John Dyson Posted December 4, 2019 Author Share Posted December 4, 2019 Here is a perfect example of DolbyA decoding being needed -- for its original purpose, and not just to correct a screw-up. Note that I have determined that the Audio Productions 2011 version of Nat King Cole Story is definitelly leaked DolbyA, and it uses the typical EQ to make it sound 'tolerable'. There are two versions of the file -- one is 'decoded', the other is 'undecoded'. You can clearly hear the hiss in the undecoded version -- which is the disk as-is. The decoded version also hasa bit of noticeable hiss, probably because the original recording did have a moderate amount of hiss itself. The original recording didn't use DolbyA, but the re-done master or the archive tape obviously did. This kind of thing is why they used DolbyA to begin with -- it was NEVER intended as an 'effects' device or some kind of corrupting 'distorter' as it seems currently to be used :-). (Whether or not it had been used as an effects device, it wasn't really intended to do so.) So, this is a snippet of 'Mona Lisa', decoded & undecoded (flac for details, mp3 for quick review.) https://www.dropbox.com/sh/622qbxxtguwyzaw/AABTDSec_JLdjdNrXryZwH9Ea?dl=0 Link to comment
John Dyson Posted December 9, 2019 Author Share Posted December 9, 2019 During some recent decoding attempts, I decided to make a change to the DHNRDS -- and the results are frightening and astounding both. Back in the olden days, when DolbyA was used for noise reduction instead of a transport/obfuscation and sound effects mechanism, recording engineers with acute hearing would often complain that DolbyA has a 'fog', that the music ends up sounding less detailed. Spurred on by circumstance and my project partner's idea, I taught the DHNRDS DA to ALMOST look back in time, and to undo some of the distortion created by the ENCODING process. Not only is the decoding distortion mostly avoided & cancelled, the encoding distortion is now at least partially cancelled. This code is almost a time machine -- I can hear the suppression of modulation distortion when compared with the original DolbyA -- not just with other decodes!!! The suppressoin cannot be complete, but is very beneficial. I have only decoded a few albums so far, other than some snippets. When I make more progress in decoding, and having material other than ABBA and Supertramp, I'll probably make some extra-long snippets available. These results are worth pushing things a little bit. The decoder has turned into a 'time machine'. Luckily it now knows enough about the expected state of the encoding device to undo some of the original damage to the signal!!! This benefit should extend to ALL feral material, eliminating any disadvantage to decoding. John Link to comment
Popular Post Samuel T Cogley Posted December 10, 2019 Popular Post Share Posted December 10, 2019 On 12/9/2019 at 4:49 AM, John Dyson said: During some recent decoding attempts, I decided to make a change to the DHNRDS -- and the results are frightening and astounding both. Back in the olden days, when DolbyA was used for noise reduction instead of a transport/obfuscation and sound effects mechanism, recording engineers with acute hearing would often complain that DolbyA has a 'fog', that the music ends up sounding less detailed. Spurred on by circumstance and my project partner's idea, I taught the DHNRDS DA to ALMOST look back in time, and to undo some of the distortion created by the ENCODING process. Not only is the decoding distortion mostly avoided & cancelled, the encoding distortion is now at least partially cancelled. This code is almost a time machine -- I can hear the suppression of modulation distortion when compared with the original DolbyA -- not just with other decodes!!! The suppressoin cannot be complete, but is very beneficial. I have only decoded a few albums so far, other than some snippets. When I make more progress in decoding, and having material other than ABBA and Supertramp, I'll probably make some extra-long snippets available. These results are worth pushing things a little bit. The decoder has turned into a 'time machine'. Luckily it now knows enough about the expected state of the encoding device to undo some of the original damage to the signal!!! This benefit should extend to ALL feral material, eliminating any disadvantage to decoding. John John, I know others have said this, but I just want to say again that I really, really appreciate your efforts with DolbyA. We don't deserve you. 👍 sandyk and rando 1 1 Link to comment
Popular Post John Dyson Posted December 10, 2019 Author Popular Post Share Posted December 10, 2019 19 minutes ago, Samuel T Cogley said: John, I know others have said this, but I just want to say again that I really, really appreciate your efforts with DolbyA. We don't deserve you. 👍 Thanks -- look, I am nothing special, really. Just got a crazy hobby... I am kind of on a 'slowdown' for a week or so. Believe it or not, this recent improvement in the DA decoder has upset me a bit. I didn't realize that it could be improved!!! Just need to let off some steam for now... I'll be back in full force in a few days... --------------- BTW -- I think that I figured out how the feral-DolbyA is leaking nowadays. I know that it doesn't leak all the time, but there are still new releases that aren't really decoded (some of the high res stuff.) The recording people can feel 'good' about the decoding, except they must not have checked the results very completely. This is probably one reason why my Nat King Cole from Audio productions is feral DolbyA.... There are so-called 'decoder' plug-ins that don't really decode. I was confounded by one, early on, when presented to me by my project partner. He said that he didn't like it, and when I reviewed it -- noticed that it didn't really do noise reduction... Also, the frequency response balance was 'plausible', but not quite correct... It produced results that couldn't compare favorably with the true pre-DolbyA source. Answer: I suspect -- but cannot prove -- that some of the so called 'DolbyA decoders' are doing the feral EQ compensation. So, basically they take the terrible frequency characteristics of true DolbyA, then convert it to 'feral-DolbyA'... Then, the recording engineer or mastering person can feel comfortable that they have 'decoded' the material, and they don't have to go to the bother of a real-time decode. (An actual DolbyA decode with the actual HW takes the same amount of time as playing the tape.) The 'feral-DolbyA' EQ compensation is as quick to do as a few 2nd order filters!!! Big difference, eh? Unfortunately, a significant qualitative deviation from the artists intent of the sound. * What also confounded me by the 'plug-in' decoder is that a true decoder takes LOTS of work, and no-one would do an accurate DolbyA decoder commercally because of a lack of market... Either the plug-in writer knew something that I didnt know about math and the complexity of a DolbyA, or something was rotten somewhere. I'd suspect that there is certainly something that was 'rotten', and that 'decoder' should be called an 'equalizer' instead. John Samuel T Cogley, sandyk and Teresa 1 1 1 Link to comment
Popular Post John Dyson Posted December 11, 2019 Author Popular Post Share Posted December 11, 2019 Just wanted to give a status report. I am working on the DA at a very slow, methodical rate. Essentially, the goal is to be very very safe while making changes. Any significant change requires very careful consideration. Doing such things in a hurry will always leave mistakes or errors. I am taking it very easy and not worrying about getting answers right away. There are two kinds of 'solutions' to problems -- 'how it is calculated', and 'the actual solution'. Because there are so many complex factors, it is impossible to calculate the exact answers to what is needed in the decoder. First, there is a theoretical -- usually fairly simple -- bit of math. That math often excludes lots of information because it is too complicated to develop a precise answer. Also, importantly, there is NO specification to work with. Next, a few intuitive adjustments -- usually using bisection, successive approximation until the distortion nulls. This is where it is very important to have very perceptive hearing. Without a certain mix of skills, any non-trivial modifying the decoder would not produce good results. This makes the decoder impossible to develop without an iterative procedure and knowing ones own limitations. Frustratingly -- 'sounds good' is not the goal. Tuning the decoder for any one sort of material does not result in reasonable results. Tuning for the best accuracy is the only goal, and luckily there are some resources for finding the best accuracy... There are very many ways for the decoder to be inaccurate. Amazing that only a 10 to 15dB gain range can cause so much distortion... But, think of it like this -- take a signal, and flip quickly between even 0dB and -6dB of gain, or modulate the gain of that signal by 50% with any tone or signal... That can make a huge amount of difference (-6dB of gain is like AM at 50% modulation level.) Gain changes at the wrong time are tantamount to distortion. All DolbyA HW, while decoding, couldn't really do the gain changes at the correct time, and it is a real pain to make 100% correct!!! However, the DolbyA HW, while encoding, is by definition -- correct. This is one reason why I would not do a 'DolbyA encoder', because there is great responsibility in creating the 'famiily jewels', and frankly, DolbyA shouldn't really be used to encode for NR purposes anymore. (There are other purposes, but not dealing with that.) So, the reason for my short disappearance/slow-down isn't just an emotional thing. It comes from a recognition that distractions are not good right now. Hurrying will only result in poorer quality. Plus -- it is nice to relax a little!!! John lucretius, rando and sandyk 3 Link to comment
John Dyson Posted December 13, 2019 Author Share Posted December 13, 2019 I am still 'disappeared', but came up with a few breakthroughs -- one was that the DA was being a little too aggressive in trying to mitigate MD, so created a bit of an indelicate sound... I am modifying the behavior to be more agile, yet still sound good. One problem was Linda R's (Youre no good), where the chorus was 'chunky', almost distorted sounding. I don't have the correct balance yet -- but it is definitely a tweaking game to mitigate MD, but also to make sure that the DA can track the signal very closely. I ran my test of a 3 times encoded copy of Dreamworld, and one version of test code where I tweaked up the anti-MD algorithms -- the result was ludicrous, but almost tolerable. That tells me that with the more weak filtering was even too strong (the 'Youre no good' problem.) The attack/release coefficients for the nonlinear attack/release code for BOTH attack/release layers is tedious, and I think that I have it tuned to work on difficult cases... The attack/release portion of the program is one of those things that should not be modified for 'effect', and must be dead-on correct. (The attack/release is not a time constant, but rather an exponential relationship that mimicks the conductivity of a diode.) There is a problem of tracking the needed attack/release to make the music CORRECT, but also making sure that during those fast gain skews, that distortion isn't created. The problem with the old DolbyA is that both the attack/release had a delay factor in it when decoding (fixed), and also the attack/release speed was poorly controlled (still a work in progress.) If you let the attack/release go nuts, then you get those foggy effects... Strange - but intuitively like the frequency/time resolution tradeoff. I'll be back in a few datys. This requires full concentration and deliberation, not just tweaking right now. This is way too complex to come up with quick answers. In the worst case, I have all approx 500 previous versions of the DA!!!! John Link to comment
John Dyson Posted December 16, 2019 Author Share Posted December 16, 2019 Got an interesting example of a destroyed stereo image by material not being properly mastered... This is the song 'sussudio' by Phill Collins. At first, both versions aren't too awfully bad... Alll I ask is that you listen to the decoded version first... Notice the nice organization of the instrumentation -- fairly orderly music with some nice effects.... * One warning, it is very likely that properly mastered digital versions of this material haven't been available until... ME!!! (some humor intended...) Now -- listen to the CD version... At first, the compression doesn't suck too bad -- it is a little more bright and sparkly sounding... All of a sudden -- continue after the mostly bass intro... Splat -- you got stuff moving all over the sound field -- what, where, which? Reflect on the sound -- the relative lushness of the properly decoded (not even mastered yet) material, and compare with the confusing morass on the undecoded version. Since I have been able to determine more accurate corrective EQ, it is becoming MUCH MORE clear the loss of good stereo image, sometimes total inability to hear anything coherent -- ALMOST LIKE A US TV COMMERCIAL type sound... This kind of thing is one reason why I can listen to ABBA, and it actually sounds like music, not so much like a TV commercial -- BLAST type of sound. Here is the material -- the DECODED version (listen first): https://www.dropbox.com/s/d7vxbrw64i8inxy/sussudio-decoded.mp3?dl=0 the UNDECODED, CD version (after listening to this, then go back and forth a few times.): https://www.dropbox.com/s/y66gtlakycmckml/sussudio-cd.mp3?dl=0 Very interestingly -- the DECODED and UNDECODED versions are directly related to each other and can be somewhat transparently converted between each other. Why would one want to listen to the UNDECODED version on a good system? On the other hand, it does appear that the dynamic range compression on the UNDECODED version would be helpful in a car or on a system that cannot really reproduce the quality of the DECODED version. Note that these are only MP3s. With the new decoder, some results are starting to be amazingly impressive. I can hear a full, very stable stereo image on ABBA now. Since there is so little interest in ABBA, I haven't been making some of the suprising stuff available any more. Frankly, even after all of the iterations, I am amazed by the quality of some of their albums. It just appears that really strange EQ has been used on some material, thereby making it difficult to decode. Other material has been fairly easy to process all along (except for the stereo imaging and lushness problems that I had with the Carpenters, findign out that some of their albums were intended to be decoded in an Mid Side config instead of Left Right -- strange, but a method to their madness.) Anyway -- I was really taken aback by the 'distorted sound' being so predominant nowadays. This nonsense is even happening on premium stuff nowadays!!! John Samuel T Cogley 1 Link to comment
Samuel T Cogley Posted December 17, 2019 Share Posted December 17, 2019 Hi @John Dyson I listened to both a few times. For my listening tests, I raised the level of the "decoded" file by 2.5dB to match the other. I've always looked at 1985 as the year that digital production in the studio eclipsed analog production. I rarely purchase used vinyl made after 1984 for that reason. It's likely digitally sourced with under-performing DACs feeding the cutting lathe (or making the tape that feeds the lathe). I see what you mean about the instruments being much more distinct in the decoded version. There is a synthesizer line in that song I was never fully aware of until today 👍 Just curious, how do you determine provenance of a recording (e.g., analog vs. digital production)? Is it just by listening? Link to comment
Popular Post John Dyson Posted December 17, 2019 Author Popular Post Share Posted December 17, 2019 1 hour ago, Samuel T Cogley said: Hi @John Dyson I listened to both a few times. For my listening tests, I raised the level of the "decoded" file by 2.5dB to match the other. I've always looked at 1985 as the year that digital production in the studio eclipsed analog production. I rarely purchase used vinyl made after 1984 for that reason. It's likely digitally sourced with under-performing DACs feeding the cutting lathe (or making the tape that feeds the lathe). I see what you mean about the instruments being much more distinct in the decoded version. There is a synthesizer line in that song I was never fully aware of until today 👍 Just curious, how do you determine provenance of a recording (e.g., analog vs. digital production)? Is it just by listening? About provenance -- analog vs digital... I don't know at all. I can guess as there was a noticeable increase in quality starting in the middle 1980s, but it could patially have been digital, better tape, etc. Also, DolbySR started appearing about the middle 1980s, and it is super effective NR, but at the cost of even more modulation sidebands... DolbySR is very effective, but IMO a step backward. It is basically making further deals with the devil. DolbyA was bad enough. The only thing that I have become fairly reliable at guessing is whether it is DolbyA encoded or not. There is A LOT of digital material that FOR SOME REASON is DolbyA encoded... I'd suspect that the reason for DolbyA encoding is that of tradition or existing processes, not so much an advantage of noise reduction. There is a LOT of material on CD that IS DolbyA encoded, yet the equivalent vinyl was not sold in DolbyA form. Then, there is the travesty of DolbyA encoded vinyl -- but really, it could be an advantage if the vinyl is subsequently DECODED by the user!!! DolbyA is very effective at reducing rumble.... It helps by about 10dB or so. The really, really tricky part about this DolbyA thing isn't so much deciding dolbyA or not, because the ultimate test is to try to decode the material -- DolbyA sucks badly when trying to decode material which is not in a DolbyA encoded state. The tricky thing is to undo the EQ that the distributor used to HIDE the DolbyA sound. Native DolbyA sound makes the shrillness of undecoded DolbyB or DolbyC sound like the true sound of a mellow Carpenters recording... DolbyA without decoding and not being obscured by EQ is fairly intense. That EQ (or I named -- 'corrective EQ') to recover the DolbyA character is tricky as hell. One can be far, far in error, yet sound almost correct. I have been trying to decode feral DolbyA material for the last 2yrs, and finally believe that I can find the corrective EQ for anything thrown at me. It isn't a 'skill' as much as finding the traditions and techniques used to hide that notoriously shrill DolbyA sound character. Some day soon, maybe after I make a bona-fide effort to get the C4 decodeer working, I intend to look back at the DHNRDS DA decoder and add the consumer features which would make a direct .wav to .wav conversion practical. That is, the corrective filter algorithms will be available as several modes, and several optional submodes. One item of note -- the 'calibration' level, that is the match of the DolbyA compressor/expander curve appears to be relatively stable on the early consumer CDs. it is almost always the magic number of -12.67dB below a certain tape reference level -- which appears to be maintained as the levels on the CDs themselves!!! To me, it is astounding, but probably true at least 50% of the consumer DolbyA recordings... Weird, huh? Oh well, anytime you have ideas -- let me know. I am going to continue finding interesting things about this matter... Also, further questions are very welcome!!! John rando and Samuel T Cogley 2 Link to comment
John Dyson Posted December 18, 2019 Author Share Posted December 18, 2019 This isn't directly on topic, but is related to the DHNRDS DA decoder itself. Fairly often, I have described the DHNRDS DA decoder as large and somewhat complex -- not mind bendingly complex however. To give an idea as to why the DHNRDS might act differently & slower than a normal DSP or HW gain control device - even any other bona fide DolbyA type SW decoder, here is a description of what is going on inside: ---- Each sample on each channel is run through 3 carefully crafted FIR filter banks, and each filter bank has approx 30 filters to splt the signal into 4 bands. (It is a brute force method to match the true DolbyA band splitting, as normal IIR filters do not adequately emulate the analog filters in the HW itself on 3 of the bands.) The LF band uses a normal fc=74Hz, Q=1.070 low pass IIR filter. The IIR filter does adequately emulate the LF filter because the differences in timing and phase between a HW and IIR filter have a minimal negative effect on that application. Normal HANN window functions were used for any FIR bandpass filters. Each sample has 32 Hilbert transforms applied to the signal, and 2 Hilbert transforms are applied to help calculate the gain control. The number of taps on each Hilbert transform can be between 2048 and 14336, all depending on the reason inside the program and quality mode. For the anti-distortion applications, ripples in the phase curve become problematic, and even the window function is carefully considered. The window function is fairly uncommon and not-well-known -- chosen after a lot of research: Kaiser-Bessel w/alpha=4.5 for the best sideband minimization. The more common HANN window function would be okay, so would the HAMMING function, but KAISER-BESSEL alpha=4.5 is the best that I could find for both the anti-MD and anti-IMD. For each sample for each channel, on each band, there are 3 diode exponential curve calculations emulating 4 diodes (in one case, there are two diodes in series, so the result is a halving of the cnoductance instead of simulating both diodes individually.) After the intricate calculation of the attack/release for each sample, there are 40 linear attack/release shaping filters amongst 40 square root and squaring operations. Each filter and math function operates on each of 8 channels (4 bands of 2 channels each) at a time. In a way, one can say that there are 320 filters and 320 square root and squaring operations PER SAMPLE. The attack/release calculation itself is dependent on internal negative feedback, which is needed to precisely mimick the high level, user visible DolbyA dynamics. For each sample, for each channel, for each band, there is an intricate 8th order polynomial which calculates the gain curve. Because of the number of calculations and for fastest operation, there are over 20 threads comprising the DHNRDS DA, where a each sample is passed amongst over 20 threads. Nine of the threads is a substantial load on the CPU when running from file to file decoding operation. The slowest single thread is the attack/release timing calculation itself, but the slowest total operation are on 8 of the 16 threads that are needed to very accurately calculate the 'newsignal = oldsignal * gain'. Avoiding the creation of modulation distortion requires well over 90% of the CPU usage when decoding a recording. Even the fastest mode, without the complex, very slow modulation distortion calculation, the DHNRDS DA produces a lower level of distortion products than the true DolbyA because the anti-IMD attack/release calculation. Currently the user cannot disable the anti-IMD mode. (The basic internal code base does allow disabling the anti-IMD mode, and if someone ever wishes to write a plug-in for decoding DolbyA material, then I would supply them a version for a very basic DolbyA decode witihout anti-MD and probably without anti-IMD abilites.) The DHNRDS DA can fully utilize an 8 core CPU for speedup, and beyond 12 cores there is a serious diminishing return. The DHRNDS can run on one core, but is slow on only one core. The DA decoder does everything practical to avoid distortion. John rando 1 Link to comment
John Dyson Posted December 22, 2019 Author Share Posted December 22, 2019 There are some new samples coming, and in the PM group there is a new 'Crime of the Century' with criticisms considered in the redo. I am looking at some Phil Collins samples, but not really happy with Carly Simon for samples -- just don't sound enticing. The Carly Simon decode is done. Linda Ronstadt is starting to sound much better though. Been working on Christmas presents -- I am very far behind because of trying to perfect the results!!! More stuff coming soon... John Link to comment
John Dyson Posted December 28, 2019 Author Share Posted December 28, 2019 Sometimes, I write abut the 'anti-modulation-distortion' capabilities of the DHNRDS DA. The terminology probably sounds like gobblty-gook, but the best way to describe is to demostrate. Note that a real DolbyA would sound even grittier, but also leave more of a compressed sound than the DHNRDS DA. I only have ABBA for the a publically available decoded reference -- I gave that demo in some private communications, but this is something that I could do publically. This is from an old Burt Bacharach greatest hits type album set. You might have to listen for the graininess later in the cut -- but the ambiance is also less stable in the no-anti-MD. It has a generally more grainy sound -- yet the FULL anti-MD has a more full sound as the recording engineer might wish. Note that the settings are IDENTICAL, and the only difference is the disabling of the anti-MD and anti-IMD portions of the code. FULL anti-MD: https://www.dropbox.com/s/yiy1yru9r7fk8rt/08. Twenty Four Hours From Tulsa-FULLantiMD.mp3?dl=0 NO anti-MD, but still some distortion reduction: https://www.dropbox.com/s/x9xf41e72qfhhxq/08. Twenty Four Hours From Tulsa-noantiMD.mp3?dl=0 John Link to comment
sandyk Posted December 28, 2019 Share Posted December 28, 2019 41 minutes ago, John Dyson said: FULL anti-MD: https://www.dropbox.com/s/yiy1yru9r7fk8rt/08. Twenty Four Hours From Tulsa-FULLantiMD.mp3?dl=0 NO anti-MD, but still some distortion reduction: https://www.dropbox.com/s/x9xf41e72qfhhxq/08. Twenty Four Hours From Tulsa-noantiMD.mp3?dl=0 John I have no problems hearing the improvement with FULL anti MD How a Digital Audio file sounds, or a Digital Video file looks, is governed to a large extent by the Power Supply area. All that Identical Checksums gives is the possibility of REGENERATING the file to close to that of the original file. PROFILE UPDATED 13-11-2020 Link to comment
John Dyson Posted December 28, 2019 Author Share Posted December 28, 2019 28 minutes ago, sandyk said: I have no problems hearing the improvement with FULL anti MD The thing that you and I have to realize (in a very respectful way) is that a lot of audiophiles haven't been listening for these kinds of defects all of the time. I truly DO NOT KNOW if/when even a perceptive audiophile hears the difference. It is a tricky thing -- my hearing has been trained specifically for the effects of IMD and MD, but I do believe -- as you say -- the differences should be noticeable. (Esp for you, since you have been helping with testing/verification/discerning results.) The more 'practical' matter WRT the anti-MD code is that it is very compute intensive, and isn't really practical for day-to-day use except for a one-off decoding operation. The lower quality modes are more doable in real-time, but don't give quite the major improvement. These last fixes (true bug fixes) have enabled me to use the normal 'high quality' mode for all but the most critical decodes. I have a decoding level scale that goes from 1 to 10, with 11 and 12 being extra better. Starting today, I have been more than satisfied when using the level '10', but am still using '12' for 'Crime'. Frankly, level 6 would be a godsend for those expecting DolbyA quality, and maybe the lower --xp mode (level 4) would be more than enough -- and decodes relatively quickly -- 2-3X faster than realtime.. Today, I created the 'disable MD' mode so that the lesser modes could be demonstrated, but there were also some anti-distortion mechanisms that would have been a lot of trouble to remove. There are all kinds of timing compensation schemes so that when there is a filter, then the timing effects of that filter are also compensated. There are some shaping filters that a true dolbyA doesn't have, which certainly produce a marginal improvement on the DHNRDS. There are several 'little' small improvements that really do help - but I didn't bother removing. However, that final 'anti-MD' and the front end 'anti-IMD' code makes the major difference that gets rid of most of that DolbyA compander sound. The mechanisms that I developed are not well known at all, and just recently (in the last 10yrs) became practical to run in real-time on PCs. Imagine an old 'big computer' like a CDC6600, DEC10 or IBM 360/158 -- it would take them a year (literally) to DolbyA decode a song, and a VAX 8600 or Intel 386 would take several days to a week. These new PC computers are literally 30000-100000 times faster than a VAX-11 and 10000 times faster than even a VAX8600, for the kind of math being done on the DHNRDS... A cray2 would start becoming practical for super high quality decoding, but would be many thousands $$$ of CPU time per song, to do the decodes. Thanks for the feedback -- finally I got up the motivation (after fixing the last bugs) to do a full disable of the anti-MD/anti-IMD mechanisms. John Link to comment
John Dyson Posted January 4, 2020 Author Share Posted January 4, 2020 Here is a simple, clear difference between the typical CD vs the decoded version (I remastered it SLIGHTLY -- just some EQ, no DRC.) The reason for the EQ is because the DolbyA decoded copies are the RAW MATERIAL, and often need a bit of a tweak for given media. I made some assumptions about vinyl (since the recording was done in the vinyl days), and produced this result. The difference is STARK -- and at first the warm sound of the CD might seem okay, but then you'll hear what you are mising as she really starts singing. Past this point, it is a matter of personal preference -- but I do not prefer the 'elevator music' sound of the CD. (Okay, elevators tend not to have much bass in their sound systems -- but otherwise the sqeezed dynamics are typical.) It might be reasonable to add some DRC to the 'mastered' version -- but that is going WAY too far for me to do it. John Linda Ronstadt - 09 - Different Drum-CD.mp3 07. Different Drum-mastered.mp3 Link to comment
John Dyson Posted January 4, 2020 Author Share Posted January 4, 2020 I wanna clarify the difference in sound between the various 'normal' versions and even my 'mastered' versions -- there is ZERO compression, and even with my 'mastering', it is just a bit of EQ that helps to audibly compensate for the lack of compression. Also, when listening to Linda's voice -- SHE GETS LOUD!!! She had a wide range... Think about it -- why do people say that she has a good voice -- is it sweet sounding? Nope, she has a STRONG voice, and a very wide dynamic range. So, when she gets loud, then she gets REALLY loud. I was astounded with the dynamics myself -- getting confused, because if her voice ended up sounding more normal, then everything else was screwed up (the chorus was all mangled.) The fact is -- she can belt it out!!! Some compression at moderately high volume levels might be helpful -- my own compressor doesn't handle her dynamics very well, but a small amount of compression in the -12 to -3dB range would be helpful. The low level cmpression like the feral-DolbyA stuff that is so common -- that just increases the piano sustain way way too far. A different compression curve would be needed for LR's stuff. John phosphorein 1 Link to comment
fas42 Posted January 4, 2020 Share Posted January 4, 2020 John, just to say that I prefer the original 😜 ... again. I have no trouble hearing the strength in Linda's voice, in that one - but in the processed variety I hear the backing losing its strength; the instruments have "withered" a bit ... Link to comment
sandyk Posted January 5, 2020 Share Posted January 5, 2020 3 hours ago, fas42 said: John, just to say that I prefer the original 😜 ... again. I have no trouble hearing the strength in Linda's voice, in that one - but in the processed variety I hear the backing losing its strength; the instruments have "withered" a bit … Frank Your reports are of no value whatsoever unless you at the very least use a conventional stereo set up, not rely on what you are hearing from tiny little speakers on a laptop with a far from ideal frequency response and a mediocre S/N ratio. If you wish to be taken seriously, you need to, at the very least, DL them and save them to USB memory, and listen to them through your Audiophile friend's system. Teresa 1 How a Digital Audio file sounds, or a Digital Video file looks, is governed to a large extent by the Power Supply area. All that Identical Checksums gives is the possibility of REGENERATING the file to close to that of the original file. PROFILE UPDATED 13-11-2020 Link to comment
fas42 Posted January 5, 2020 Share Posted January 5, 2020 Luckily, I don't have a hangup about needing to use an "audiophile approved" rig to hear what's going in with some recording, Alex 😜 ... IME, it's too much work trying to listen through the flaws of the playback chain, and I just get annoyed with the exercise. A very simple setup that presents the core of the sound, without damaging the treble too much, is good enough - extra volume brings in a whole set of other issues, which get in the way of determining things ... sandyk and Teresa 2 Link to comment
John Dyson Posted January 6, 2020 Author Share Posted January 6, 2020 Here is the full set of snippets for Linda Ronstadt Greatest hits. Notice the piano in Desperado having a normal trail off instead of sounding like the sustain pedal is depressed.. (Desperado is what I was writing about before -- confused that I didn't make the demo available!!) The first cut also has a horrible chorus -- it seems like it is correct? Don't really know. The hiss should be pretty minimal also. https://www.dropbox.com/sh/oarqeasbh8hqh4t/AABQ7JhfCQ84j7QDP8VKAlK8a?dl=0 Link to comment
Confused Posted January 12, 2020 Share Posted January 12, 2020 I was listening to this via a CD rip earlier. It just sounds wrong to me, very thin and shrill. If I had to guess, I would say that this is a pre emphasis issue, maybe just poor mastering, maybe "FeralA"? The You Tube clip attached sounds much the same as my CD rip, any thoughts? In comparison to the above, this sounds more correct to me..... Windows 11 PC, Roon, HQPlayer, Focus Fidelity convolutions, iFi Zen Stream, Paul Hynes SR4, Mutec REF10, Mutec MC3+USB, Devialet 1000Pro, KEF Blade. Plus Pro-Ject Signature 12 TT for playing my 'legacy' vinyl collection. Desktop system; RME ADI-2 DAC fs, Meze Empyrean headphones. Link to comment
John Dyson Posted January 12, 2020 Author Share Posted January 12, 2020 9 minutes ago, Confused said: I was listening to this via a CD rip earlier. It just sounds wrong to me, very thin and shrill. If I had to guess, I would say that this is a pre emphasis issue, maybe just poor mastering, maybe "FeralA"? The You Tube clip attached sounds much the same as my CD rip, any thoughts? In comparison to the above, this sounds more correct to me..... Very probably, it is feralA or an 'enhanced'. In this decoding attempt, note the relative lack of hiss just at the beginning of the intro. Here is an attempted decoded snippet (time location specified in snippet name.) I will send you the entire piece privately. Also, I did two decodes, this one is a little more bright, but the first one I did was a little more natural. I think that the pop recording would be done on the brighter side. My result seems a little upper midrange heavy, but that is just a matter of finding the correct corrective EQ, which sometimes requires trying several selections on the same album to make the right noice. I used 'vanilla' MF parameters and a 2nd choice HF set of parameters to make the material sound more like a pop recording. Elsewhere, I have lamented about the lack of reference material that has been properly handled, but otherwise I try to do the best that I can. Someday (hopefully soon), I'll make the DHNRDS easier to use for feralA, even though it will still be command line. At least, it won't need SOX to do the EQ, it will be able to do the EQ itself, with selected modes/parameters instead of needing to specify the filters from scratch. Give me a short while to send the full result privately, I have to clean-up my dropbox space. John prince-snippet-55secto110.mp3 Link to comment
John Dyson Posted January 17, 2020 Author Share Posted January 17, 2020 Minor bit of 'good news'. I think that *finally* I got a good decoding formula for all of the 'The Carpenters' original albums on CD. (early CD releases of the albums.) The formula wasn't difficult, it is just tricky to choose the correct set of parameters. It almost requires *soul searching*, because the results of incorrect parameters can be as minimal as an undesired tone control setting. In the examples -- note the fullness and stereo image reconstructed from the 'undecoded' harsh ness. This might be the kind of sound that started the 'harshness' complaint those 35yrs ago. I have documented these for future full decoding efforts. Each album is different, but the rest of the albums weren't too hard. The 1972 album CD release is generally the worst and very aggressively too bright. When listening to the very bright sounding material, it distorts my hearing and it is so easy to make mistakes when listening to properly decoded material. The compensating feralA EQ is very strong. (-6dB at approx 9kHz, -12dB at approx 12kHz, the highest that I have seen.) Please play the 'decoded' version first. If listening to 'undecoded' first, it might make the 'decoded' one sound a little dull. But, the 'decoded' one is NOT dull!!! I am very close at deciding on the correct built-in helper formulas for feralA decoding EQ. it is looking like there will be a default EQ and default calibration level that works fairly well. Then there will be some mode switches for the MF part, and a mode each for the 3k/6k and 9k/12k eq, with one gain for each pair. That is about as simple as it can be made to be. There will be a short several page note that describes suggested EQ settings for material that I have found. The decoding is becoming VERY robust, and minor errors don't produce unlistenable results (in fact, better than the feralA itself, even with minor setting errors.) John 04 - It's Going To Take Some Time-undecoded.mp3 04 - It's Going To Take Some Time-decoded.mp3 Link to comment
The_K-Man Posted January 17, 2020 Share Posted January 17, 2020 I posted this question in this thread several days back, but it never went through(thanks a lot, repealers of Net Neutrality!! 🤦♂️ ) I was wonder when, how, and why, mistakes surrounding Dolby A encoding decoding during playback for transfer to CD started to occur: 1. Encode but no decode 2. Decode but no encode Etc. And also, what audible clues to listen to on my CDs. Thanks! Link to comment
John Dyson Posted January 17, 2020 Author Share Posted January 17, 2020 Just now, The_K-Man said: I posted this question several days back, but it never went through(thanks a lot, repealers of Net Neutrality!! 🤦♂️😞 I was wonder when, how, and why, mistakes surrounding Dolby A encoding decoding during playback for transfer to CD started to occur: 1. Encode but no decode 2. Decode but no encode Etc. And also, what audible clues to listen to on my CDs. Thanks! I am so sorry if I didn't answer the question, I don't remember seeing it -- various things have been going on. Encoded but no decode actually seldom happens, but instead they do an 'Encoded', but extra 'EQ' that hides the nasty, intense DolbyA sound. So most commonly, we have these (in order); A0) Encoded but EQed to sound okay. A1) Encoded and correctly decoded. B) Encoded but not decoded. C) Decoded but not encoded. C would seldom happen, because it usually sounds really bad for a DolbyA equivalent decoder to 'decode' uncompressed material. Sometimes material compressed with other than DolbyA can sound listenable when decoded, but 'C' is a serious mistake. I included a copy of 'C' above because I wouldn't have to carefully review the results. (Unreliable hearing right now.) B sounds pretty bad. It is harsh! The worst album that I have is 99Red Balloons from Nena which has both CD pre-emphasis and DolbyA encoding. With just the raw DolbyA, is pretty ugly. How do you detect feralA or DolbyA? It is tricky -- a practiced ear for the HF compression and that isn't very reliable based using my own hearing. Also, you can check for hiss at the lead-in, intersyllable or fade-out -- sometimes that is a pointer to too much hiss, therefore possiibly undecoded DolbyA (in feralA form or native DolbyA.) Also, looking at the spectogram, bring the background hiss up so that you can just barely see it for the 3k-10k range. Sometimes, DolbyA where it was actually used for NR purposes will show a small increase in hiss above about 12kHz. (That shows the extra 5dB of compressionn gain for the 9k+ range.) My hearing is corrupted right now, so making examples for overt comparisons might be unreliable right now ( could easily botch it.) Here is the previously posted snippet/song "its gonna take some time', but decoding the alrleady decoded material -- but I used the pristine flac version for the source instead of the mp3. Also, the material is alread compressed, so the excess expansion isn't so terrible, and the middle frequencies are essentially saturated (the gain is the maximum 0dB at apprpox -25dB input, so is basically pinned on.) Notice that the worst of the loss is in the highs and stereo image damage. * A true DolbyA would be worse sounding. (Note that my previous decode might not have been 100% accurate, my hearing was starting to fail earlier today. I keep on trying to bounce between two versions -- but the sibilance is better on the version that I previously uploaded, but has a bit too much 'growl'. When cleaning up the 'growl', then the sibilance gets a 'hitch' in it. So, I uploaded the best of several attempts -- and when listening to the entire album, the results are pretty good -- just that one song bothers me) . 04 - It's Going To Take Some Time-decoded-doubledecoded.mp3 The_K-Man 1 Link to comment
Recommended Posts
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 accountSign in
Already have an account? Sign in here.
Sign In Now