|
Post by johndyson on Jan 6, 2020 21:57:37 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. The current state of Linda Ronstadt is that it appears that the decoding is correct, and was very close to where I want it to be. I can hear one flaw that should be corrected -- the high end is a little bit too hot. I'd guess bettween 0.75 and 1.5dB too hot in the 9-12kHz range. Such a boost gives a slight brittle sound, but mild enough that a slight tweak downwards on a tone control could help. I just checked using a filter of -0.75dB/Q=0.50 at 12kHz seemed to wrap it up nicely. I'll work on replacing the files in the repository -- about 1Hr ETA. (I just woke up, and it is like my hiearing is recharged after waking up... Yesterday, I couldn't hear the hot high end.)
It *is* amazing what even a +0.75dB error can do. Actually, the correction at 12kHz with the filter that I specified is NOT -0.75dB, but instead that is the maximum gain change -- but at 12kHz, the gain change is -0.375dB!!! The need for these little changes isn't uncommon at all. The DHNRDS DA itself is about +-0.35dB, and that is TOO MUCH. I biased the error on the low side, but still maintaining accuracy over time and vs signal level is tricky. Instantaneously, the DHNRDS can have a higher error, like during fast gain changes. If it would track a DolbyA perfectly, it would produce more distortion, and the 'improvement' would be heard as a DECREASE in quality. *Fast gain control is a weird thing.
* EDIT: the files are updated. For the passively curious, I also uploaded some snippets of the same thing (truncated and only mp3) to the following location (remember that the online mp3 decoder further degrades mp3 beyond the normal lossage): Also -- I got as much out of the recording as I can. There are flaws -- the chorus is too hard sounding, and that *is* partially because the DHNRDS does not soften the recordings as would be expected by a true DolbyA. True expert users of the old DolbyA technology appear to have known how to partially compensate for the softening (the DolbyA fog), so the DHNRDS will show the results of the compensation.
John
|
|
|
Post by johndyson on Jan 8, 2020 5:03:48 GMT 10
I've been getting good reviews on the ABBA results (only demoed Ring Ring so far), and also the Carpenters 1970-1976 are being decoded right now. I just sampled a few of the cuts immmdiately as they came out of the decoder, and got some good, positive feedback. The decoding is NOT perfect yet -- I have learned a lot about possible defects. Also, I am learning, studying the defects in these first full attempts with the new technique of undoing the feral-DolbyA. I'd suspect that even right now, the review would be pretty good.
I don't want to waste anyone's time when I am going to do another set of decodes over the next 1-2 days. When they are done, I will make the available for the most critical review to people here on request. Usually, I do simple tests over on another forum -- getting casual feedback and interest. I take this forum more seriously - so when the more refined results are produced, I'll offer them for review and maybe just because of cursory interest.
One item of note: the Carpenters decodes are coming out pretty darned good. They are NOT where I want them to be, but they are close... When they are done, they might be nice to add to an existing Carpenters collection -- REALLY... (I wish Alex's hearing was working better for him right now!!!)
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 11, 2020 15:24:45 GMT 10
If any member is interested in hearing a few sample tracks from John's decode from Brasil 66 ( The Look of Love etc.) please send me a PM. It is from a CD that appears to have been sourced from a multi channel recording.
Alex
|
|
|
Post by johndyson on Jan 14, 2020 9:34:09 GMT 10
If any member is interested in hearing a few sample tracks from John's decode from Brasil 66 ( The Look of Love etc.) please send me a PM. It is from a CD that appears to have been sourced from a multi channel recording. Alex Brasil'66 came out (even though improved), "okay". The album that sounded esp good to me was Herb Alpert. I am not sure that others would feel it to be as good, but something about it seems 'special'. It might be the 'quad' stuff, not sure.
With so little experience with the multichannel stuff, I might have created too much seperation.
The problem with these feralA decodes is that they entail messing with the stereo width (often doubling it - multiplying S of M+S times two), and then squeezing back to something other than 50% (It seems like it is either 50%, 59% (a sqrt(2) related value) or 70.7%. They mess with the image for the feralA because the compression changes the perception of stereo, so they modify the stereo after filtering the raw DolbyA sound to produce semi-listenable signal.
I wouldn't mess with the stereo image if I didn't have to. without knowing the exact formula, there ends up being a kind of subjectivism that I am not good at doing -- I might be missing some kind of stereo image perception ability. I have good tone and signal perception, but not good stereo.
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 14, 2020 10:29:12 GMT 10
HI John.
No, that's how it was intended to sound back then in the SQ 4 channel mode etc. as evident in stuff I had back then, including a logic enhanced SQ decoder as well as some Enoch Light material that was later converted to stereo for CD release, but retained that sound to a large degree.
Regards Alex
|
|
|
Post by johndyson on Jan 31, 2020 22:04:59 GMT 10
I am sorry that I have been gone for a few days. Been working more on the software side. I found a few major problems with the feralA decoding, one is that the MS matrix that I was using had a bias on encoding vs. decoding. Suffice to say that the M and S channels had equal and opposite calibration errors. Most of the time, the effect was minor, but it did effectively make the decoding a little less accurate than should have been. Alex has some newer examples if you want to listen, but the only recordings 100% updated at the point of the posting is 'Rumors'.
Also attached is a block diagram for a typical feralA decode, of course one item in the block diagram is some kind of accurate DolbyA decoding means...
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 Feb 1, 2020 6:30:34 GMT 10
Hi John Many thanks for the update. If any member wishes to hear samples of John's corrected version of Fleetwood Mac- Rumours, please send me a PM
Alex
|
|
|
Post by ROWUK on Mar 5, 2020 19:15:08 GMT 10
So, due to a medical issue, I am spending this week at home and have time to listen more seriously than in December/January/February where concerts got in the way. There are several things that are interesting to me.
1) In the corrected recordings the solo voices NEVER move forward in the stereo playback geometry, they all recede some - except for Karen Carpenter whose voice gets a bit less reverberant when compared to my original CD "Carpenters Their Greatest Hits". ONJ and Carly Simons voices are moved too far back which I attribute to excessively applied reverb in the studio. 2) In all the corrected recordings, everything below 200Hz improves remarkably. Bass drum and electric bass are much better detailed in respect to pitch and length of the note. 3) in all recordings there is far more "texture" in the midrange, voices are certainly more "palpable", left to right/front to back effects are much greater. 4) In all of the recordings the original applied reverb is much more natural - the sound stage is also much larger, extending beyond the speakers - but only for specific instruments. Solo voices are nailed down in the stereo perspective and easily localized. 5) I am absolutely amazed at the dynamics and colors present in the recording. The recording engineers work was not as primitive as we often hear. Why the audiophile press did not run amok all these years shows how deaf that they are 6) FadeOuts - I notice now for the first time how inconsistent that the engineers faded out - often very unevenly 7) I notice a very strong "adjacent room effect". This is an effect that happens when the playback is in one room and listening in the next - with an open door between. Some recordings (Johns) sound like they are being played live but being listened to through a window. None of the originals have this effect except the Carpenters Their Greatest Hits. ALL OF JOHNS RECORDINGS DO! 8) Each of the recordings now has an obvious optimal playback volume in my room. Slightly too loud or too soft, then it sounds unnatural. This made me turn up some tracks and turn down others
Individual recordings: The Tijuana Brass was a great disappointment - not because of the processing, rather because now I hear how "weakly" the music was played. Up until now it was happy mush, now it is very clear how simple the recordings and primitive the effects were.
Fleetwood Mac (recording from 30 Jan 20)- some minor mistracking on the female solo vocals. Bass - unbelievable bass! Guitars and voices have the right size - on the LP their mouths extend between both speakers - very unnatural when the voice is not localizable. John fixed that somehow!
Carpenters - sibilance comes and goes. This is certainly not an artifact of the decoding as it is the same on the LPs but on the CD there is ALWAYS sibilance. The decoded versions make this music so much more enjoyable - even if Karens voice is sometimes not as sexy.
Carly Simon - vocals wander depending on the track from 1 meter behind the other instruments to WAAAAAAAY back. They are consistent inside each track - the voice does not wander in the cut. The voice is never too low in volume, just unnaturally behind the band.
Olivia Newton John - her Barbie doll voice is not my cup of tea. It is like she never gets above "mezzopiano". Sometimes it would be great to have a real womans voice! The decoded recordings do draw me in because the band is so fantastic. I never noticed this highest quality playing before. The arrangements are great. Listening is like reading the score.
Linda Ronstadt - "It's so easy" and Tumbling Dice have the most recessed voices of all of the recordings. Carmelita is "a bit" further forward. I question this rendering however as the solo voice changes timbre based on volume - gets brighter when louder - more than a voice does naturally. "Sorrow lives here" is WONDERFUL! The piano tone is consistent throughout, the voice of a real woman in real space. "I will never marry", "Blue Bayou", Maybe I'm right", Old Paint is like Carmelita with LRs timbre changing with volume.
I will get to Supertramp later this week. I have another recording to spend some time with today - a new cover of the band Chicago performed by Russians. The group is called Leonid and Friends and the playing is incredible!
In any case, John, this is a TREMENDOUS effort with INCREDIBLE results. You have unlocked a door to this music and I certainly hope that a venue will be created to get this goodness out to a large group of customers.
|
|
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 Mar 6, 2020 13:51:06 GMT 10
So, due to a medical issue, I am spending this week at home and have time to listen more seriously than in December/January/February where concerts got in the way. There are several things that are interesting to me. 1) In the corrected recordings the solo voices NEVER move forward in the stereo playback geometry, they all recede some - except for Karen Carpenter whose voice gets a bit less reverberant when compared to my original CD "Carpenters Their Greatest Hits". ONJ and Carly Simons voices are moved too far back which I attribute to excessively applied reverb in the studio. 2) In all the corrected recordings, everything below 200Hz improves remarkably. Bass drum and electric bass are much better detailed in respect to pitch and length of the note. 3) in all recordings there is far more "texture" in the midrange, voices are certainly more "palpable", left to right/front to back effects are much greater. 4) In all of the recordings the original applied reverb is much more natural - the sound stage is also much larger, extending beyond the speakers - but only for specific instruments. Solo voices are nailed down in the stereo perspective and easily localized. 5) I am absolutely amazed at the dynamics and colors present in the recording. The recording engineers work was not as primitive as we often hear. Why the audiophile press did not run amok all these years shows how deaf that they are 6) FadeOuts - I notice now for the first time how inconsistent that the engineers faded out - often very unevenly 7) I notice a very strong "adjacent room effect". This is an effect that happens when the playback is in one room and listening in the next - with an open door between. Some recordings (Johns) sound like they are being played live but being listened to through a window. None of the originals have this effect except the Carpenters Their Greatest Hits. ALL OF JOHNS RECORDINGS DO! 8) Each of the recordings now has an obvious optimal playback volume in my room. Slightly too loud or too soft, then it sounds unnatural. This made me turn up some tracks and turn down others Individual recordings: The Tijuana Brass was a great disappointment - not because of the processing, rather because now I hear how "weakly" the music was played. Up until now it was happy mush, now it is very clear how simple the recordings and primitive the effects were. Fleetwood Mac (recording from 30 Jan 20)- some minor mistracking on the female solo vocals. Bass - unbelievable bass! Guitars and voices have the right size - on the LP their mouths extend between both speakers - very unnatural when the voice is not localizable. John fixed that somehow! Carpenters - sibilance comes and goes. This is certainly not an artifact of the decoding as it is the same on the LPs but on the CD there is ALWAYS sibilance. The decoded versions make this music so much more enjoyable - even if Karens voice is sometimes not as sexy. Carly Simon - vocals wander depending on the track from 1 meter behind the other instruments to WAAAAAAAY back. They are consistent inside each track - the voice does not wander in the cut. The voice is never too low in volume, just unnaturally behind the band. Olivia Newton John - her Barbie doll voice is not my cup of tea. It is like she never gets above "mezzopiano". Sometimes it would be great to have a real womans voice! The decoded recordings do draw me in because the band is so fantastic. I never noticed this highest quality playing before. The arrangements are great. Listening is like reading the score. Linda Ronstadt - "It's so easy" and Tumbling Dice have the most recessed voices of all of the recordings. Carmelita is "a bit" further forward. I question this rendering however as the solo voice changes timbre based on volume - gets brighter when louder - more than a voice does naturally. "Sorrow lives here" is WONDERFUL! The piano tone is consistent throughout, the voice of a real woman in real space. "I will never marry", "Blue Bayou", Maybe I'm right", Old Paint is like Carmelita with LRs timbre changing with volume. I will get to Supertramp later this week. I have another recording to spend some time with today - a new cover of the band Chicago performed by Russians. The group is called Leonid and Friends and the playing is incredible! In any case, John, this is a TREMENDOUS effort with INCREDIBLE results. You have unlocked a door to this music and I certainly hope that a venue will be created to get this goodness out to a large group of customers. Robin Thanks for your indepth listening analysis that confirms that John is on the correct path. Unfortunately, some of the industry are pissed off with John for exposing sloppy practices that in effect deprived customers of their right to hear the highest possible quality on the CD format.
Kind Regards Alex
|
|
|
Post by johndyson on Mar 7, 2020 8:37:22 GMT 10
I have been very slow in responding -- but have lots of things 'working on me' right now. However, I do have some answers for Robins' kind reviews!!!
About Tijuana Brass -- I do worry that I might have botched something. It feels like I should be able to get better sound out of it -- it is still one of the 'background worries'. For different reasons, the ABBA albums also 'worry' me.
The Carpenters are very different from album to album. Also, I finally latched on to some actual vinyl rips... Some of the songs appear to be total remixes on the LP vs. digital. Top of the world is especially egregious. RIght now, one of my 'problem child' albums is the 1970 Carpenters. They did some things differently than EVERY other copy of FA material, and I am JUST THIS MINUTE trying to reverse engineer it. It is NOT important by itself, but a general solution is important.
I have been botching the stereo image because the FA process has a varying amount of stereo image manipulation. You know how fast compression can distort a stereo image? That is exactly the kind of thing that they try to counteract, but then the amount of stereo image maniuplation does have some apparent standardization, but I have been making mistakes. As you very obviously noticed, I tend to err on the side of overly widening the image. That is ALL my fault and my lack of mastering type skills.
As I have mentioned to Alex -- my goals do not include any kind of mastering. The goal is to present as close to a purely decoded DolbyA master tape as can be derived from the materials. As soon as I get into 'mastering' the material, other than small amounts of LF or HF tilt/rolloff compensation, I have done the OPPOSITE of my goals. It is the listener who uses the decoder, or the mastering engineer who is trying to resurrect a consumer FA CD -- those are the individuals who are the best at mastering.
Sometimes, privately, I rant about avoiding any kind of 'mastering', probably to the point of being irrtating. Technically, doing a DA or FA decode would be a part of mastering, but that is the only part (other than small amounts of tilt compensation) that I allow myself. Beyond that, I am 'perverting' the recording, thereby undoing the progress that I (all of us) are trying to make.
I consider everyone who comments, helps, advocates as a productive part of this process. This problem NEEDS to be opened up, and the frustrated consumer who might be purchasing $10k worth of equipment, trying to improve the sound -- those customers are already behind the eight ball because of the recording quality.
My goal: fix and recover the music... Simple as that. I am part of the process, the DA/FA decoder is also part of the process. All of us, those who are looking at this thing, trying to figure it out, including Alex, Robin and the others -- we are working on the problem together.
There is ZERO profit motive here, If any money is changed hands, it might be small $300 PROFESSIONAL nusance fees, but I won't turn someone down who wants a copy. This especially applies to consumer use. I am considering disabling the license manager for the FA mode, therefore for most consumer and experimental professional uses, it will be free to use... PERIOD. I have to think about it, but that is a political matter for me, not my own motivation.
As it is right now, the planned 10Mar release will work with or without a licnese file if the --fa switch is used (FeralA.) This is ONLY about the music, it is only about the recordings.... PERIOD.
Thanks, Robin, Alex for your feedback and 'support'... I am working just this minute, right after hitting the 'post quick reply' button!!!
John
|
|
|
Post by cj66 on Mar 7, 2020 14:41:52 GMT 10
Still reading and paying attention with great interest John. I've just been very short on time during our busiest time of year. We're now slowing down so I will have some time to enjoy music again, maybe too much time with all the Covid madness going on!
I never cease to be amazed by the huge efforts you're putting in here for playback pleasure, only to be applauded and appreciated.
|
|
|
Post by johndyson on Mar 17, 2020 5:53:12 GMT 10
Hey -- got a new toy (or a better old one.) Just ran off the best ever 'Crime' using the latest decoder. Also, there is some stuff in 'Quiet' that has been revealed to me. If anyone is interested, I'll make them available -- the 'Crime' is 'velvet', and can hear details in 'Quiet' that were smeard over before. (They aren't 'good' details, but show greater resolution in the signal.)
Also, just ran the ABBA 8 albums -- abba as you have never heard them. Of course, I cannot post the address of the small respository, but if interested -- private message me, and I'll make them available also.
The decoder is really getting kind of easy to use.
Sincerely!! 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 Mar 20, 2020 8:14:19 GMT 10
John now has a further improved version of Supertramp-Crime of the Century in 24/88.2 format. If anyone would like to hear a sample please send me a PM
Alex
|
|
|
Post by johndyson on Mar 20, 2020 21:01:09 GMT 10
John now has a further improved version of Supertramp-Crime of the Century in 24/88.2 format. If anyone would like to hear a sample please send me a PM Alex Most frustratingly, I might have figured another/better set of decoding parameters. This stuff is sometimes tricky!!! I'll make sure that Alex gets it ASAP.
When undoing the EQ, how much is too much, and how little is too little?
I am working on it right now.
John
|
|
|
Post by ROWUK on Mar 22, 2020 0:40:02 GMT 10
Too much and too little is when we have audible compression/decompression artifacts. I am pretty sure that the sweet spot is pretty big where we do not have the artifacts, but some leeway for EQ. That seems to be the toughest - what is "better" and what is only different? I need a relatively long period of time to sort stuff like this out as I have to listen with various moods and times of day. When I listen to these cuts in "analysis mode", I am primarily listening for artifacts. When they pass this test, then I leave analysis mode and let the general sound "acclimate". That means listening for enjoyment and letting "things" modify or disturb perception. John, you are simply kicking Ass with this. John now has a further improved version of Supertramp-Crime of the Century in 24/88.2 format. If anyone would like to hear a sample please send me a PM Alex Most frustratingly, I might have figured another/better set of decoding parameters. This stuff is sometimes tricky!!! I'll make sure that Alex gets it ASAP.
When undoing the EQ, how much is too much, and how little is too little?
I am working on it right now.
John
|
|