'FeralA' decoder -- free-to-use

[jabbr]

--fcs="2,-46,fcx=G+*" --pif9k=-2 is also really really good

 

The "tinkles" in "The Lamb Lies Down on Broadway" come in at 13 seconds

The vocals in "Carpet Crawlers" are clearer and more immediate

The background tape hiss in "The Waiting Room" is gone, and the HF percussion is very clean

[John Dyson]

4 hours ago, jabbr said:

--fcs="2,-46,fcx=G+*" --pif9k=-2 is also really really good

 

The "tinkles" in "The Lamb Lies Down on Broadway" come in at 13 seconds

The vocals in "Carpet Crawlers" are clearer and more immediate

The background tape hiss in "The Waiting Room" is gone, and the HF percussion is very clean

Good, I am glad that it is working better.

 

The decoder, as of the V1.6.6P release, the decoding algorithms arent getting modified unless there is a serious bug.   The EQ is corrected again (I actually fixed a bug in the P release WRT EQ), and  I really do not want to break anything by 'fixing' something else.

 

There are a lot of adjustments available, but most of the time only a few --pi* type adjustments are needed, and the arguments to fcx (like fcx=G) need to be changed from time to time.

 

You'll notice that a lot of the commodity things that I enjoy listening to (Carpenters, ABBA, Warwick, Bread, etc), all need a very similar EQ.   Carpenters albums vary, but all seem to have a similar formula.

 

The trickiest thing is the 'classical' mode.   Don't be prejudiced about the name 'classical', but it is basically a mode that changes how the stereo image is modified.    Much of the time, if you think that you need an extra '+' or '*' on the argument to 'fcx=G' or 'fcd=G', it just might be you need 'classical'.

 

When 'classical' is needed, but not used -- one side effect is that instruments and vocals loose some of their spatial locality, but that often is difficult to discern.  The other side-defect is that it feels like the highs are slightly blunted.   If 'classical' is needed, and there were otherwise 'blunted' highs, then magically they might become more distinct.   The 'classical' mode is only one item in our quiver of tools.

 

The most common usages appear to be:   fcd=G, fcx=G, fcd=G+, fcx=G+, fcd=G+*, fcx=G+*, and then possibly add classical mode.   I have a few recordings that want 'fcd=classical' (no G.)    Sometimes, if you think if a recording needs a '+' and/or '*', then given 'classical' a try instead -- it might clear things up more cleanly, but might not.

 

For the --pi switches, then the (--pif3k=-1, --pif6k=-1, --pif9k=-1) appear to be most common on material that I listen to.   Sometimes the --pif6k needs to be skipped, and sometimes --pif12k needs to be added.   Usually, try starting with the 'f' version of the EQ (which means full) like --pif3k=-1.   If you need a little more brightness, then try using the 'x' variant (which means 'xtra' instead 'full'.   'full' is the complete EQ curve, but 'x' only goes to 21.5kHz.)   Without 'x' or 'f', the top freq depends on the base freq. 

 

Without 'x' or 'f', the 3k,4.5k,6k EQ stops at 9kHz,  everything else (9k,12k,15k) stops at 18k.   If you want 6k to stop at 18k, there is no direct command, but if you do --pi6k=-1, then --pi9k=-1, then the two concatenated EQ:  6k->9k + 9k->18k will give you the same as 6k->18k.

 

John

 

[John Dyson]

For those of you who have good copies (usually Sweedish Polar or European Polydor, but sometimes Discomat) of ABBA albums, I have a foolproof formula for decoding them and getting as clean, bright and accurate copies that is possible.   Some of the non-pristine disks might need skipping the 'classical' switch.   Some non-pristine copies have had an extra DolbyA encode/decode cycle, therefore a slight loss of quality -- but generally everyone that I have tested is vastly improved by decoding...

 

Warning -- these settings are not 'finalized' for listening given certain taste -- some might want some bass boost, slight midrange boost (like this:  --pe45=2,0.75 --pe1k=4,0.75), some might want a slight additional treble cut, resonance -- --pe9k=1,-1.5,1.0, or resonance with less cut: --pe9k=K,0.75 --pe9k=1,-1.5,1.0.   There is also the option for a non-resonant cut at 9k like: --pe9k=K,-1.5.  Personally, I prefer the slight MF/midrange boost, and the full -1.5dB K mode cut at 9kHz a;pmg with a -0.75dB '4' mode cut at 12kHz, but tastes vary.   The 'K' mode is good for getting rid of the excessive edginess on some of the vocals...   Sometimes adding an additional slight cut at 12kHz of about 1/2 of the amount -- maybe 4 mode instead of K mode can be better  (It depends if there are one of the slightly overly bright vocals or not.)

* 'K' mode and '4' mode are special mixes of 2nd order filters that are good at cancelling out the negative effects of any one Q value.

 

Important note:  Use the 2nd order EQ (the --pe switches) for getting the best personal 'sound', but the 1st order EQ (--pi switches) for proper post-decoding EQ.

 

The decoding is as close to perfect as I have ever heard (no excess IMD, etc.)   Even SuperTrouper (the song on the SuperTrouper album) is pretty darned good.

 

 --fcs="6,-50,fcx=classical,gG" --fcs="2,-70,fcx=classical,gG" --pi4p5k=-1 --pif9k=-1

 

Note that this is as close to a full decode that I can reliably get.   I believe that adding '--fcs="1,-80,fcx=classical,gG"' would be a totally full decode, but there can sometimes be troubles on certain albums.  Using only the first '--fcs' command comes close, but leaves some lingering compressions at low levels -- giving a more edgy sound.

 

Note also -- the use of the new 4.5k 1st order EQ.   It was the key to get the correct balance on certain albums, including ABBA.

 

John

[John Dyson]

1 hour ago, John Dyson said:

For those of you who have good copies (usually Sweedish Polar or European Polydor, but sometimes Discomat) of ABBA albums, I have a foolproof formula for decoding them and getting as clean, bright and accurate copies that is possible.   Some of the non-pristine disks might need skipping the 'classical' switch.   Some non-pristine copies have had an extra DolbyA encode/decode cycle, therefore a slight loss of quality -- but generally everyone that I have tested is vastly improved by decoding...

 

Warning -- these settings are not 'finalized' for listening given certain taste -- some might want some bass boost, slight midrange boost (like this:  --pe45=2,0.75 --pe1k=4,0.75), some might want a slight additional treble cut, resonance -- --pe9k=1,-1.5,1.0, or resonance with less cut: --pe9k=K,0.75 --pe9k=1,-1.5,1.0.   There is also the option for a non-resonant cut at 9k like: --pe9k=K,-1.5.  Personally, I prefer the slight MF/midrange boost, and the full -1.5dB K mode cut at 9kHz a;pmg with a -0.75dB '4' mode cut at 12kHz, but tastes vary.   The 'K' mode is good for getting rid of the excessive edginess on some of the vocals...   Sometimes adding an additional slight cut at 12kHz of about 1/2 of the amount -- maybe 4 mode instead of K mode can be better  (It depends if there are one of the slightly overly bright vocals or not.)

* 'K' mode and '4' mode are special mixes of 2nd order filters that are good at cancelling out the negative effects of any one Q value.

 

Important note:  Use the 2nd order EQ (the --pe switches) for getting the best personal 'sound', but the 1st order EQ (--pi switches) for proper post-decoding EQ.

 

The decoding is as close to perfect as I have ever heard (no excess IMD, etc.)   Even SuperTrouper (the song on the SuperTrouper album) is pretty darned good.

 

 --fcs="6,-50,fcx=classical,gG" --fcs="2,-70,fcx=classical,gG" --pi4p5k=-1 --pif9k=-1

 

Note that this is as close to a full decode that I can reliably get.   I believe that adding '--fcs="1,-80,fcx=classical,gG"' would be a totally full decode, but there can sometimes be troubles on certain albums.  Using only the first '--fcs' command comes close, but leaves some lingering compressions at low levels -- giving a more edgy sound.

 

Note also -- the use of the new 4.5k 1st order EQ.   It was the key to get the correct balance on certain albums, including ABBA.

 

John

I thought that I'd supply the exact CDs that I was having great success using.

One item of note:  Sometimes the best sounding CD when directly playing isn't the best version to decode.

Sometimes manipulated CDs sound better, but that isn't what you want to start with to decode!!!

You want the closest to original tape as you can get.  There is notable generation loss even on copied tapes,  if they go through an extra encode/decode cycles.

It is almost random whether they did a full encode/decode or a raw tape copy (which is preferable):

 

These are the most clean versions, least compression, least manipulated that I have found so far

 

ABBA                           "PolydorKK P33P 20056"
Arrival                          "Polar POLCD 272"
RingRing                      "Polydor - 843 642-2"
SuperTrouper              "Polydor POLCD 322"
TheAlbum                    "Polar POLCD 282"
TheVisitors                  "Polar  POLCD 342"
VoulezVous                 "Polar POLCD 292"
Waterloo                     "Polydor - 843 642-2"

 

John

 

[bogi]

19 hours ago, jabbr said:

--fcs="2,-46,fcx=G+*" --pif9k=-2 is also really really good

 

The "tinkles" in "The Lamb Lies Down on Broadway" come in at 13 seconds

The vocals in "Carpet Crawlers" are clearer and more immediate

The background tape hiss in "The Waiting Room" is gone, and the HF percussion is very clean

Hi jabbr, which release of "The Lamb Lies Down on Broadway" it is about? This one (1986) or other?

https://www.discogs.com/Genesis-The-Lamb-Lies-Down-On-Broadway/release/5021076

I like the old Genesis, maybe I could try it too.

[sandyk]

4 hours ago, bogi said:

Hi jabbr, which release of "The Lamb Lies Down on Broadway" it is about? This one (1986) or other?

https://www.discogs.com/Genesis-The-Lamb-Lies-Down-On-Broadway/release/5021076

I like the old Genesis, maybe I could try it too.

I even located some flac files from Genesis - The Lamb Lies D ...  Disc 2 Japan  VJCP-68097 1974

although I only listened to the title track.

[jabbr]

6 hours ago, bogi said:

Hi jabbr, which release of "The Lamb Lies Down on Broadway" it is about? This one (1986) or other?

https://www.discogs.com/Genesis-The-Lamb-Lies-Down-On-Broadway/release/5021076

I like the old Genesis, maybe I could try it too.

VJCP-68096-097 "Japanese Paper Sleeve Remasters"

 

I find that at 3 layers the HF detail is stripped out. I am not sure I have the post decode equalization perfect yet.

[John Dyson]

2 hours ago, jabbr said:

VJCP-68096-097 "Japanese Paper Sleeve Remasters"

 

I find that at 3 layers the HF detail is stripped out. I am not sure I have the post decode equalization perfect yet.

It can be really tricky on certain material to get great results -- there is even the possibility of

1) a missing capability in the decoder.

2) it just might only have two layers in the recording.

3) the recording might have been manipulated after FA encoding (or might not be using the apparently standard scheme)

 

and, maybe a fourth --

compression, when not excessive, isn't always bad.  One or two layers of lingering FA encoding (esp only one) doesn't always sound bad!!!

I believe that the 'badness' of FA encoding starts appearing when there are multiple layers (compressors) interacting then producing a ragged, ratty sound.   When I knowingly leave one or two layers in a recording, the result can be improved by removing the others, and a little bit of non-interacting compression isn't terrible.

 

There is a big difference between plausible results for ones own personal listening and a true remaster.   Even though I might sometimes have low opinion of those 'professionals' producing the mush on the CDs and downloads that we buy,  actual 'mastering' a recording is a skill.   There are some genres of material that I feel pretty comfortable that I can do a really good job finding the correct settings -- but other music that I had never listened to, I screw up royally.

 

The decoder technology (someday) needs to be made even easier to use.   The few 'settings' are starting to be reasonable for semi-technical and technical people to use, but still can be tricky.  It is easy to chase one rabbit after another down rabbit holes -- I do suggest that unless one is really having fun, the project of trying to decode material can be counter productive to the goal of improved enjoyment.

 

My own decoding efforts are focused on *learning* how to inform others on using the decoder.   This is a LOT to learn, and my enjoyment of music is greatly diminished because of the testing, listening - try to learn to explain things and document them.

 

The decoder works very well now, but just might (as I mentioned above) be missing a capability or two.   I'll probably give 'Lamb' a try again in a day or so -- maybe to figure out some plausible settings, or more important, see if the decoder needs a new capability that I hadn't thought of.

 

Most importantly -- people using the decoder should try to ENJOY whatever improvements that come reasonably easily!!!

Also, some day (hopefully soon), we'll restart the idea that @lucretiuscame up with -- a list of successful decoding information.

 

I am 150% overloaded trying to finish the project, so I'll only be sporadically adding some decoding results, and sometimes even correcting/improving them.   My guess is that I'll be somewhat freed up in about 2months, and I am not being aggressive...  The decoder mechanics are truly perfect -- I have run tests and gotten fantabulous results that I never really expected to work so very well.   There are simply logistics keeping me busy, and every delay is possibly turning off another potentially interersted user...

 

Really good news:  a person who does recordings from time to time, who HATES this project, finally listened again to the results after maybe a year or two -- he said -- the results don't sound as bad (paraphrased), and pretty much shut up about his technical criticism.  Now he (miscreant industry person)  says that no-one wants the better quality anyway (paraphrased.)   I think that digusting attitude contributes to the reason that we are getting MUSH almost every time we do a download or purchase media for the music that we love.

 

 

John

 

[jabbr]

24 minutes ago, John Dyson said:

and, maybe a fourth --

compression, when not excessive, isn't always bad.  One or two layers of lingering FA encoding (esp only one) doesn't always sound bad!!!

I believe that the 'badness' of FA encoding starts appearing when there are multiple layers (compressors) interacting then producing a ragged, ratty sound.   When I knowingly leave one or two layers in a recording, the result can be improved by removing the others, and a little bit of non-interacting compression isn't terrible.

 

It’s one thing to remove compression and another to remove sounds.

 

Like Photoshop, it’s one thing to remove noise and another to remove wrinkles.

 

Every actual sound b the recording must be preserved.

[John Dyson]

10 minutes ago, jabbr said:

It’s one thing to remove compression and another to remove sounds.

 

Like Photoshop, it’s one thing to remove noise and another to remove wrinkles.

 

Every actual sound b the recording must be preserved.

I am starting to believe that the 'Lamb' recording might have only a few layers.   However, I will be looking at it again so also - just to help...  I'll communicate directly with you, and then you can determine if there is any improvement...   'Lamb' is your project :-), but I'll try again to see what I can do -- might even need something additional in the decoder.   However, again -- as it is -- there just might only be two layers in it, maybe even one.

 

Sometimes one can try to decode one layer too much, and the results aren't always terrible either.   Expansion of +-5dB to +-10dB at low levels, when not actually decoding, is sometimes a barely noticeable bit of damage.

 

John

 

[lucretius]

13 minutes ago, jabbr said:

Every actual sound b the recording must be preserved.

 

Perhaps John will correct me if I'm wrong, but I don't think that's possible.  That is, "good" sound sharing the same frequency as "bad" sound (to be corrected), will be somewhat affected.

[jabbr]

9 minutes ago, lucretius said:

 

Perhaps John will correct me if I'm wrong, but I don't think that's possible.  That is, "good" sound sharing the same frequency as "bad" sound (to be corrected), will be somewhat affected.

Signal recovery amidst noise is another topic. 

[sandyk]

10 minutes ago, lucretius said:

 

Perhaps John will correct me if I'm wrong, but I don't think that's possible.  That is, "good" sound sharing the same frequency as "bad" sound (to be corrected), will be somewhat affected.

 That seems highly likely. Sometimes you will need to tolerate a bit of the bad sound to preserve most of the good sound though .

[John Dyson]

Here is an anecdote about my own tilting at the ABBA windmill -- I seems that ABBA sometimes used a variant of FA for producing their recordings in the studio.  So, when I try to find the correct number of layers, sometimes the mixed version has FA in it.   It is very easy to chase ABBA recordings too far, and try to decode too many layers.

 

The 'Carpenters' recordings have been successfully decoded, and even the Linda Ronstadt recordings are coming out VERY GOOD.   However, ABBA has kept on giving me troubles.   I sometimes feel like I am getting close, and even succeed...  Then, I listen to the entire album, and one or two recordings is over-decoded.

 

Before I have to move-on and get some more useful things done, I am going to punt, and limit my # layers to 6 on ABBA, and sometimes 5 layers...   The most egregious is the song 'Waterloo' -- it appears to have been mixed with several layers of  FA (or something similar.)

 

There is definitely something odd going on -- no trouble anymore with most recordings, but there are still some tricky ones like the albums that @jabbris working on.  I might just be lucky (except for ABBA) that the music that I listen to is easiest to process?

 

John

 

[Skip Pack]

It's also worth noting that each of our environments adds a great deal of complexity to our attempts to understand what we confront here. I have been primarily using digital source material from files, initally via Squeezebox, since late 2005. My system, mostly DIY, has evolved and become more or less improved to suit my old (now 74) ears. I am generally focused somewhat away from the material John has used for development and testing, but I'm a fairly eclectic music lover.

 

My question here is, 'Have I developed my current system to ameliorate the very issues that John is addesing in NHRDS?'

 

I will continue to use versions to see what happens and to develop my understanding, because I think that there are a lot of tools contained in NHRDS that an expert can use selectively to get good results.

 

Skip

 

System:  Moode Audio on RPi4/Shanti PS -> Soekris Dac1321 -> line level passive filter network for Open Baffle contouring -> Neruochrome That Driver Preamp SE to BAL -> Neurchorme Mod-86 power amp -> Altec 605A drivers in open baffles with 18" woofers in H-frame boxes

[sandyk]

16 minutes ago, Skip Pack said:

My question here is, 'Have I developed my current system to ameliorate the very issues that John is addesing in NHRDS?'

 I.M.E. (and one and a half as well), you can markedly reduce noise and distortion, and reduce sibilance a little ,as well as markedly improve the soundstage by fixing up most of the Earth issues in a typical desktop PC, but you can't get rid of the damage done by dollops of Feral Dolby-A . 
https://audiophilestyle.com/forums/topic/60381-hdd-to-case-bonding-uptick-in-sq/?tab=comments#comment-1084599
 I am utterly convinced that John's current S/W is also removing some of the distortion products generated by the actual PC used to rip and save the files.

 

[John Dyson]

21 minutes ago, sandyk said:

 I.M.E. (and one and a half as well), you can markedly reduce noise and distortion, and reduce sibilance a little ,as well as markedly improve the soundstage by fixing up most of the Earth issues in a typical desktop PC, but you can't get rid of the damage done by dollops of Feral Dolby-A . 
https://audiophilestyle.com/forums/topic/60381-hdd-to-case-bonding-uptick-in-sq/?tab=comments#comment-1084599
 I am utterly convinced that John's current S/W is also removing some of the distortion products generated by the actual PC used to rip and save the files.

 

FA acts like an NR system to some extent -- actually more powerful than DolbyA itself.   So, certain disruptions in the signal (mostly noise type things) can be somewhat suppressed by FA.   When looking at the reverse engineered design of FA, my guess is about 20dB of NR can be provided.   SO SOME impairments can be improved -- then, look at the impairments that give real troubles:

 

On the other hand, frequency response changes are deadly.  It only requires about a 1dB deviation (or less) to make decoding the material very tricky.   This MIGHT be part of the problem that @jabbrcould be seeing.   It only takes small deviations of certain parameters to make a mess of decoding.  Something might be able to be 'decoded' to 4 layers, might indeed have troubles going beyond 2 layers.

 

Level variations, where the CD was normalized by the distributor -- per track normalization is worse than normalizing the entire CD/digital download, are also very troublesome.   Given straight, non-normalized recordings, the calibration levels are seldom outside of -50, -49, -46, -44.5dB.   As soon as the signal is normalized (level changed from when encoded), then all bets are off on the normal numbers.

 

This tells me that it is likely the FA encoding happens very close to creating the master using for pressing, creation of CD, etc.   This also says, don't be disappointed -- *TADA* any FA signal has likely been in the analog world!!!  I don't know of anything but Analog DolbyA HW that can produce the accurate DolbyA encoded signals that I am seeing.

 

This whole thing disgusts me -- because we have effectively been lied to about the quality of some recordings.  My position is that of supporting my friends who are customers of the industry, the decoder is only a side effect of my quest to get good copies of the recordings.

 

John

 

[John Dyson]

I think that I have figured out how to REALLY distribute high-res and corrected recordings.   There are special rules where a distributor who sells a modified copy of a recording, then that modified copy has to be available for others to sell.   There are archane rules like that.

 

However, think about this:  legally sell licensed copies like HDtracks does, but additionally give away a copy of a special version of the decoder along with encrypted 'best results' decoding parameters.   The end user can then use their simple, fuzzy, HDtracks version, or have access to the improved versions by locally decoding the material?   Since the copy of the recording is not modified, then there is no loss because of the requirement to give away the improved version.

 

Instead of giving away a vanilla copy of the decoder like I am doing here, and I always intend to make it available -- source code sometime in the future.  The 'music distributor' version would give away a copy of the decoder  that accepts an encrypted version of the decoding parameters, and place the 'distributor' version under a punitive license.  Even the potentially 'derived works' decoding parameters don't need to be put under tight licensing, so there is no encumberance there either...

 

This would be a way of 'freeing the quality' in the recordings while someone can make a business of it.   (I don't do business -- I intend that someday, someone is allowed to make money from the decoder, just with personal attribution.)

 

Would this work legally?   Technically, I know it would...   Would people go to the trouble of using an automatic procedure on Windows or whever to get a better copy?

 

John

 

[sandyk]

8 hours ago, jabbr said:

VJCP-68096-097 "Japanese Paper Sleeve Remasters"

 

The .flac folder of files that I found was mislabelled by the OP. 

 The cover photo shows that it is from VJCP-68096-097

[John Dyson]

I am learning how to use the decoder -- like everyone else here...   The difference might be that there are some things that aren't well documented...   Trying to learn what is needed, so I can better document it...

 

Anyway -- have a tidbit...   If you are doing decoding operations, and the sound ends up a bit 'thin'.   I wanted to remind of the --pi1k switch.   I added it for a reason, and wasn't sure if/when it would be needed ( I understand a lot of what is going on, and surmised that it might be needed.)

 

The pi1k switch really improved the Rumors and Fleetwood Mac album with Rhiannon on it.     Trying to remove the thin-ness by doing something like --pe1k=4,0.75 or --pe750=4,1.5 or somesuch sometimes is not quite right.   When it is useful, the --pi1k=1 switch is just right.

 

(Note:  the pi1k switch is a 3dB 1st order bass boost starting at 1kHz,  the  --pe series of switches are 2nd order switches with Q values or mixed Q values.   On the --pi series, there is no 'Q' associated with it.)

 

John

 

[John Dyson]

Minor update/bugfix release:  V1.6.6R, truly a minor release.

https://www.dropbox.com/sh/1srzzih0qoi1k4l/AAAMNIQ47AzBe1TubxJutJADa?dl=0

 

I found that the pi1k, pi750, pi500 are more useful than I had originally estimated.

 

Improved/tweaked more correctly: --pi1k

Added: --pi500, --pi750

 

Each of these are 500Hz, 750Hz, 1kHz 3dB 1st order bass boost.

There is only 0,1,2,3 -- no bass cut for now.

 

John

 

 

 

 

[John Dyson]

Gang -- I wanted to forewarn you about an industry like pushback, esp by the culprits of the FA encoding.

Here is some ammo:

 

The compression is lined up on 10dB increments.   There are reasons for doing it that way for DolbyA, even if not DolbyA, but these compressors look just like DolbyA.   If you don't line up on the right calibration mark, then there are odd inconsistencies in the dynamics -- even though mostly not critical.  For best results, you gotta line up on one of the standard calibraton levels.   Note that the calibration levels are another data item in favor -- the base tape EQ is typically -13dB, and the typical highest calibration number is based on -1.5dB lower, -3dB lower, -6dB lower, and -9dB lower.   Sometimes, I have seen -10dB, but not sure.  Also, there are cases where the lowest level isn't based on -13dB but instead based on -23dB.

 

The compression attack/release is JUST LIKE DolbyA.   I can produce a special version of the decoder, with slightly different attack release -- the results always suck.  The attack/release has to match a DolbyA *perfectly* or rather than sometimes tricky to get rid of pumping and strange artifacts, instead it is IMPOSSIBLE if the specs don't match.

 

The gain curves are JUST LIKE DolbyA.   I mean, exactly.   If the gain curves are too different, then lining up the frequency response is impossible (low levels more dead than high signal levels), and sibilance becomes TOTALLY borked if the gain curves don't match.

 

There are a few other esoteric comments that I can make -- the above three items are enough to demonstrate the GREAT liklihood that the FA is DolbyA based.

It has taken YEARS to make the DolbyA accurate enough to run 8 times to do an FA decode.  Any small error will create shrill or dead results that cannot be tuned out.

 

Been getting pushbacks from know-it-alls, and those who benefit from keeping the MUSHED UP recordings the way that they are.   This makes the remastering business easier to do!!!

 

John

 

 

 

 

[John Dyson]

Some cool Fleetwood Mac examples/snippets:

I keep these up for about 3days.

 

Rhiannon:

Decode parameters: --fcs="4,-50,fcx=classical,G+"  --fcs="2,-60,fcx=classical,G+" --pif6k=-1 --pi9k=-1 --pif12k=-1

https://www.dropbox.com/s/yb1pwwzsfz3mgns/04 - Rhiannon-rawCD.flac?dl=0

https://www.dropbox.com/s/1ynvbex5wkkcx5t/04 - Rhiannon-DECODED.flac?dl=0

 

You Make Loving Fun:

Decode parameters:  --fcs="4,-50,fcx=classical,G+"  --fcs="2,-60,fcx=classical,G+" --pi500=1 --pix6k=-1 --pif9k=-1 --pif12k=-1 --pe9k=3,-0.75

https://www.dropbox.com/s/veg5jodmdzaq09q/Fleetwood Mac - Rumours - 08. You Make Loving Fun-rawCD.flac?dl=0

https://www.dropbox.com/s/x4b8jivc2a5o9y1/Fleetwood Mac - Rumours - 08. You Make Loving Fun-DECODED.flac?dl=0

 

The settings seem long, but notice that each set of decoding parameters is similar.  I have found this to be true on

many groups, where decoding is similar.   This is NOT true on the Carpenters and ABBA.

 

John

 

[John Dyson]

There is a new release coming, with more ease of use improvements, further narrowing down the settings for decoding parameters, and a minor fix for permissions on Windows systems.

 

I have been busy with a commercial version (still non-profit), and also further ongoing improvements.

 

I know that the decoder, even though it is easier to use than the earliest versions, it needs more ease-of-use refinement.   Been getting help from good people.

 

My guess -- about 1wk for the new release.

 

John

 

[John Dyson]

New version of the decoder is available...   There have been a few test releases, one even went to a major studio -- they are going to get this  one also.

 

A lot of 'automatic' stuff has been added -- it will be easier to get really great results.   Using the decoder is NOT trivial, but it is probably easier to learn to use than a new TV set or new cell phone.

 

The big problem has been historically that the EQ for decoding operations has been a 'science project'.   Science projects can take away from fun, even though a lot of audiophiles do get into science projects for trying to improve their systems.   Alas, the worst quality problem isnt' with their system, it is with the quality of the recordings that people have.

 

The decoder undoes the noxious compression used on consumer recordings -- this is NOT just a regular expander, but undoes the specific kind of compression often used on consumer recordings.

 

The program is a windows (or Linux) command line program.   Sorry about the command line thing, but I don't do GUI programming (well, at least since the late 1990s.)

Decoding normal recordings is now even simpler:

 

1) Choose needed calibration level, usually from -50, -46, -54.5, -60, -44.5, -49

(Initially, the differences don't sound that significant, eventually, you'll hear them)

2) Choose the decoding EQ (usually 'fcf', sometimes 'fce')

3) Choose the number of layers...  Recordings are usually made with between 4 to 8 layers, but if you try 4 layers at first, you'll probably be okay.

4) Choose post decoding EQ...

 

Does this seem like a lot of steps?   Well, once you are used to it, probably can make the correct choice in under 2-3minutes -- probably 20 seconds.   Shouldn' thave to run the decoder everytime listening to a recording, but run the decoder to produce your listening copy, so maybe 2-3 times for each of your favorite recordings to make that 'perfect copy'.   When you hear the improvement, you'll likely want to use it more often...

 

Initially, it is hard to hear the damage in consumer recordings -- once you hear the more clean versions, then the regular CDs really do start sounding ugly and compressed.  I am not saying that you SHOULD use the decoder, but it can be eye opening.  The bad sound in consumer digital recordings is the reason why I stopped my audiophile hobby back in late 1980s.  I finally admitted to myself that the CDs were NEVER going to sound good...

 

* Using the decoder might be like eating from the 'tree of knowledge' in the garden of Eden, you might end up realizing something that is best not to know -- most CD sound really bad.

 

====================================

First -- using the decoder is actually easier than the steps above imply -- typical classical can be done like this:

* these settings are almost always very similar, the --b switches are optional, compensates the recording

These settings were used for Andre Previn/London Symphony Rhapsody In Blue...  Changes for other classical

recordings are usually minor (if any.)

 

(full-out quality)

da-avx --infile=in.wav --overwrite --outfile=out.wav --tone=-50 --b15kxxk --b18k --fcs="8,auto,fce" --pi1k=1  --fw=classical

(simpler, still good results)

da-avx --infile=in.wav --overwrite --outfile=out.wav --tone=-50 --fcs="6,auto,fce"  --fw=classical

 

(In the above examples, I might be suggesting 'too many layers'.   The first example shows '8' layers, but 4,5,6 might be better to begin with!!!)

 

====================================================================================

Pop is a little more complex, but from recording to recording, the EQ is similar.  All you gotta do on the '--pi=' switches is move the numbers around until it sounds right.  There are almost 'standard settings', and the Usage guide talks about it a little bit, more docs to come.   However, doing the process can be pretty much fill-in-the-blanks based on the following:  (This would be Abbey Road, 2009 remasters) -- Many recordings use EXACTLY the same settings, but sometimes you gotta tweak:

 

 da-avx   --infile=in.wav --overwrite --outfile=out.wav --tone=-50  --fcs="6,auto,fcf"  --pi1k=1 --pi=v4 --pi=hf5

--pi=v3 or --pi=v4 are most common.   --pi=hf3 or --pi=hf5 are most common.

 

Doing the decoding operation is really very simple now.   The Usage doc isn't really enough to cover all cases, but might get you started.

Once you try a few times, maybe ask for help at worst, it starts to be easy to get good results.

 

Here are possible disappointments:

1)  Loss of ambience.   This can be caused by incorrect tone settings, too much decoding (too many layers), or perhaps you are used to the compression.

2)  Less bright sounding...   The FA compression DOES brighten recordings, but have you ever heard that gritty sound in person?  Real instruments ARE bright sounding, but not grimey and gritty like most consumer recordings are.  The decoder creates very 'bright' sounding results, but they are not as compressed, so sound different.

3)  Vocal sibilance -- sometimes the recordings are EQed before FA compression -- part of the EQ boosts sibilance.   There is a switch available: --as=++ (give one through 4 '+' signs to mitigate sibilance.  If you must, you can use the dynamic sibilance control, e.g. --as=++9 is a good starting point.)

4) The mix will sound different.  Different parts of a recording will have different levels.   Vocals are boosted, yet sibilance is decreased on FA recording.  Be ready for more natural vocal sounds.

 

Possible improvements:

1) For those without overwhelming tinnitus -- much less hiss on older recordings.

2) Much more stable stereo image (the consumer CDs tend to have very damaged stereo image, and wobbles around at micro-levels.)

3) Real bass instead of the strange compressed stuff.

4) Sounds more like a recording session instead of a bunched-up garbled thing like most consumer recordings.

 

The decoder is REALLY easy to use now -- usage and settings are a combination of technical correctness (which is easy to get now), and preferred sound character, which is extremely variable!!!

Current version:  V1.6.8C.

Current version for the Usage guide: V1.6.8C

The Usage guide is 100% up-to-date, but we do know some more tricks based upon experience.

There is also a Linux version.   The decoder works more naturally on Linux as it was written with a Unix sensibility, but does work okay on Windows also.

 

Just ask questions here (or message me privately) if you have troubles.   The decoder is as simple to use as possible for a command line program.

Repository: (Repo is kept up to date with latest versions -- I don't post every release in the forum, but the repository is up-to-date):

https://www.dropbox.com/sh/1srzzih0qoi1k4l/AAAMNIQ47AzBe1TubxJutJADa?dl=0

 

Also attached:  the Usage guide.   Admittedly inadequate, but can get you started if motivated.   Refer to the larger reference manual in the repository for installation, etc.   You won't need a 'License file' as mentioned in the full manual, the decoder doesn't need a 'License file' for what we are talking about here!!!

 

I haven't been watching the groups as much lately, been working on the decoder.   But, I'll be back around more often to answer questions soon.

 

Also, I'll provide some examples in the next few days -- INFINITELY better than before, and with the last change in the last day or so, even better yet.   You won't believe how badly damaged most recordings are -- and there is a LOT of room for improvement...   The DHNRDS FA decoder is NOT perfect, but will often produce suprisingly better results than the originals.

 

John

 

 

 

 

 

Usage-V1.6.8C.pdf Dyson-Hess_NRDS_Manual_v0-7-1a-p.pdf da-V1.6.8C-Linux.txz da-release1.6.8C-win-2020-11-12.zip