|
Post by johndyson on Dec 29, 2019 8:37:22 GMT 10
I think that I prefer the 'before' (the first one.). The mild compression/enhancement brightens it up for me. I might have not gotten the settings quite right either...
Mono did surprise me also.
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 29, 2019 9:03:57 GMT 10
John I thought that your version sounded a little better when I turned the volume up, but neither version is remotely like the quality from the master of Fever, or for that matter more recent versions of Black Coffee from either Norah Jones or Claire Martin. They were capable of doing WAY better back then !
Kind Regards Alex
|
|
|
Post by johndyson on Dec 29, 2019 21:34:06 GMT 10
I think that we just came up with a new improvement for some of the 'difficult-to-decode' feral DolbyA materials. Alex made me aware of some quality problems with the decoding attempts on the early Carpenters recordings, at least the 1970 and 1971 albums, and probably the rest.
Previoiusly, when decoding the Carpenters correctly WRT the DolbyA, the result was still 'soft' sounding. We found that subseqent, post-decoding EQ was needed to make the recordings sound more close to the original. I resist doing subsequent EQ after decoding because I do not like to 'master' materials. This intention to avoid mastering isn't an arrogant choice, but is a considered decision to be 'true' to the original recording. That idea that the 'decoding result' is the desirable 'original recording' is often wrong, because some recordings are apparently EQed before DolbyA encoding.So, one of my important rules hasn't changed, which still IS: don't do any mastering, except sometimes I try to recover NATURAL bass or get rid of some kind extreme problem on the high end (usually resulting from mixing for vinyl.) I don't like making major changes. However, since I do need to get results that sound close to the original, I have to open-my-mind to post-decoding EQ, but not as 'mastering', but instead to complete the decoding effort.So, we have 100% found that at least part of the Carpenters albums as distributed to consumers need both the corrective EQ and the post-decode EQ. Along with that, the Carpenters albums need to be decoded (all of them) in the mid-side decoder configuration. ABBA has also been trouble. ABBA has been my 'problem child' all along. Where I can get good quality decodes from something like Supertramp after several days of hard work, I haven't been able to get fully 'clean' results from ABBA, maybe until now. JUST TODAY, JUST NOW, I realized that at least several ABBA albums *also* need to be decoded in mid-side mode instead of the straightforward normal stereo left-right mode. Chorus singers line up into a more stable position when decoding ABBA in the M+S mode, which is definite proof of the need to use the mid-side matrix on the input and output. Also, very critical comparisions are telling me that post EQ is also needed.So, some of the big groups who did lots of recordings for some reasons have been treated to the mid-side processing choice. I haven't seen this on Supertramp. Today, I intend to chase this ABBA thing again for a while. It might be interesting to hear the eventual results... Everyone reading this, please remember that my goal is to be *faithful* to the recording, the recording engineer and the artist, and not to be an artist myself. I am trying to prove that there just might be a market for true audiophile copies of recordings, and not just placating the public with 'high res' as being the ultimate quality...John
|
|
|
Post by johndyson on Dec 30, 2019 11:29:24 GMT 10
Amazing announcement about 'Crime' and a lot of other decodes... I am starting to find that more and more decode attempts were suboptimal because of the L+R vs M+S decoding.
At least the 'Crime of the Century' album was apparently encoded in M+S mode instead of L+R. There are some technical advantages, basically being less dependent on the DolbyA unit being balanced, and also the difference signal will tend to be lower level, therefore more likely increasing the signal-to-noise ratio.
So, by using the M+S decoding configuration (quality maintained only if the encode/decode configurations are the same), there is a slight improvement in SNR. Here is what I mean by using M+S:
signal from tape -> convert L/R to M/S -> decode -> convert back to L/R.
The M/S can be converted to L/R by L = (M+S)/2, R = (M-S)/2. The L/R can be converted to M/S by M = (L+R)/2, S = (M-S)/2.
The problem, for those who actually look at th relation is that one of the conversion (L/R -> M/S) or (M/S -> L/R) need an additional factor of two, or the levels are messedup... That is just a minor detail that I resolved.
The resulting decode of 'Crime' appears to be slighlty better. I'll be posting it soon.
John
|
|
|
Post by johndyson on Dec 30, 2019 14:36:33 GMT 10
Wow -- this new 'Crime' superficially sounds very similar to the previous version... It is possible to give very clear A/B comparisons to show more detail!!! I didn't think that it could get any better (again!!!) I am working on finishing it up. It required about 4-5 decodes to get the EQ the same (it was actually easy to do, but requires lots of cross checking -- that is, detailed.) THe EQ values are very normal. THe M+S decode almost normalized the recording in many respects.
Next time I post, there should be a new 'Crime' available -- I hope!!!
John
|
|
|
Post by johndyson on Dec 31, 2019 14:49:32 GMT 10
Here it is -- the Major-Mega-Amazing decode of 'Crime of the Century'. This is very serious, and any loss of distortion did NOT result in a loss of detail. In fact, it is very likely that an A/B comparision will make the loss of detail in previous versions quite suprising. (I didn't check this specific version, but several previous versions using nearly the same parameters.) This is the result of approx 50 decoding passes of at least a portion of the album. -- STATUS REPORT: There was one major bug in the previous decodes, which stupid me didn't see (needed to decode in M+S instead of L+R). This M+S thing fixes a lot of transient issues -- probably not noticeable without A/B comparisons. Using Q=0.50 filters instead of Q=0.707, thereby getting better transient response, but the high end freq range is the same between previous releases, but will sound different/better stereo. Vastly improved DHNRDS DA decoder (part of what took so long to get this release out.) Proof of the pudding - listen to the 'tooting tone' move across the stereo field at the beginning of cut 03, Hide in your shell. Previous versions did it poorly. Also, even though the vocals are just as intense, notice no edgy distortion unless reproducing a 'hitch' in the voice. Fantastically stable stereo image!!!
No-one in their right minds would spend the time, effort to produce the incredible detail and balance with minimal distortion and precise clarity from the recording.
Please have fun -- this is a 'once in a lifetime' decoding effort, and 'impossible' decoding result!!!
--- RECORDING: Please contact either John or myself via PM for the Download links
EDITED by Alex.
|
|
|
Post by johndyson on Dec 31, 2019 15:21:57 GMT 10
Important update -- made a mistake on the upload, picked the wrong files. I double checked and found the error, so as of this posting time, the files should be correct. Very important - do NOT be disappointed about the lack of vocal distortion -- the spectrum goes all the way up to 20kHz very nicely!!! The extreme distortion free sound is impossible without the latest version of the DHNRDS DA decoder... Richard is getting another release tomorrow!!!
John
|
|
|
Post by ROWUK on Dec 31, 2019 22:57:00 GMT 10
So, I have listened to the "new" Crime album and technically I hear no decoding artifacts. There is bass in abundance. The upper register is very present but with no exaggerations. I have no sibilance issues in the voices. Could it be that the decode was lowered in absolute level? I have to turn it up a lot more than with other albums.
The solo voices are very much on the "bright" side of natural (signature Supertramp sound) but here brighter for instance than the last Quietest Moments decode. The piano is lovely. The saxophone and clarinet are spot on natural. I have no reference for electric bass or drum set but they fit naturally into the sonic picture.
I do have some issues with the production however. The voice and saxophone are positioned back a meter or two (nice) but the hand drums (not the drum set) are in my face.
Sometimes the strings sound phasey or distorted - but there is always something going on at the same time like the solo sax that is perfectly clean so it is surely the original master. Maybe they aren't even real strings. I couldn't tell.
Comparing to vinyl. Well, there is no comparison to my pressing. Johns' restauration is simply better in every way! Congratulations!
Still, the Quietest Moments decode is still my favorite. The technical merits of the decode made me want to listen to it a LOT more often. The Crime is nice, but I am not drawn into it in the same way.
|
|
|
Post by johndyson on Jan 1, 2020 0:42:01 GMT 10
So, I have listened to the "new" Crime album and technically I hear no decoding artifacts. There is bass in abundance. The upper register is very present but with no exaggerations. I have no sibilance issues in the voices. Could it be that the decode was lowered in absolute level? I have to turn it up a lot more than with other albums. The solo voices are very much on the "bright" side of natural (signature Supertramp sound) but here brighter for instance than the last Quietest Moments decode. The piano is lovely. The saxophone and clarinet are spot on natural. I have no reference for electric bass or drum set but they fit naturally into the sonic picture. I do have some issues with the production however. The voice and saxophone are positioned back a meter or two (nice) but the hand drums (not the drum set) are in my face. Sometimes the strings sound phasey or distorted - but there is always something going on at the same time like the solo sax that is perfectly clean so it is surely the original master. Maybe they aren't even real strings. I couldn't tell. Comparing to vinyl. Well, there is no comparison to my pressing. Johns' restauration is simply better in every way! Congratulations! Still, the Quietest Moments decode is still my favorite. The technical merits of the decode made me want to listen to it a LOT more often. The Crime is nice, but I am not drawn into it in the same way. --- Thanks, and thanks for the analysis -- I can really understand what you are writing about. All too often people give feedback (critical or positive) that uses language that is difficult to interpret, but your feedback is as clear as day. As a result of your feedback, I AM thinking about the voice and saxophone positioning, and believe it or not I JUST MIGHT have a modicum of control over it -- could be a 1/2 dB or so difference in midrange. There is a kind of EQ BEFORE decoding that might be a little in error (I felt the most recent to be as close as I have ever gotten, and will as always, consider your comments as very very very important.) Unless I can make any other improvements, which might result from the positioning matter, this version of Crime REALLY REALLY is the final version. I keep the decoding formulas, and if there is a need for a direct decode into another format, or starting with an unmolested high-res version of the same material), then I'll be able reproduce the same results, but with whatever source improvement or target format that might come up in the future.
I plan to work on both Breakfast and Quiet in the very near future (days, not weeks.)
Also, there were some major breakthroughs in the decoder which also allowed doing some really clean decodes of the Carpenters. I did find, with some feedback from Alex, that UNFORTUNATELY, the Carpenters absolutely need some post-decoding EQ, which I seriously avoid. Post-decoding EQ is tantamount to 'mastering', which I try to avoid because 'mastering' encroaches too easily on the artist's intent. Note: it appears that many of the Albums need the same amounts of post-decoding EQ. Knowing that the EQ should be close to the same -- I'll do an iterative effort to optimize the EQ based upon multiple albums. My ear for the actual, real world sound of music is not practiced, so it is a good idea for me to be very conservative in my decisions about EQ.
The mastering issue is one area were I do try to avoid. If major modifications of the decoding results are made, then artistic aspects of the recordings can be affected. I have developed a skill of detecting DolbyA decoding artifacts, so can come fairly close to correct decoding even with molested source material (EQ for distribution or loss of calibration levels.) I do not normally make any changes beyond those corrective in nature, and very seldom make substantive changes to the character of the recordings.
So, after the currently running re decodes of the Carpenters today, and during a parallel effort working on some Linda Ronstadt decodes, I'll be working on 'Breakfast' and 'Quiet' using what I have learned from "Crime".
Will keep you (and Alex) and everyone else informed as to progress, etc. New ideas always welcome.
John
John
|
|
|
Post by ROWUK on Jan 1, 2020 2:27:47 GMT 10
John, being a professional trumpeter does put me at a disadvantage with pop recordings. Recording engineers almost always have their "interpretation" of voices and instruments that has little to do with the natural sound and space that we get on stage in classical music or in a nice room during rehearsal.
The measure for me in your work is "would I buy this type of restoration", "does it allow me to listen with less distraction". Currently my opinion is very much that the tool is only an extension of the engineers respect and taste. The 3 Supertramp definitely qualify as a hearty YES - buy them (in spite of "Crime" being a half generation further along technically). I do have my doubts that the owners of the material share your altruism - otherwise we would have gotten something much different in the first place. These decodes most certainly rekindle my long lost interest in this music.
Abba and Carpenters are also groups that are of great interest due to the use of acoustic instruments. Other groups of interest would be Steely Dan or The Alan Parsons Project.
One possible venue would be HD Tracks. Currently they have a lot of interesting stuff in the library but I am equally disappointed and surprised at the "quality" that they offer. I am sure that they are only offering resampled masters. I am not aware of any restoration attempts by them.
P.S. I have taken a second listen to Crime decode after playing the LP. I am pretty sure that there was some major remastering - not just EQ on top of Dolby A. Your decode has more space around the instruments (something my Vinyl playback is especially good at)
|
|
|
Post by johndyson on Jan 1, 2020 3:45:30 GMT 10
About your comment about 'space' around instruments -- WOW!!! Man you have really good hearing. You are the first person *ever* who has elucidated that exact notion. Most people describe the benefit as 'more detail', but not in these very-precise terms. I search high and low for such instances where there is improvement of detail around the instruments or audio event. Mitigating the loss of detail was an original design goal of the DHNRDS, and it is so gratifying to get feedback describing that specific difference.
The 'space around the instruments' matter is at least partially a direct result of modulation distortion. As the modulation distortion gets worse, then the details around individual 'events' in the music starts being increasingly obscured . The decoder itself had a serious bug where it previously had left about 1/2 of the modulation distortion in the material, so there was an improvement in detail beyond normal DolbyA, but not quite at the level of the most recent version of the DHNRDS DA decoder. After I detected that bug -- a really stupid programming botch, literally a typing error, then alot of ugly 'veil' around the recorded signal went away. After this recent fix, I had expected that there would be noticeable improvement, but not as clearly described as you have.
DolbyA units NEVER avoided creating the modulation distortion because of the impossibility in hardware. Back in the early DolbyA days, perceptive recording engineers and their musician cohorts noticed the loss of detail when using DolbyA NR units. I had originally intended on removing 'all' of the modulation distortion 'veil', but my programming botch left 1/2 in.
The most recent decoder changes since the last demo were:
1) fix DHNRDS DA modulation distortion processing 2) improve accuracy of attack/release 3) correct some filtering that used to help a little with modulation distortion (useful because of the MD bug). Extra filtering no longer needed, bad tradeoff now. Removing filtering vastly improved Carpenters decodes...
4) minor improvement in frequency response flatness 5) (recording specific): revisit the corrective EQ in and out of the decoder.
You just 'made my day' by noticing the improved detail around the instruments, etc. A huge amount of the DHNRDS program is dedicated to keeping the detail while gain control is actively bouncing around...
Thanks so much!!!
John
|
|
|
Post by ROWUK on Jan 1, 2020 18:50:51 GMT 10
John, I and most of my musician friends have no idea about audiophile listening. Things like additional detail really have no context unless we attach that to specific event. We listen to instruments as we know them live. Each instrument has a signature “size” (grand piano =big and diffuse, Trumpet small and directional). Certain instruments like the piano, we do not physically get close to. There is physical space around it. Never ever in real life does it sound like the trumpet is playing from the same location as the piano - they are always at least a meter away. What tells us how far, are the early reflections. Naturally, in a studio situation with only direct microphones, space is destroyed - there are no early reflections. In the Crime and Quiet instances, it appears that the piano was properly recorded. There is a sense of size and space. What I noticed and especially liked was the clarinet and saxophone sounds. The clarinet was smooth and whispy but the saxophone had that explosive honk that often gets eaten in the mix. Those instruments had believable size.
In any case, I don’t have golden ears and basically only compare recordings to my experiences with the real instruments in spaces designed for the type of music that we are playing. Big halls or churches for big music, smaller, more intimate spaces for chamber music. I have no references for electronic instruments like keyboards or electric guitars or bass. How much drum set cymbal is correct? What does a real non-amplified bass or kick drum sound like? The voices are manipulated, so they also only provide basic clues, although there are standard facets of the voice like articulation, core and halo that remain even with heavily processed tracks. The acoustic instruments each have their signature. That is my area of “expertise”.
I have a lot of respect for what you are doing. I wouldn’t know where to start.
At the end of the day, I want a coherent performance. Morons in the mix only raise the level of tension when listening because artifacts draw away from the musical message.
|
|
|
Post by johndyson on Jan 1, 2020 22:09:13 GMT 10
John, I and most of my musician friends have no idea about audiophile listening. Things like additional detail really have no context unless we attach that to specific event. We listen to instruments as we know them live. Each instrument has a signature “size” (grand piano =big and diffuse, Trumpet small and directional). Certain instruments like the piano, we do not physically get close to. There is physical space around it. Never ever in real life does it sound like the trumpet is playing from the same location as the piano - they are always at least a meter away. What tells us how far, are the early reflections. Naturally, in a studio situation with only direct microphones, space is destroyed - there are no early reflections. In the Crime and Quiet instances, it appears that the piano was properly recorded. There is a sense of size and space. What I noticed and especially liked was the clarinet and saxophone sounds. The clarinet was smooth and whispy but the saxophone had that explosive honk that often gets eaten in the mix. Those instruments had believable size. In any case, I don’t have golden ears and basically only compare recordings to my experiences with the real instruments in spaces designed for the type of music that we are playing. Big halls or churches for big music, smaller, more intimate spaces for chamber music. I have no references for electronic instruments like keyboards or electric guitars or bass. How much drum set cymbal is correct? What does a real non-amplified bass or kick drum sound like? The voices are manipulated, so they also only provide basic clues, although there are standard facets of the voice like articulation, core and halo that remain even with heavily processed tracks. The acoustic instruments each have their signature. That is my area of “expertise”. I have a lot of respect for what you are doing. I wouldn’t know where to start. At the end of the day, I want a coherent performance. Morons in the mix only raise the level of tension when listening because artifacts draw away from the musical message. BTW, I am not a 'golden ears' at all either, but an EE/DSP/computer person who has always had perceptive hearing. I gave up on the audiophile hobby back in the late '80s or early '90s because the 'ultimate' distribution medium -- CD, they usually sucked.
I couldn't describe the sound other than 'shrill' or 'something wrong'. The big turn-off happened in the early '90s where I had purchased a CD from Sheffeild Labs 'I have the music in me', and it was so shrill as to be unlistenable. I previously had the vinyl some 10yrs before, and loved it. During moves or whatever, I had lost the album, but wanted to replicate the experience in the CD. After a myriad of disappointments with other CDs, I started giving up. I thought that we would never get the true promise of 44.1k/16 bit digital sound...
Roll forward to 2012, I was casually listening to an ABBA Gold CD, and with a more complete set of listening/mental/technical tools -- noticed LOTS of HF compression... After analyzing that CD for a while, realized that DolbyA was not decoded somewhere & somehow.
After many years of learning and experimentation, and Richard Hess joining the effort, who gave me access to DolbyA units -- the effort came together and we ended up with the creation of the DHNRDS DA DolbyA compatible decoder.
So -- I am FAR FAR from being a golden ears, and this has been a long journey. The amazing thing to me -- is that the terrible infestation of DolbyA into consumer material had not been mentioned ANYWHERE FOR 30yrs until recently. What happened to the investigative press? DolbyA was never intended to reach the consumer's ears!!!
John
|
|
|
Post by johndyson on Jan 3, 2020 9:36:59 GMT 10
Here is a status report on what I am doing WRT DHRNDS DA decoder & the various albums that I am trying to decode:
DHNRDS: Numerous little tweaks/bugfixes to help zero-in the timing between the gain control & the signal, making sure that they match in the time domain. Also, slightly widened the attack/release to more carefully encompas the signal events -- this decreases MD and other kinds of distortion if tuned-in precisely. Both efforts are complete and fully optimized. (The improvements on the Carpenters decodes come mostly from this 'gain control' signal effort.)
ALBUMS: I jump around a lot. With the Linda Ronstadt special on last night, I did a marathon decoding effort on "Greatest His" - carefully finding the correct *pre-decoding* EQ, but also tried to produce a better post-decoding EQ. I severely resist doing true 'mastering', but some albums really need help. The either/or decision is in Alex's hands right now. I cannot decide on the two versions created from earlier today.
Supertramp:
I am doing Supertramp on a album by album basis, and 'Breakfast' is good -- but not stellar, and having troubles with the vocal enhancement in the 'Quiet' recording. They did a huge boost in the 9kHz region, and it is so very tricky to get that right -- that is, leaving the enhancement, but getting rid of the associated distortion.
Carpenters: The decoding is as close to perfect as possible -- gotten rid of the edgy vocal enhancement effects, but I need to look at potential post-decoding EQ (it just might be necessary per some examples provided by Alex.)
ABBA: Super good progress. The mastering and/or distribution encoding was devilish. All are M+S encoded, but on top of that, most of the recordings had a 20% widening of the stereo image AFTER EQ. That is, the signal needs to be narrowed in the L+R sense AND converted to M+S before decoding. Additionally, about 1/2 of the albums had a different calibration between the M and S channels. The S channel calibration sometimes needs to be 6dB lower (which means 'S' is assumed to be a lower level, which is sometimes true.) I am doing the final prospective decode of ABBA *right now* (all 8 albums), waiting for the decodes to come though the process, at absolutely the highest quality mode. Initial test listening shows very good, smooth sound. Good stereo, none of that grainy sound that is sometimes manifest on ABBA.
---- After Linda, I expect the ABBA results to become available to review. After that, I am going to focus 100% on Supertramp.
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 Jan 6, 2020 8:44:52 GMT 10
UPDATE.
John is currently working on a Linda Ronstadt album as a tribute, however he has in the process uncovered a few more tricks that were used back then. Among other things, it has become obvious that Linda had a very powerful voice back in those days before she aged and heartbreakingly developed Parkinson's disease. There was a recent TV special on this sad development.
|
|