMT 10
Thanks for the feedback. There is a new 'Crime of the Century.' I took some feedback seriously -- there was one flaw -- not really obvious, because of the difficulty in finding references. One example that someone showed me is indeed a DolbyA decoded, digital copy. The DolbyA digital copy has the 'DolbyA fog', so has a smoother sound. However, there is another mechanism in the DHNRDS DA that can falsefy detail -- it is a numbers mistake by me. Basically, there was too much expansion in the 74-2788 frequency range (commonly called 80-3k or the MF/midrange.) This excess expansion was slight, about 0.5dB over the 10dB range. It was just enough to be noticeable on 'Rudy'. Also, there has been some dynamics and further fog reduction, mostly noticeable on 'Breakfast' for example. I don't know if the dynamics improvement/fog reduction is noticeable on Crime -- it is deeper than my ability perceive on 'Crime'. So, here is a better decode of 'Crime' 88.2/24 (I am listening to it right now): www.dropbox.com/sh/lz3pjtqii0ktef1/AAB7Otv0HCmVgzbTekSOmlEPa?dl=0Also, a decode from the Linda Rostadt 1977 album... Note that it comes from a feral-DolbyA CD and not a master tape. THere is some manipulation on it, but it is definitely WIDE dynamic range. I didn't touch this other than what was necessary to decode, and a reduction to 44.1/16 (sorry) because of space limitations, the material is less pristine, but nice anyway: www.dropbox.com/sh/5z6xl6ig0wpp5tf/AAAxhcqNbnqhR04tAcX0Tl22a?dl=0
|
|
|
Post by johndyson on Dec 24, 2019 5:16:00 GMT 10
Wow -- I found the magic sauce to make Sergio Mendes and Brasil'66 sound pretty darned good. I have been looking for the magic sauce for the last several years. It sounds very interesting compared to the feral DolbyA CD. I plan to get the decoding done this afternoon. It WILL be worthwhile to listen!!! (I can hear the material in a 'preview' mode with about 90% of the quality, but I like to send better copies to friends.)
John
|
|
|
Post by johndyson on Dec 24, 2019 11:31:08 GMT 10
|
|
|
Post by ROWUK on Dec 24, 2019 18:58:45 GMT 10
John, I have a question. Are you going to make this a tool that can be purchased by restoration facilities, hi-res retailers or even major labels? Do you plan to offer this service commercially? Or is this a just a pet project based on a hobby? At what time would royalties become an issue? Or is so much different after conversion, that it is yours?
The reason that I ask is that it has become apparent that undoing crimes of the century (undocumented electronic manipulation) needs respect and ears that YOU have, but those that would normally use the tool got us into this mess in the first place. I question if they would have the common sense, desire, capability of discriminative listening and respect for the artist. Would major studios need to retain “golden ears” to get it right? How many golden ears are even out there? Did the studios not have Dolby decoding available and did they decide to not decode to make playback on cheap systems more spectacular?
In any case, I feel lucky and honored to be part of this at this stage. Your decodes have changed my listening habits for these classics. Before, it was just fun. Now I can stare at my speakers for an hour and soak up details and have fun at the same time.
I do not search for audiophile artifacts. The pleasure comes from freedom of unnatural distraction. It would be interesting if we could get someone with a real original master tape for comparison...
|
|
|
Post by johndyson on Dec 24, 2019 22:35:51 GMT 10
My comments below are unfiltred and raw... I am not accusing or intending to hurt anyone. My motives are altristic, and I have never expected to make money on the project. So, the cost/price of the decoder is very low (a few $100 for a perpetual license, free for a 1,2,3 month trial version -- willing to add months to consumers for free.)
1) My decodes are only shared amongst 'friends'. I try to limit to a few people -- I watch the numbers reading the forums/etc. I ask that people already have copies or purchase copies of the recordings to keep things straight. The copies should not be distributed widely, and if I felt that I would impact sales negatively, that would be contrary to my goals. My interest is in improveing sales and encouraging better quality. If I find too many people 'stealing' instead of 'doing the right thing', I would pull back very aggressively. 1a) I love people to share the results to possibly wake consumers up -- push the high res and other distributors to produce better quality stuff. 1b) Even though the DHNRDS can do a very superior job over and above the DolbyA HW, it is still mostly better for material to be decoded with real DolbyA than this feral-junk being sold. 2) The DHNRDS is for sale through Richard Hess. He and I give out timed-out versions for free. Lookup our very primitive web site DHNRDS.com, or contact Richard if you already know him. The project is a little depressed because of the general low standards of the distributors and the consumer acceptance of feral-DolbyA. (That is, DolbyA material which is EQed, then sold to consumers as 'product'.) 3) The DHNRDS is very difficult to use with 'feral-DolbyA' material -- there is a total of one person in the world who has the patience to do it (me.) 4) The DHNRDS is very easy to use with DolbyA master tapes, and perfect results will result from working directly with pristine master tapes. 5) Per my discussions with Richard and other people in the recording industry - it appears that there would be a terrible mess and poor profits to start selling the material on true commercial basis.
And finally -- I plan to produce a consumer verison of the DHNRDS which is easier to use for decoding the feral-DolbyA material. It will never be as simple as decoding with master tapes, but it can be made to be much easier to use. I can train on an over-the-net one on one basis to get someone started decoding feral material now, but with a few improvents, a very advanced audiophile will be able to do decodes themselves (with a little initial help.) The damned feral-EQ usually has a similar shape on most recordings, but some are very different and some expected to be decoded in an M+S matrix, but input and output from the process are L+R... The decoder has already been improved to do the M+S matrix internally, but the feral-EQ correction filters still need to be added... So, this is most of the story, without any censoring -- telling you as much of the truth as I can (the dirty side as well as the good things!!!) John .
|
|
|
Post by johndyson on Dec 25, 2019 7:56:15 GMT 10
About 'Breakfast' -- I am always chasing perfection. Found a possible improvement -- will be done before Christmas morning USA EST time. If I don't upload it by then, then it will be as-is. However, I think that I found a minor nit.
John
|
|
|
Post by johndyson on Dec 25, 2019 17:13:02 GMT 10
A new 'Breakfast' is ready. I fixed the slithering snake sound (it is kind of cool in a way... but was wrong.) This now sounds correct. I felt there was something wrong, but took mew a day or so to figure it out. This decode (and future) will result from the new DA decoder as I described after the link to the new Breakfast: www.dropbox.com/sh/tgrzo0b2zguon2n/AADghinVq4r17TfZnGSc6bSna?dl=0-------------------------------------------------------------------- Here is an idea of the improvements: Notes for V1.0.0I Improvements: Better selection of input hilbert filters. Changing to the Hann window on the input hilbert filters. Kaiser-Bessel doens't help the input filters. Moving to Hann smooths the noise in the math precision. Select double precision math for only the longest/highest quality mode. Double precision is less helpful for short Hann windowed filters. So, there was a bit of tweaking optimization in the input Hilbert anti-IMD detectors, but no changes to the anti-MD gain multiplication code. Lower quality modes should have less IMD than previous versions. Very careful adjustment so that the 0dB response is more flat, and there is no longer a general curve downwards to about -0.50dB at 20kHz. The MF band expansion was excessive -- caused some really audible problems in 'Rudy' from Crime of the Century/Supertramp. Slightly faster decoding in middle/high quality modes (using normally documented modes --xtra, --xtraplus and --xpp. The super-normal xpp=7 or higher are faster until the --xpp=11 and --xpp=12, which are 'better', but not faster. Design tuning: It is very important for the decoder to have an attack time that mimicks the DolbyA, but only as fast as it can be before creating splats. It is a bit of artfulness to optimize the fastest attack times. There is a careful mix of RMS and average filters to emulate the DolbyA behavior with maximum faithfulness and quality. Quality and faithfulness to the sound is more important than emulating the DolbyA bug for bug. Previous versions (before the slippery version) had slightly too harsh attack... Instead of depending on a simple time constant, there is a combination of time constants intended to provide the fast attack up to a certain point, then a fixed approx 9kHz filter -- this attack signal is synchronized at 1/2 the time of the approx 1kHz RMS filter array (specially constructed for total linear phase)... The resulting output of the filters is zeroed to exactly the sample point for the DolbyA compatible detector. By using this sophisticated filter array, there are no operative nonlinear effects in the detection -- unlike normal diode type detectors, and therefore no IMD in the output down to a certain frequency. Each mode has progressively more resistance to lower and lower frequency IMD. (An interim version used a super highly optimized attack speed that minimizes distortion, but was so controlled that it didn't sound correct. A rescinded copy of Breakfast In America example was produced with the 'overly silky smooth' sound. It was a mistake. This slimey version didn't have the 9kHz filter portion of the Hilbert detector filter -- which caused sound to be 'too smooth'. An earlier verison of the 9kHz portion of the filter input was too harsh -- this one is linear and tweaked to match exactly what the DolbyA compatible attack/release calculation wants to see.) Usage optimization: Create a mode 12 that maximizes all filtering for highest quality. It eliminates the need to specify an additional input-filter mode for maximum distortion control. Add an --ms mode for mid-side decodes without needing to use input/output sox scripts for the conversions to/from mid-side matrix as needed on some Carpenters, Brasil'66 and a few others.
|
|
sandyk (RIP Alex, 1939 - 2021)
Global Moderator
Posts: 226
About Me: Retired ex Principal Telecommunications Technical Officer with 43 years at Telstra (Australia)
I am also a Moderator in Hi Fi Critic Forum
Electronics hobbyist for >65 years with DIY projects including Loudspeakers, Stereo FM tuner, S/W Regen Receiver, Superhet AM ,
Synchrodyne PLL AM tuner (Phase Lock Loop),Stereo Tape Deck, Amplifiers including I.C. types, Class A, Class AB 100W/Ch. (ETI5000) 240W/Ch. Mosfet (AEM6000) ,several DACs , numerous PSUs including VERY low noise (<4uV) types etc.for myself and friends
Audio Industry Affiliation: NIL
|
Post by sandyk (RIP Alex, 1939 - 2021) on Dec 25, 2019 17:48:46 GMT 10
A new 'Breakfast' is ready. I fixed the slithering snake sound (it is kind of cool in a way... but was wrong.) This now sounds correct. I felt there was something wrong, but took mew a day or so to figure it out. This decode (and future) will result from the new DA decoder as I described after the link to the new Breakfast: www.dropbox.com/sh/tgrzo0b2zguon2n/AADghinVq4r17TfZnGSc6bSna?dl=0 Many thanks John I have just started downloading it. Enjoy Xmas with your family, and try to relax for a little while if your brain will let you .
Best Wishes Alex
|
|
|
Post by cj66 on Dec 25, 2019 20:14:14 GMT 10
Hi John,
I have DL'd Crime... (and The Logical Song to see what you mean by your comments) but not sure when I'll get a chance for a proper listen, in the meantime have a great Christmas and New year.
|
|
|
Post by johndyson on Dec 25, 2019 20:40:46 GMT 10
Enjoy listening, and Merry Christmas again. Got more 'stuff' coming!!! It is so hard to choose the *best*. I LOVE some material, but it misses the mark. The newest decode for The Best of Brasil'66 is better yet (there are SO many combinations of feral DolbyA EQ. Even though the previous Herb Alpert also Brasil'66 albums were okay, the newere versions FINALLY have the proper ambiance -- real pain in the b*tt sometimes.) I haven't uploaded these new ones, but got more 'better' stuff also.
The Nat King Cole recordings are really good -- but unfortunately, but the time peoople get their own copy of the Nat King Cole Story, it has that evil feral DolbyA compression. I wonder if it is the distributor or the mastering making the bad choice? Actually, might not be a 'bad' choice because a lot of listeners are now used to the compression... I know I am tilting at wind mills.... I'll probably make Nat King Cole available in the future, but if someone is really interested it can be sooner rather than later.
No matter, I really hope that people enjoy these results -- and SOME DAY, maybe will be able to produce a consumer usable decoder for feral material. If everyone had true DolbyA masters, the DA decoder would be so easy to use, but this messed up stuff is really tricky.
John
|
|
sandyk (RIP Alex, 1939 - 2021)
Global Moderator
Posts: 226
About Me: Retired ex Principal Telecommunications Technical Officer with 43 years at Telstra (Australia)
I am also a Moderator in Hi Fi Critic Forum
Electronics hobbyist for >65 years with DIY projects including Loudspeakers, Stereo FM tuner, S/W Regen Receiver, Superhet AM ,
Synchrodyne PLL AM tuner (Phase Lock Loop),Stereo Tape Deck, Amplifiers including I.C. types, Class A, Class AB 100W/Ch. (ETI5000) 240W/Ch. Mosfet (AEM6000) ,several DACs , numerous PSUs including VERY low noise (<4uV) types etc.for myself and friends
Audio Industry Affiliation: NIL
|
Post by sandyk (RIP Alex, 1939 - 2021) on Dec 25, 2019 21:17:21 GMT 10
Hi John,
I have DL'd Crime... (and The Logical Song to see what you mean by your comments) but not sure when I'll get a chance for a proper listen, in the meantime have a great Christmas and New year.
Hi Chris I haven't had a chance to listen to the whole BIA album, but the Logical Song is markedly better than the previous in all respects Alex
|
|
|
Post by johndyson on Dec 28, 2019 3:42:12 GMT 10
OMG -- I was just doing a code review to figure out what needed to be done to morph a copy of the DA decoder into the Telcom C4 emulator. I found some SERIOUS bugs in the DA decoder... About 1/2 of the anti-MD code was actually functioning and the LF and lower MF bands (that is 0-74, and 74-850Hz) didn't have the anti-MD distortion cancellation enabled AT ALL. This would explain the difficulty in getting good, clean decodes on certain material. Of course, you know what this means to the audiophile world? Yep -- new copy of 'Crime of the Century' and 'Breakfast in America'. I am listening to 'Crime' (01 School in particular), and it is really coming out clean/smooth.
Isn't it amazing? I have been having some brain chemistry problems, which makes me incredibly mistake prone -- I just get irritated and frustrated while trying to concentrate more intensively. It is these kinds of things that happen when I try so very hard, knowing what needs to be done, but I forget in the middle of programming. There was literally 1/2 of the anti-MD code with an option that I forgot to enable... I probably lost concentration for a second, and then the mistake ensued. THis is SOOO frustrating, but these incremental presents are like new Christmas presents every few weeks? :-).
The Crime/Breakfast redecodes are my highest priority -- will probably have the final reesults tomorrow, but they are really beautiful.
Alex came up with a DolbyA decoded comparison of some Carpenters stuff that I decoded. They did some serious post decoding mastering on the material (lots of EQ), but additionally, you can hear the advantages and disadvantages of the DHNRDS DA. It is relentless about grabbing detail and avoiding the DolbyA fog. Problem: what happens when material has been mixed where it has been optimized for the DolbyA fog? Answer: the DHNRDS sounds too intense. Instruments that sound a little compressed on the true DolbyA sound bright and very clean on the DHNRDS. Sibilance is also suppressed by the effective HF compression when decoding with the DolbyA.
Actually, for some recordings, it appears that using the DolbyA (or disabiling all of the anti-MD and anti-IMD on the DHNRDS) are better than running the DHNRDS in fully quality mode. The stuff was mastered when expecting that there would be the DolbyA fog, and they appeared to compensate for it!!!
For really clean material (e.g. Crime, and most other good material), the DHNRDS in full quality mode is probably nicer though...
John
|
|
|
Post by johndyson on Dec 28, 2019 9:55:01 GMT 10
Here is a potentially interesting comparison: DolbyA decoding, vs DHNRDS DA in full quality mode vs. DHNRDS DA with very reduced quality. Note that the original purpose of the DHNRDS DA was NOT to be better than DolbyA, but I decided that it must be!!! DolbyA decoding -- vinylrip DHNRDS full quality -- DHNRDS fullantiMD DHNRDS no anti-modulation distortion -- DHNRDS noantiMD The big difference between the three, other than the slight 'compression' & EQ difference in the DolbyA decoded version, and the graininess in the DolbyA and noantiMD verison, the full antiMD will have much less grain or grind in the sound.. The anti-MD DOES mitigate distortion, might be subtle at first. After a while, you can hear the intermodulation in all but the full anti MD versions... FULL QUALITY DHNRDS DA: www.dropbox.com/s/8wm8y839knw6z7j/08. ABBA - Me And Bobby And Bobby's Brother-DHNRDS-fullantiMD.flac?dl=0 NO ANTI-MD DHNRDS DA (but some distortion management): www.dropbox.com/s/dsqwyl76jttv6q3/08. ABBA - Me And Bobby And Bobby's Brother-DHNRDS-noantiMD.flac?dl=0 VINYL RIP/DOLBYA decoded www.dropbox.com/s/b4mz00iacil8u3v/08. Me And Bobby And Bobby's Brother-vinylrip.flac?dl=0 John
|
|
sandyk (RIP Alex, 1939 - 2021)
Global Moderator
Posts: 226
About Me: Retired ex Principal Telecommunications Technical Officer with 43 years at Telstra (Australia)
I am also a Moderator in Hi Fi Critic Forum
Electronics hobbyist for >65 years with DIY projects including Loudspeakers, Stereo FM tuner, S/W Regen Receiver, Superhet AM ,
Synchrodyne PLL AM tuner (Phase Lock Loop),Stereo Tape Deck, Amplifiers including I.C. types, Class A, Class AB 100W/Ch. (ETI5000) 240W/Ch. Mosfet (AEM6000) ,several DACs , numerous PSUs including VERY low noise (<4uV) types etc.for myself and friends
Audio Industry Affiliation: NIL
|
Post by sandyk (RIP Alex, 1939 - 2021) on Dec 28, 2019 10:48:24 GMT 10
Here is a potentially interesting comparison: DolbyA decoding, vs DHNRDS DA in full quality mode vs. DHNRDS DA with very reduced quality. Note that the original purpose of the DHNRDS DA was NOT to be better than DolbyA, but I decided that it must be!!! DolbyA decoding -- vinylrip DHNRDS full quality -- DHNRDS fullantiMD DHNRDS no anti-modulation distortion -- DHNRDS noantiMD The big difference between the three, other than the slight 'compression' & EQ difference in the DolbyA decoded version, and the graininess in the DolbyA and noantiMD verison, the full antiMD will have much less grain or grind in the sound.. The anti-MD DOES mitigate distortion, might be subtle at first. After a while, you can hear the intermodulation in all but the full anti MD versions... FULL QUALITY DHNRDS DA: www.dropbox.com/s/8wm8y839knw6z7j/08. ABBA - Me And Bobby And Bobby's Brother-DHNRDS-fullantiMD.flac?dl=0 NO ANTI-MD DHNRDS DA (but some distortion management): www.dropbox.com/s/dsqwyl76jttv6q3/08. ABBA - Me And Bobby And Bobby's Brother-DHNRDS-noantiMD.flac?dl=0 VINYL RIP/DOLBYA decoded www.dropbox.com/s/b4mz00iacil8u3v/08. Me And Bobby And Bobby's Brother-vinylrip.flac?dl=0 John Agreed that the full Anti MD mode sounds better than even the Vinyl rip
|
|
sandyk (RIP Alex, 1939 - 2021)
Global Moderator
Posts: 226
About Me: Retired ex Principal Telecommunications Technical Officer with 43 years at Telstra (Australia)
I am also a Moderator in Hi Fi Critic Forum
Electronics hobbyist for >65 years with DIY projects including Loudspeakers, Stereo FM tuner, S/W Regen Receiver, Superhet AM ,
Synchrodyne PLL AM tuner (Phase Lock Loop),Stereo Tape Deck, Amplifiers including I.C. types, Class A, Class AB 100W/Ch. (ETI5000) 240W/Ch. Mosfet (AEM6000) ,several DACs , numerous PSUs including VERY low noise (<4uV) types etc.for myself and friends
Audio Industry Affiliation: NIL
|
Post by sandyk (RIP Alex, 1939 - 2021) on Dec 29, 2019 8:06:18 GMT 10
|
|