|
Post by jsdyson on Sept 4, 2018 16:06:28 GMT
I improved a few of the demos -- firstly, I found a better source for Dreamworld. The new one that I uploaded appears to be one generation less of DolbyA encoding/decoding. The previous one appears to have not previously been decoded properly -- the DolbyA tone was probably incorrectly set. Now, it doesn't gate as badly when fading down, and also the sound is less muted. I have never heard a perfectly clear/clean version of Dreamworld. Some of the ugliness can be hidden by using lots of compresssion, but that is it's own kind of evil -- so I didn't do that kind of thing. Also, I didn't like how terribly overcompressed 'What About Livingstone' sounded, so I cleaned it up a little with my RMS dBlinear multiband uncompressor (expander). When listening, it does appear that they used a single band (or a small number of bands) compressor on the material -- so my multi-band didn't 100% match the compression, but came close and really restored some of the dynamics. On the repository, there is both the original DolbyA decoded example, and also an uncompressed (don't like to use the term 'expanded' when it is actually a recovery/correction) at a mild level. Add-on: I also uncompressed 'Thank You for the Music, Doris Day version'. It was also a bit too compressed, but I uncompressed it more mildly than What about Livingstone. Hope you enjoy the examples. THEY REALLY DO SOUND GOOD.
Repo: spaces.hightail.com/space/xghqJodgrjJohn PS: Info about the 'uncompressor': it isn't a typical expander. It dynamically calculates the attack/release time based upon the best choice for the sound quality. The expansion ratio is set, but isn't really the true 'expansion ratio'. It is dynamically calculated also for optimum sound quality. Also, the 'uncompressor' doesn't skew the frequency response like a normal aggressive expander would do -- the dynamics processing is dB linear, but has some aspects of pure linear at times. The dB linear/RMS is more of the style of a THATcorp/dbx style RMS scheme -- and in fact the same C++ classes are being used for my dbx type I and type II decoder. However, those C++ classes have much more logic in them for dynamic adaptation. The depth of expansion and the expansion ratio resists crazy or overly aggressive behavior, and even though I used rather mild settings on the expander (there is really only one dynamics setting -- everything else is dynamically calculated), much higher settings strongly resist creating artifacts -- the uncompressor is quite capable of expanding the material very aggressively without surging or pumping -- but too much expansion doesn't sound natural.
|
|
|
Post by jsdyson on Sept 2, 2018 13:06:28 GMT
Did a major rework of the examples. Firstly, the matrix for fixing the 90deg was still suboptimal, now I have an optimal 90deg conversion, and also, the EQ for the voices was all wrong in my previous examples -- so I brought the voices back up front. No major amount of re-EQ, but a few dB here and there can make a huge difference (for the vocal range, all it took to fix the voices was treble boost +1.5dB at 2.2kHz and +1.5dB at 3.3kHz.) Take a much prettier (yet) listen for the examples. Got rid of most of the stereo image problems left over in previous examples, and also (as I noted above) brought the voices out more clearly. I am still trying to choose a flac example (the clearest possible), but the mp3 examples aren't too bad. If you notice, even the start of SuperTrouper is clean and the loud chorus approx 64seconds, or louder at approx 80secs into the song doesn't devolve into a blob. Examples repository (MUCH IMPROVED)!!! ALSO ADDED A HIGH QUALITY FLAC FILE!!!spaces.hightail.com/space/xghqJodgrjThe source CD is ABBA Gold 1992, which is DolbyA encoded (the reason for the harsh sound with boosted highs.) ABBA should sound somewhat intense, but shouldn't sound as intense as that CD -- and it is that CD that encouraged me to write my DolbyA decoder. Subsequently, I found a lot of material in consumer hands is DolbyA encoded. It is almost like having access to a master tape as long as one has a DolbyA decoder!!! The bash shell script (using the sox audio tool) that fixes the 'fuzz' -- does the 90deg thing -- (only works well after DolbyA decoding): function rm15 { fn=$1 shift sox --combine=merge \ -v 0.50 "|sox -v 0.50 \"${fn}\" --type=wav --encoding=floating-point - remix 1v1.0,2v1.0 1v1.0,2v-1.0 sinc -n 4095 -21k " \ -v 0.3535 "|sox -v 0.50 \"${fn}\" --type=wav --encoding=floating-point - remix 1v1.0,2v1.0 1v1.0,2v-1.0 hilbert -n 4095 " \ -v -0.25 "|sox -v 0.50 \"${fn}\" --type=wav --encoding=floating-point - remix 1v1.0,2v1.0 1v1.0,2v-1.0 sinc -n 4095 -21k " \ --type wav - remix 1 4v1.0,6v1.0 remix 1v1.0,2v1.0 1v1.0,2v-1.0 rate -v 96k }
|
|
|
Post by jsdyson on Sept 1, 2018 17:52:59 GMT
Thank you so very much!!! I did try cleaning up my decodes a little more also -- I only did a 50/50 90deg/0deg before. The new decodes on the site (just done an hour or so ago) are full 90deg, and sound much better yet. The DolbyA project is being looked at recording pros, and JUST MIGHT have some interest from various major archive sites. It has taken a VERY LONG time to get the decoder working as well as it is. The biggest boost that I got is a recording pro who uses the DolbyA units all of the time, and we have decided to produce DBX I/II decoders and DolbyB/C also. We might do the C4 (most likely), and someday try the DolbySR (which is a major complex piece of equipment -- if DolbyA is tricky, then DolbySR crazy-tricky.) Do try the latest uploads (soon after the previous post) -- they are very further improved. Here is the repo again. I am RIGHT NOW hearing the ABBA snippets as clear as I ever have, and I use them for test material much of the time. Most of the examples are MP3, but I am going to figure out how to upload a good quality flac (if I can -- the mp3s actually lose quality now.) The flac versions are actually noticeably better in some cases. spaces.hightail.com/space/xghqJodgrjJohn
|
|
|
Post by jsdyson on Sept 1, 2018 12:37:57 GMT
I have been keeping some 'best possible' decoding examples of the 1992 ABBA Gold release on my repository. The location is below. Firstly, my DolbyA decoder has been 'beat into submission' to be better than a real DolbyA by myself and a recording engineer who deals with DolbyA units all of the time. We got either as-good-as or better results on by far most material with the SW decoder, and the SW decoder can do lots of things to grab pristine audio from the orginally DolbyA encoded material. So, this is the result of DolbyA decoding of 1992 ABBA Gold, plus phase shfiting the 'S' of M+S, which seems to really clear up the basic material to the level of the early vinyl releases (sans vinyl noise and freq response oddities of the vinyl process.) Along with the phase shift in M+S, the decoder goes to extreme/heroic measures to keep from damaging the audio (linear phase filters throughout, as the gain control is applied, the UNDESIRED intermodulation products are removed, etc.) The result on my repo cannot be any better than the master tape, but is better than any normal DolbyA can do. Please enjoy -- I know that I have made incremental progress in my projects, but this should truly be enjoyable... I did attempt processing the already damaged 'Dreamworld' copy that I have. It required uncommon settings to be able to play relatively cleanly. There is definitely something wrong with that recording, but the decoder tried to keep it as clean sounding as possible. spaces.hightail.com/space/xghqJodgrjJohn
|
|
|
Post by jsdyson on Aug 31, 2018 15:33:43 GMT
I need to conditionalize the comment about 90deg phase shift. I noticed instead (and I forgot about my unexciter software -- which tries to undo the effects of the Aphex exciter), it seems also to improve the sound quality. So I am not sure if it was just a 90deg phase shift for the mid channel or use of an aphex exciter that would supposedly 'improve' the sound (only causes a pernicious kind of distortion.)
John
|
|
|
Post by jsdyson on Aug 31, 2018 11:55:09 GMT
Some people might know that I use ABBA for one of my tests cases on an important project that I am working on. However, this is more about ABBA than my project.
Since I use the ABBA material for tests, I have become somewhat technically familiar with the quality of various versions of the ABBA distributions, and some have been very disappointing. The best ones are the least accessible to the normal listener today -- both the vinyl versions and the DolbyA encoded CDs (including the 1992 ABBA Gold CD -- it IS DolbyA encoded, at least the one that I have.) And, of course vinyl isn't very available to most people any more. (You CAN get the DolbyA encoded versions relatively easily -- but decoding them is where the accessability is impeded.)
Before someone thinks -- oh, but my favorite copy, thus and thus is good -- well, nothing that I have found (including a few Polar/Polydor releases, and also the ABBA 'the complete studio recordings) are as good as the old vinyl distributions (the vinyl itself might be damaged, but the quality of the recording is pretty good) or properly decoded DolbyA encoded versions.
First, there is some kind of damage done to the masters that seems to have occurred since the early vinyl releases. Some of this damage can be undone, but is a rather extreme technical measure for most people. I explain in the paragraph at the end, which you can ignore if you want. This damage is IN ADDITION TO the frustrating DolbyA encoding.
The digital Polar/Polydor (not all are the same, but I am speaking in generalities right now) apparently were not DolbyA decoded very well, but seemed to come from a reasonably good master tape. In one or two cases -- e.g. one of hte Polar Supertrouper copies -- was done significantly better than a lot of US domestic versions.
The most potentially beautiful copy, but also the most atrociously over processed version is the 'The Complete Studio Recordings'. If they hadn't overcompressed it so severely, it would have been better than any other copy that I have. The crest factors and the peak-average ratios are atrociously small, and indicate a very good quality and too aggressive compressor being used :-). The recordings are 'pretty', but have lost some of the good parts of the ABBA sound. I have tried 'uncompressing' the TCSR recordings -- with some success, but could never regain the proper crest factors and peak-average ratio (I call it 'uncompressing' instead of 'expanding', because the goal is to undo an effect, not to add a new one.)
Except for the vinyl, and superficially pretty bad quality -- the releases like ABBA Gold 1992 are actually the best, if you can DolbyA decode them. Unfortunately, ABBA & DolbyA are not very nice friends -- ABBA is incredibly tricky for DolbyA to decode because of the mixed high pitched voices, and otherwise intense HF sound. That creates huge amounts of intermodulation, and is one reason for many ABBA releases sounding too compressed during 'crescendos' in their singing. My DolbyA compatible (not licensed or encouraged by Dolby labs) decoder DOES do a fairly good job of undoing the DolbyA encoding (much better than DolbyA HW), but still it would be much nicer to have clean copies of the pre-encoded master tapes to work with. (The compression effect when encoding/decoding difficult material for DolbyA seems similar to mild overload of tape itself... Such intermodulation effects do sometimes cause a sound that is similar to gain compression.)
So many of the ABBA releases have been needlessly damaged, and after the some 5-10copies of each song/recording that I have (probably more than that number), it is so sad to hear all of the major differences in the releases, yet KNOWING that it sounded much prettier in the studio. The saddest thing (but potentially most fortunate in the longer term) is that so much pop music has been delivered in DolbyA form (that is one reason for the 'harsh' digital sound that so many people are used to nowadays), and the listener base has become accustomed to the unnatural sound of DolbyA compression without decoding. (When any so-called 'experts' claim that material that I am suggesting is not DolbyA encoded and DolbyA encoding is too ugly to distribute -- well, I agree with part of the point -- the DolbyA material should NOT be distributed in that form. And, I am dealing with tape experts who know DolbyA very well -- and it is pretty clear that a LOT of material IS being distributed in DolbyA encoded form because of cost savings and/or laziness.)
I still remember the time when recordings could sound 'lush' instead of HF-hyped. In fact, they used to use DolbyA units to make voices sound clearer (back in the day when such devices were less common.) By leaving material DolbyA encoded, it makes it sound clearer (sometimes too unnaturally clear), but also tends not to boost the highs in a way that will cause severe clipping (because it mostly just boosts the lower signal levels, and mostly affects the highs except at very low levels for the mid frequencies.)
John
PS: Explanation of part of the distortion between early ABBA releases and the DolbyA encoded copies:
Apparently, there is something odd being done to the M+S encoding of the recording. Normally, we think in terms of L+R (left and right), but sometimes M+S(mid and side) are useful versions of the material. M+S can be easily calculated from L+R by very simple relationship: M=(L+R)/2, S=(L-R)/2. Note that side is the difference between L+R, and that is where the oddness appearst to be. I found that by shfting the Side channel by 90degrees, that the material clears up. I am not sure the best place to do the shifting, but I think that it is best to do the shifting after DolbyA decoding. It massively clears up the results, and produces quality very similar to the old vinyl relases (without the vinyl distortions and noise.)
|
|
|
Post by jsdyson on Aug 1, 2018 2:44:11 GMT
Gang -- My DolbyA decoder is now being targeted to Archivists and recording pros BOTH. Some of you already know that I have been uploading demos from time to time -- well, these are worth listening to. NOTE: as of a few hours after this original post -- I did an update to the repository -- made an error and grabbed the wrong files for the demos. These WILL blow you away!!!On the site, there are two short files that compare using the high quality DolbyA compatible decoder with an actual Polar CD release (or Polydor?) Anyway, the difference clarity is striking (but, of course not perfect.) I have uploaded three examples, SuperTrouper, UnderAttack and MamaMia -- full quality, only hit on the quality is mp3 encoding at 256k. Strikingly clear - previously impossibly clear and accurate.I am making the very bold claim that the quality that you hear when playing these examples is as good as the studio before encoding in DolbyA. If you compare with non-hyped versions the examples are about as perfect as can be extracted from the material. The ABBA studio TCSR recordings are not really fair -- that material is incredibly heavily processed with crest factors of 4 and peak-rms of 13-14dB. Music usually needs crest factor of 6 or higher and a peak-rms of at least 15-16dB to sound normal. The demos have peak-rms of about 18-20dB, and crest of at least 8. However, the important thing is the recovered detail. For example, the chorus during SuperTrouper and MamaMia actually sound like voices instead of a 'blob'. Do a real comparison -- the short Mama Mia 30second examples do not do it justice. ('orig' is the original CD, 'decoder' is the DolbyA compatible decoder working on something close to a master tape.) I demoed Under Attack because people who know about how attack/release times work -- Under Attack is kind of tricky to get right, and also the Dolby level has to match pretty well. I am not totally happy with 'Under Attack', but it is pretty darned good. Please enjoy the demos -- they have to be erased in a few days (maybe a week.) However, these are copies that exist no-where else because of the crazy heroic means used in the DolbyA compatible decoder to extract everything in the recording. Demo location: spaces.hightail.com/space/xghqJodgrjJohn
|
|
|
Post by jsdyson on Jun 27, 2018 8:48:49 GMT
Really worth listening -- TOTALLY cleaned up SuperTrouper with no added compression, 21dB peak to average ratio, 11+ crest factor. The spectrum all the way to 20kHz. No audible distortion, very natural sounding. Dreamworld as clean as you have ever heard -- no added compression, 18dB peak to average ratio, 7.98 crest factor. The spectrum all the way to 20kHz. Most of the excessive sheen is gone. PS: I just re-uploaded with a slightly different processing -- I added the unexciter to remove most of the evil Aphex exciter nasal sound, and slighty changed the EQ. The processing sequence from the master is: DolbyA -> unflatten -> (uncompress on SuperTrouper & Dreamworld) -> unexciter -> mp3/very mild EQ (no more than 3dB anywhere.) location: spaces.hightail.com/space/wU6nwJD4bWEnjoy -- John
|
|
|
Post by jsdyson on May 29, 2018 20:06:59 GMT
Something kind of interesting and has been bugging be every time that I have listened to the Super Trouper recording on the album since day one (at least back to the 1980s.)
Remember the song, and approximately 62 seconds into the song there is a chorus that starts, and is somewhat understandable but too compressed. Then going on to 80 seconds the chorus goes nuts to the point where it starts becoming difficult to understand. It is obviously over compressed to the point where the singing has morphed into a sound effect.
I have been playing the studio recordings through proper DolbyA compatible and DBX Type I compatible decoding, and the severe 'squashing' is gone. I am not touting the decoders, but rather that SuperTrouper could have been signficantly more clear all along!!! The very beginning, there is a little bit of roughness - but I haven't heard any severe roughness recently after usng some DolbyA decoding on relatively clean DolbyA encoded material. However, even then the chorus is too distorted.
I have tried all kinds of measures to recover the chorus, but something has severely damaged the material in almost every case except for the complete studio recordings version (after being decoded.) It is the FIRST TIME that I have heard the chorus in a way that seems to make sense to me.
SuperTrouper is such a nice song, and almost every copy (until now) has been distracting. The studio recordings are miracle, but need some special work to gain access to the beautiful ABBA music.
John
|
|
|
Post by jsdyson on May 28, 2018 22:09:59 GMT
I have a copy of TCSR -- and know some technical aspects that most people don't realize. I did some research and some people have reported that it is TOO COMPRESSED or doesn't sound good -- things like that. I can tell you why, and this is almost 100% proven. The TCSR (at least some of the copies) were distributed with both DolbyA and DBX type I noise reduction. Using both systems together is henious, but can provide BEAUTIFUL sound if you have the abillty to decode it. Some of the software that I have been writing (and it works) does a accruate DolbyA decode, and the new DBX Type I decoder is working very well. There are technical aspects which can make it a challenge, but the challgening issues have been fixed. On my Hightail repository, there are some example decoded ABBA recordings from the TCSR. These still have aliiasing from the earlier version of the Type I decoder. The new version has the aliasing mitigate, and slowly (I mean 10-30minutes), I will add some of the aliasing-mitigated versions also. The alising gives a hard edge to voices -- and removing makes them natural. So, if you have TCSR, YOU HAVE DIAMOND IN THE ROUGH. Here is the respository site: spaces.hightail.com/space/UntM4LCdcm
|
|
|
Post by jsdyson on May 26, 2018 3:36:30 GMT
Please try name of the game again -- this listening experience WILL BE RELIGIOUS. Listen to the entire thing -- the stereo image will knock your socks off. This is the result of my project and some additional work on audio procesing. Any previous demo PALES in comparison. I am currently processing the entire ABBA corpus in my posession with the technique. Simply spend 10 minutes. It will be worth it. I must pull back the demo soon because it is the complete song. The entire song must be listened to for the beauty and image. spaces.hightail.com/space/UntM4LCdcm
|
|
|
Post by jsdyson on May 26, 2018 2:56:28 GMT
You really MUST listen to these. I found a much better source -- and applied DolbyA and a DBX style RMS expander (and a bit more 'love'). I found a wonderful package of ABBA music in my collection. I was SO surprised. Of course, it doesnt' sound like this until the proper processing. This is REAL -- and you gotta listen. Kisses of Fire (kof.mp3) Name of the Game (nog.mp3) Repository: spaces.hightail.com/space/UntM4LCdcmIt really is worth the 5minutes to check these out. I have been processing about 50-100 ABBA songs right now, and cant wait to listen to all of them. These were test samples, and not fully complete. Can you imagine what they will sound like when they are done? John
|
|
|
Post by jsdyson on Apr 18, 2018 22:53:49 GMT
Well -- good news is that I have been improving them more and more. What you might hear in an hour from now will be 10X better than last night (really.) I haven't normally been working on the ABBA stuff -- again (for a few months) until now. ABBA is especially interesting to me because it is so tricky to process. If you look at the AG1 repository (one of the respositories above), I also have some Petula Clark stuff. It is SOOO pretty. But they did some stuff to the ABBA recordings to make it sound loud on AM radio -- and the ABBA is very difficult to make it sound right. However, to try to unwind some of their tricks is quite tedious and actually a little difficult (I didn't know ahead of time what they did.) As this is being written, I currently have some vastly improved ABBA GOLD 1 examples, but I intend to update the SOS and SuperTrouper examples within about 1/2 hour (needs more sidechain compression.) But, from a sound smoothness standpoint, the SOS & SuperTrouper examples really sound pretty good now (as of about 30-60minutes ago.) The AG2 still needs more work (in progress -- running some expansion and compression right now, about 3-5minutes per song), and should have AG2 examples updated in about 1Hr. THIS IS NOT EASY -- because I am working to make it prettier -- not just showing what my best source material sounds like. It is hard to explain what they (ABBA) did -- one thing is that they added stuff (somehow) to the L-R signal at 90deg. It sounds kind of like distortion on high quality equipment, but probably sounds more dense on AM radio. Also, apparently they used an Aphex Aural Exciter, which gives a nasal quality to the sound. It sounds louder but isn't really prettier -- I wrote code (a single program) that undoes some of the L-R and Exciter sound -- but that isn't the most difficult thing. Perhaps one of the more difficult things is that the music is an acid test for my psuedo-DolbyA decoder (don't let the adjective 'psuedo' imply that it doesn't 100% decode DolbyA!!!) The density of the 2kHz to 15kHz is pretty intense, and really tests the tracking of the 3-20k and 9-20k bands on the decoder. Also, any mistakes in the attack/decay are so easy to tell -- I am currently running an experimental version of the psuedo-DolbyA, and it still does have troubles with UnderAttack -- even though the mainline version has NO trouble with UnderAttack. (I am working on some very esoteric improvements -- further mitigating sources of intermodulation.) So -- in some cases my examples might not be as good as possible (yet.) I am intending that the psuedo-DolbyA be used to recover old archives of music -- and do the best possible job -- even better than the real units. One thing that might really confound you (it did confound me) is that perhaps 1/3 of the old music is still DolbyA encoded as it is provided to the end user. A simple example might be the Carpenters from HDtracks. Most people might not detect the compression artifacts on the music distribution -- but if you listen carefully, you'll notice TOO MUCH COMPRESSION, and more hiss than should be. A DolbyA decoder can remove the hiss and recover the original dynamics in the music. This is even on a supposedly pristine music distribution!!! A Queen CD that I purchased about 3yrs ago -- still DolbyA encoded!!! It just keeps on happening. The small ABBA examples that i have on my repository mostly are just a little more transparent than the original and have a little greater dynamic range. The original material is still quite compressed -- but when they make the disk/CD or whatever -- they often do quite a bit more compression. Even the Polar CDs are generally VERY compressed relative to the original material. PS: I now have the examples pretty much as they are going to be for a while. There is still a bit of an EQ issue (not much though), and I reverted the DolbyA decode to the standard version for UnderAttack. (It is pretty good now.) PPS: Some new repositories -- ABBA in some of them. These are mostly demoing my DolbyA decoder -- really proving the noise reduction/etc. Site1: spaces.hightail.com/space/z3H68lAgmJ Site2: spaces.hightail.com/space/ko2yTjF5YY
Have fun!!!
John ..
|
|
|
Post by jsdyson on Apr 18, 2018 19:41:22 GMT
The direct URLS that I posted above are no longer valid -- but the URLS to the repositories are correct!!! I reran the mastering again -- got a lot more aggressive, and changed some of the recompression to side-chain instead of fully inline (sounds better yet.) Most of my goal is been trying to get rid of the harsh/confused sound -- and I found that simply being more aggressive with the 'distortion removal' phase works wonders. I was being too conservative by using a factor of 0.707 instead of the much better sounding (as I found) 1.0 removal of the quadrature L-R signal.
The results now are good enough that I am actually happy with them for my casual listening. I don't find enough defects to distract from my enjoyment. There is still one defect that I hear to be easy to remove, but figure that the improvement wont' be worth the 2-3Hrs of computer time (it has about 1.5dB too much between 2.2kHz and 9kHz.) That slight boost doesn't sound like alot, but is noticeable. In fact, in a car or when listening to speakers, might be desirable.
The defects removed since the last upload include further removal of 9kHz on up distortion (sounds reasonably clean now.)
Also, instead of using one setting for every song, a few of the songs are processed slightly differently (other than the DolbyA threshold, which as been different for groups of the songs forever.) Dancing Queen, SuperTrouper, SOS and Waterloo are now processed with significantly different parameters.
The current results (I have pristine flac files -- sound better, believe it or not), are good enough to keep in my archives now.
I might work a little bit on UnderAttack and see what else I can do to I am a City and SuperTrouper, but right now all are pretty darned smooth and ABBA-like sounding.
John
|
|
|
Post by jsdyson on Apr 18, 2018 1:03:59 GMT
Okay -- I took this as an opportunity to re-EQ the selections... I tend to have very variable hearing, and I think that I had too much HF -- so I dropped 6.5kHz by about 1.0dB and 9kHz by 1.5dB. On my Windows Box (which I seldom use), it is running Win10, and I tried FireFox, Chrome, Edge and IE... Seems like they work, but let me try something else. I'll try to supply direct URLS for a few selections: MamaMia: spaces.hightail.com/space/nXCmV47em2/files/fi-81e15cab-212d-4a7b-9eea-57187e45aaa2/fv-79decac8-de70-48cf-89a8-af0a3440717f/MamaMia.mp3TakeAChanceOnMe: spaces.hightail.com/space/nXCmV47em2/files/fi-21817ead-b8ea-4ad0-b4dd-73171ccf8f8f/fv-8ca254b8-ed10-4b90-9f48-84c68c386651/TakeAChanceOnMe.mp3SuperTrouper: spaces.hightail.com/space/nXCmV47em2/files/fi-3535664c-4c3e-49c6-b1c0-d6b725b1399a/fv-09db262f-c396-4ef4-b1a4-24d4dc85f417/SuperTrouper.mp3SOS: spaces.hightail.com/space/nXCmV47em2/files/fi-f11e10d5-2ae9-4452-a39c-6dd2c8c54fac/fv-741233e1-5177-4e8d-af6b-e4c3d068f7a5/SOS.mp3HoneyHoney: spaces.hightail.com/space/G5zKvkAr6a/files/fi-529cb28a-493f-4160-b70e-dabf2b221875/fv-33ba6fb5-bf6d-4593-ae06-661e8ed1961c/HoneyHoney.mp3UnderAttack: spaces.hightail.com/space/G5zKvkAr6a/files/fi-6457d683-beec-4f6a-83ee-29bc7a2ef234/fv-b66d114f-8f2c-4fe1-901d-9cf8fa07d0c0/UnderAttack.mp3DreamWorld: spaces.hightail.com/space/G5zKvkAr6a/files/fi-ad973de8-1e49-41ec-ba25-790bd10bdea8/fv-7759dc87-5617-48e2-9dc7-bb971629e460/DreamWorld.mp3I normally do all of my work on Linux/FreeBSD type OSes -- much easier than Windows (focus on the work and not the messed up GUI choices.) Again, if there was too much HF before, it is worth giving another try. The DolbyA decoder was slightly misadjusted for UnderAttack, but still sounds okay. It takes about 10minutes per song to process (not for the DolbyA, but for the un-Exciter to undo the Aphex Exciter.) The psuedo-DolbyA runs in realtime up to 256k samples per second, and I normally run it at 96k and 192k, even though it works well down to 44.1k, and actually functions reasonably (however far from meeting spec) at 32k. The processing is as follows: Pristine DolbyA encoded 2trk copy -> psuedo-DolbyA decode -> un-Exciter -> removal of 90deg distortion on M+S (on alot of ABBA stuff -- esp Dreamworld) -> 1:1.17 expansion, 1:1.21 expansion on M+S -> 3Band compression @ 1.7:1, 3-5dB depth -> 8Band compression at 1.4:1, 1-2dB depth -> soft limiter 2:1, 1-2dB depth -> hard limiter inf:1 (maybe 1dB once in a while) -> 96k flac master -> mp3 copy The compressors have totally dynamic attack/decay times, and fit the contour of the music envelope (inst attack time, 5msec -> 2second decay time, depending on dynamics and freq. The expanders are 'wild west' with totally variable attack/decay times in both the linear and dB domains. In the linear domain, the transient recovery is done by curve fitting, but the dB domain provides the long term attack/decays. The decay time can be as long as 1dB/sec, but is usually MUCH quicker than that. The compressors/expanders are designed (like the psuedo-DolbyA) to have almost ZERO intermod, and where there is the necessary intermod because of the physics of the gain control operation, it is actually filtered when possible. (You can actually see the difference in the intermod sidebands when doing a spectogram -- the anti-intermod methods actually do work.) Even though it is kind of hard to tell -- the ABBA stuff can be pretty intense on the highs, you might notice that the sound is less 'fuzzed' in the HF range. The voices are SLIGHTLY more distinct than when using normal compressors/expanders -- but the problem is that the kind of processing done on each CD/album can be VERY different... Even using the apparent SAME master tape, the mastering engineers get very different sound by doing matrixing, parameteric eq and different kinds of compression. I am not a good enough mastering engineer (I am not one at all) to get perfect sound, so my work product is the software that does the processing and requires the REAL EXPERTS to utilize!!! John
|
|
|
Post by jsdyson on Apr 17, 2018 21:05:57 GMT
Got REALLY good news -- my psuedo DolbyA decoder is now being told (by some recording pros) to me that it is better than a real HW DolbyA. I have also enabled some new anti-intermodulation features that expand the quality further. Also, I have an anti-Aphex-Exciter software pacakge, which helps to remove some of that Exciter sound (makes the sound clearer, removes some nasal quality.) The bottom line, I have also found a way of getting rid of 90% of the distortion on my copy of Dreamword -- sounds really good. Have the examples online at the following location: spaces.hightail.com/space/G5zKvkAr6aAlso have some other examples in: spaces.hightail.com/space/nXCmV47em2The really, really difficult thing is to get the ABBA sound, but clarify it more. My psuedo-DolbyA decoder has less intermod than any HW can produce (really). and also matches the gain curves very closely (I have measured 0.5dB deviation against a real DolbyA unit.) Please enjoy -- I dont think that UnderAttack, Dreamworld, SOS or especially SuperTrouper can be presented any more clearly.
|
|
|
Post by jsdyson on Mar 28, 2018 13:13:03 GMT
I learned some 'interesting' things about some of the technical aspects of ABBA's sound. They apparently used an early model of the Aphex Aural Exciter, where the early version was a phase scrambler which probably gave both the dense sound and the illusion of more intense high frequencies... This is most helpful in the AM radio era, where the BW available was in the range of 4-7kHz in general. Good and fun news: I have been working on a DolbyA decoder (really works, been proven -- sounds better than the Satin and maybe even the real DolbyA!!!), un-Exciter, and also an un-compressor. Some of my results are A LOT of fun. The most technically interesting are the fairly accurate removal of the Exciter processing, and the most fun is when ABBA is expanded by a very unobtrusive uncompressor... (I call it an uncompressor, because it works like no other expander that I have EVER heard about, and I do lots of research.) Technically interesting: spaces.hightail.com/space/nXCmV47em2 Fun sound: spaces.hightail.com/space/pUBEacRM3YNote that this software is not just a toy -- been getting help from a recording engineer who knows his stuff (not at liberty to divulge names yet), and also been doing iterative improvements for quite a while. This is a research project that might just be some fun when hearing the music better and better. Been also working on recordings from other groups, but back working on the ABBA situation for a few days. Really -- ABBA never sounded better. It is still not perfect, but it is doing better and better. PLEASE ENJOY!!! John
|
|
|
Post by jsdyson on Mar 25, 2018 13:43:11 GMT
This is your local 'ABBA clarify/cleanup' crackpot again. I just got some REAL intelligence that one reason for the ABBA sound was that the might have been using an early generation of the "Aphex Aural Exciter". Of course, I took that bit of intelligence as a challenge. So -- I did some further research/reading old magazines from that era/etc, and got enough information that I might be able to undo some of the Aural Exciter damage. It didn't take long to write an un-Exciter, to bring the clarity/lack of fuzz back from the music. Anyone who would like the source code can have it -- it is just a few SOX scripts, but will not work as well on more recent devices. There are several phases of correction -- one is to un-Excite, but there are some further passes that do a filtering operation that biases the signal towards being a signal with analytical characteristics ( a special phase behavior that tends towards what simple, natural things do.) Normal electronic device impedances are such an analytical function also -- when you know that behavior, designing wideband impedance matching (for example) becomes easier. Well -- way, way off topic. Most people reading this like ABBA -- and so do I (maybe less than some, maybe more than others.) So, after the un-Excite filtering, and the regularization of the signal towards the natural signal behavior (tends to remove un-natural distortion sources), you might be able to listen to the clearest ABBA from normal music distributions EVER. Even though I might be a bit of a crackpot -- the examples that I am providing are VERY SPECIAL and VERY CLEAR. Take a listen and enjoy. The site below mostly just has ABBA, and the most recent recovered copies. I have to say that the technology that undoes the distortion here is a breakthrough and REALLY works very nicely. SuperTrouper is my benchmark, and frankly, there is very little distortion left on it. spaces.hightail.com/space/nXCmV47em2
|
|
|
Post by jsdyson on Mar 19, 2018 21:32:54 GMT
I really need some listening help -- really!!! I have some slightly cleaned up ABBA stuff with the full names. What i am most interested in is FEEDBACK on the music files that have the '-p' in the filenames. These are from AGold, and they are cleaner, but sound heavier. I need to know what it sounds like to you'all. I really need input!!!
Wow -- this is an interesting project cleaning up the ABBA sound. I was always envious of the sound of the old ABBA videos, and that is kind of what I am trying to achieve. I just made some other major improvements. They did a real number on the phase of the music at HF-- that is what causes that dense sound from ABBA. I keep iteratively trying to unravel that dense sound. Here is another step of unraveling. You might want to EQ the results a little. The are at least doing one thing that I am trying to undo -- they are apparently moving the HF from the inphase to the quadrature and vice versa. That gives that extra density, but also causes a bit of a distorted sound.
This latest unraveling seems to be pretty good on the older songs, but not so good on the newer ones.
Give it a listen. Some of these might be worth keeping (on my repository.)
John
|
|
|
Post by jsdyson on Mar 19, 2018 9:56:04 GMT
IF YOU DOWNLOADED EARLIER -- TRY AGAIN... I made a mistake in the previous version -- it should now be clear like I suggested. I am not providing everything that I promised, but that is due to the time constraint and my previous error. However, this should be quite pretty sounding.) If you really like 'Name of the Game" -- you might like the version that I created. This was from a Spanish language CD release that wasn't as processed (or decoded) as an American/European release, then subsequently DolbyA decoded (with a vastly superior -- minimal Intermodulation decoder), then removal of some of the harshness from the 90deg phase thingie (used to be used for quad, but in ABBA seems to be used to make them loud sounding.) The sound is impossibly better and quite unprocessed other than basically proper decoding and a minor cleanup (no NR, no 'peaking or heavy EQ', no gating, no expansion other than the DolbyA.) This is as natural sounding as one can get from the completely assembled song. spaces.hightail.com/space/nXCmV47em2
|
|
|
Post by jsdyson on Mar 19, 2018 3:09:40 GMT
Please hold up for about 1HR. I screwed up some of the processing (however minor). It will be fixed by approx 9:00AM EST USA time, 19Mar2018. (I just fixed the error -- the major problem is corrected, but I couldn't provide everything that I wanted. It is still clearer than ANY OTHER VERSION. The previous mistake had a filter that wasn't intended.) I have too many things going on!!! I have some of my own remastering test results that you might find to be very entertaining. My new processor/decoder does things that are down to the point of theoretical ideals regarding intermodulation (due to gain control), etc. I know that the techno-mumbojumbo is meaningless to most, so I won't bother you with it. This is as clean as possible a rendition unless you have the multi-track masters. This is as good as any ABBA that I have ever heard, and you'll hear things that are no longer smushed together, while also having the true bass from the original (I had to contour the bass/midrange and oddly change the 90deg HF a little to get rid of the blaring.) Note that I didn't do any heroic processing or special NR. These examples are as reasonably pure as I have ever heard. Again, ABBA isn't my project, but I use their music (a lot like Suzanne Vega's music being used for designing MP3 itself.) Please enjoy -- I'll keep these up for a few days... I have tried to format the files so that they can be played online for now. I suggest that listening to 'mamamia' might be the best first choice. The reason is that the sound is VERY different than expected --- it is all there, but is much more natural sounding. You might even be surprised that Dancing Queen is also not quite so scrunched sounding. I did NOT expand the music in any non-standard or artistic way, but simply did what was necessary to master the music. This has the ABBA sound, but not the ABBA blaring sound. I found out that some of the over-intensity was hidden in a special electrical place in the signal -- so one thing that I did was to back-off some of that hardness. A simple tone control type thing was not used to remove the hardness, but it wasn't any kind of fancy AGC thingie either. spaces.hightail.com/space/nXCmV47em2PLEASE ENJOY!!! John
|
|
|
Post by jsdyson on Feb 11, 2018 23:12:28 GMT
I have a truly beautiful collection (much better than ever before) of some ABBA stuff. I also have my latest pseudo-DolbyA decoder posted on the site below (source, win64 binaries and linux/x86 binaries (both 32 & 64).) All of my previous cleanups of my best copies of ABBA had problems with an esoteric DSP problem called 'aliasing'. The gain control was mixing with the audio, because of nonlinearites were producing components outside of the sampling nyquist frequency -- this causes harshness similar to the beginning of SuperTrouper on many ABBA releases (that isn't really aliasing, but is similar.) The examples that I have produced (from the most pristine sources that I could find) are truly clearer yet than anything that I have heard -- less 'harshness'. I am NOT an artist, but only a lowly engineer (I mean EE/DSP/Software, not recording engineer), and so am doing the best that I can do. I want people to enjoy. I have also (finally) restricted my MP3 encoding to comply with my respository site, and can play the mp3s directly. There is a definite loss of quality due to the conversion between 24 bit flac/96k to 48k mp3 @192k -- I have measured some differences, but they still are very representative. spaces.hightail.com/space/pG4t4ZFnyBUPDATE: I just re-uploaded the latest processed versions. Please note that Dreamworld is about the best that I have ever heard, and Under Attack is also pretty darned good. I have left almost all of the original 'sound', and absolutely no re-compression on these. My latest anti-aliasing expanders have been used -- the pseudo-DolbyA is even more advanced than the version that I have published on the repository... WARNING: be careful about setting your levels. In some examples, there are few hints about how loud the level is set -- so just be careful and be ready for loud signal levels.
|
|
|
Post by jsdyson on Jan 29, 2018 10:50:54 GMT
The following is a pointer to the best yet reprocessed (and VERY vanilla) cleanup of the best ABBA sources that I can find. These results include music coming from DolbyA encoded copies that I got 10+yrs (prob 20yrs) ago (not DolbyB or C), and I haven't really been able to really listen to them until recently (DolbyA sounds a lot like DolbyB, except with more compression on the low/middle frequencies -- so without decoding, sounds quite overcompressed, but not as bad as DBX.) The major thing that I have to contribute is that unless DolbyA is decoded on exactly the same unit at the same temperature, with exactly the correct threshold setting, then the results leave a bit of distortion. I have reworked the software to be able to conform to a wide range of encoder characteristics and have tuned (painful amount of listening) to match the 'dolbyA level' to a reasonable degree. Also, have applied a light amount of expansion/NR to bring the music closer to recent sensibilities. The most major criticism might be the sort of EQ that I used... I didn't do heavy EQ, but tamed some of the harshness in the sound. If you want to hear more of the HF sometimes associated with ABBA, I can make that available. Here is the site -- everything except Dreamworld has been reprocessed within the last 12Hrs as of 6:00AM EST (US) in the morning. There was no extreme expansion being used, and because of the very special kind of expansion the freq resp balance is darned close to the original after expansion (again, I did do a small amount of EQ.) The EQ cant really be 100% corrected after the fact, because part of it was done before processing, but a small amount of treble boost might bring it closer to more common sound character. spaces.hightail.com/space/pG4t4ZFnyB
|
|
|
Post by jsdyson on Jan 22, 2018 11:30:31 GMT
Man -- I am so embarrassed. II had lost some of my hearing and it returned (illness) -- so very embarassed as to some of the examples that I had created. My processors (both of them) had a fatal flaw in part of the implementation. There was a very esoteric DSP design problem which caused lots of HF intermodulation products to be produced (hash in the sound.) This was especially bad for ABBA -- two voices with moderately high pitch mixing together. Normally, their sound is interesting, and it sounded ok with my limited hearing. A day or so ago, the hearing came back and I got about 10dB more hearing at 10kHz. Listening to my examples -- YUCK!!! I did some research, and found some problems -- most DSP engineers would not have any idea as to the problem, but there are some subtle things that have to be done when working with audio and fast gain control. Slower gain control software which has nice, easy 10-20msec time constant attack times don't have this problem. The expander software has to have 1msec or faster attack times to sound correct -- and that is where the problems become obvious. There is a complex interaction between the audio and the level detection mechanism -- and 'rectifiers', 'absolute value functions' or things like that seem to be much worse than 'square law' detectors. I didn't realize this an a 'feeling' sense because I couldn't hear the problems. So -- I studied the problem -- learning a lot, and developed some schemes which provide essentially the same effective attack/release needed for proper sound, but also mitigate the effects of the extreme hashy intermod created as a side effect of the gain control detection process. Even the simple pseudo-DolbyA processor had some problems -- even though that cool notch that Ray Dolby designed into it really helps, it doesn't do enough to fix the problem (apparently some mixing between the sampling, signal and the nonlinear detection function), which adds to the hash. My first examples -- with NO tuning, extra processing, no EQ, no matrixing or anything like that are coming off of the assembly line. The sound is VERY 'vanilla', and isn't meant to impress other than to show that the hash is pretty much under control. The ABBA songs in the following repository have been 'fixed', and I am populating the repository as the music is coming off of the processing assembly line. There are some other songs on the repo, but it is the ABBA songs which have been fixed so far. spaces.hightail.com/space/pG4t4ZFnyBPerhaps the most interesting thing now is that when I distribute the new pseudo-DolbyA, it contains some of my technology that I wasn't planning on divulging (helps to fix the intermod.) John Dyson
|
|
|
Post by jsdyson on Jan 20, 2018 16:05:53 GMT
Listened to a few mashps -- pretty good. Innovative. Good news -- I have corrected the problem (a serious mis-synchronization of the attack times, causing a raspy harshness that my ears couldn't hear.) I have carefully reworked the synchronization, and have VERY CLEAR and fairly non-raspy results!!! These aren't really 'mashups', but are rather pure copies of some ABBA songs. These are incredibly pure, and are almost distortion-for-distortion the same as the Polar versions, except no compression or noticeable EQ. Also, you'll notice actual bass in these copies. The normal releases appear to rolloff the bass starting at about 100Hz, but that just isn't happening in my versions. The worst one (and my most aggressive test case) is SuperTrouper, but you'll find (esp the start) is MUCH clearer in my version. Again, I GOT RID OF THE DISTORTION... spaces.hightail.com/space/d4fpzoa0zD
|
|
|
Post by jsdyson on Jan 20, 2018 2:46:25 GMT
I do understand your position. I mean NO slight to claim that what I am doing is to recover what B&B & Mr T did in the recording process -- my own work results from frustation knowing that someone in the custody chain of the recordings had distorted them so terribly. There is NOTHING wrong (in fact, perhaps beneficial) to add the work from various recordings together to produce something interesting. When adding pieces of differing recordings, in fact, might be useful to bring the source materials closer to the original -- THEN reprocess them to perhaps match better. So, the output of software like my 'restoration processor' or others similar MIGHT be beneficial for the merging process.
In some cases, the quality of the recordings that we (esp in the US, with the loudness wars) get is atrocious. Sometimes, the recordings are made aggressively loud by the original artist (obviously, Shake it Off was apparently done in that way -- I can remove much of the compression and hear the components fairly clearly -- but the 'sound' was done in the studio/artist controlled recording.) I don't care if the music had been processed to meet the artistic goals -- but in the case of some ABBA stuff, it is apparent due to the EXTREME variation in sound quality from manufacturer to manufacturer, ABBA recordings as presented to the public have INCREDIBLY variable quality and some APPEAR to be significantly different (in a bad way) from the original.
ABBA music is already processed to the hilt -- and does sound pretty good when listened to in a relatively pristine condition (Polar comes close to the most unprocessed copies that I have heard, but still has some compression, some EQ and apparently some matrix shifting of some kind.) Some of the nonsense dostributions that I have gotten here in the US borders on USELESS. If you listen to examples that I placed on the Hydrogenaudio Forums, you can hear an exact comparison between the Polar and a US release -- and then I have a snippet of the most virgin copy of ABBA recording that I have (the 'virgin' copy appears to be DolbyA encoded -- so is also compressed.) On this matter, my processor is not even an issue, but rather the issue is incredible cheat against ABBA music lovers, and even the ABBA team itself.
The 'most virgin' copies of the music that I have are pretty darned clean sounding -- and where I am perhaps slightly misguided, I have gone a step further to minimize/undo some of processing done on those clean copies also. At that point, my other, 'signal processing', hobby comes into the forefront. By attempting to undo damage to the ABBA recordings, I might actually be adulterating it. I might be guilty of actually damaging copies of the music that come close to the intent of the artists -- however, my major defense (or defence for my European/UK/OZ friends :-)) is that a little additional processing can bring what I have produced back to essentially what is on my cleanest copies. That is, I am trying NOT to take away very much from the recordings, and perhaps bring them up A LITTLE to a 2000's sensibility as to what is expected in the technical qualities of the recordings. I have NO interest in changing the artists basic intent.
So -- I guess what I am trying to say -- I am as far from a music artist as one can get. Those who have the talent to create new music (or successfully merge multiple pieces of music), go far beyond what I am capable of. I have REAL artistic limitations, but I am happy (and almost driven) to try to add what few abilities to the world that I can. I am a techie through and through, I love my work, but also understand (deeply understand) my limitations. I am interested in what is going on, and will probably listen to some mashups and enjoy them. On the other hand, I am trying to help/benefit the music listening community as a whole (and also the producers) with my very limited music creative ability (but I am quite discerning.) My ability to discern and create from a technical standpoint is what I can offer!!! Take care -- and have fun... John
|
|
|
Post by jsdyson on Jan 19, 2018 14:43:36 GMT
My big FRUSTRATION when listening to music (esp ABBA) are the huge differences when playing the exact same recording from different CD/record manufacturers. My own goal is to be able to hear what B&B and Mr Tretow actually put out, and not what the record companies morphed it in to. So, I might be misusing the normal term 'remix', where I am mostly taking different aspects of the same recording and moving them around a little -- plus (before the 'remix' activity) doing what is necessary to recover what B&B & Mr T created in the studio (of course along with A&AF.) Some CDs have come out that sound pretty nasty -- even to the extent of real distortion.
So -- I don't mean to confuse when I use the term 'remix'. I am not mashing up or remixing from different sources. There MIGHT be some cases where I could take different copies (from different sources) of exactly the same recording and extract the best of both, then recombine. I haven't done that yet, but have considered it for a few songs. The synchronization effort to do that at the DSP level is a little tricky, but I'd do it if it made the music sound more natural with respect to the recording sessions.
John Dyson
|
|
|
Post by jsdyson on Jan 18, 2018 21:54:35 GMT
Gang -- I'd been reading other threads on this board, and noticed a thread 'dear' to my heart -- the abba sound. Even though my 'restoration processor' has been helpful in decoding the ABBA sound, it really isn't the key step. Geesh, even the Carpenters have used some of the processing that my processor can undo -- the compression isn't really the only key to 'the ABBA sound'. I have a formula which is expressible with a SOX/shell script which can be helpful to see what they have done. I am not going to provide it unless there is a lot of interest (it isn't long, but it is technical.) The key has been related to applying a 90deg phase shift and a 10msec time delay. The 90deg phase shift kind of allows mixing the sound in a way that it doesn't cancel, and the 10msec delay causes the very dense sound when summed back to the audio. I haven't derived the exact formula, but I created a decoding version that does work in most cases (as you can hear by some of my examples.) It is important to start with a pristine copy or the math probably won't work -- so the Polar releases or other obscure releases which weren't mis-processed are probably your best bet. I have an ancient, non-english release which contains about the cleanest copies that I have heard -- and that is what I use, very likely because the cost of finalizing that version wasn't deemed to be worthwhile. I haven't tried the Polar release yet for the test. Even though I don't know their exact procedure, it is probably like: (take the signal delayed to match the hilbert 90deg transform -- but don't use the transform), then (take the signal witih hilber transform PLUS 10msec), add the signals together -- and voila -- you have close to the ABBA sound. I haven't checked the 'encoding' procedure, but I am using a similar inverse with good results. These tests and associated scaling parameters take time to find out (by trial and error), so that is the only reason why I haven't provided an exact encoding script. I am willing to share my decoding script, but undoing the general steps above are the way that I have cleared up the voices. It is NOT fun using these tools, but it is nice to be able to hear the voices so clearly. ADDON: Just tried the Polar release -- there are some hiccups, but comes close to working okay. The biggest problems is that they did some enhancement and some compression which messes up some of the math, but the algorithm still seems to apply. So, to sum it up -- the key to that 'denseness' in the ABBA sound APPEARS to be related to careful using of a 90deg phase shift and 10msec delay. In the olden days, the 90deg phase shift circuitry was used in SQ/QS and other quad encoding schemes. So, IT JUST MIGHT have been implemented by using a quad encoder partially. I am not sure, but I do have some very pretty results removing some of that aggressive denseness in some examples. The examples are files that have the word 'clean' in them. I also have a cleaned up Dreamworld that uses some of the same techniques (my original copy is SUPER compressed and very smeary sounding -- challenging to clean up.) spaces.hightail.com/space/pG4t4ZFnyBJohn
|
|
|
Post by jsdyson on Jan 18, 2018 21:50:13 GMT
Okay -- will do.
|
|
|
Post by jsdyson on Jan 18, 2018 14:54:20 GMT
Really good news -- my site as described above has a much better copy of Dreamworld. I found the correct settings on the processor to get it to REALLY clean up Dreamworld. It required no magical multi-pass operations, and I got the processor to adapt automatically (after specifying a higher/faster expansion mode) to the highly compressed and shrill original copy that I have. The processor simply needed some relaxation of some artificial internal limits. It (my copy of Dreamworld) isn't perfect, but is amazingly close to correct!!! ONE ITEM OF NOTE -- some of the copies that I have produced might benefit from a 9kHz or 12kHz lowpass filter (single pole.) In sox, just specifying 'lowpass -1 9k' or 'lowpass -1 12k' might be beneficial. There is ZERO attenuation of the HF in my copies, and so might tend to sound a little 'metallic'. People are often used to a lot of final processing -- perhaps to soften the sound a little or compress. There is no compression in most of my demos except for what was done in the studio. Frankly, some of the studio compression is also obliterated by the 'restoration processor.' Also -- just a few minutes ago (this is an additional note), I added a slightly prettier version of Dreamworld, called Dreamworld-remix. It is slightly different, where the spatial arrangement is a little different and a different EQ that makes it a little more like a Merry-go-round in sound. Also, I added a compressed version, so it sounds a little more typical-ABBA, but still with some 'merry-go-round'sound. Here is the site again: spaces.hightail.com/space/pG4t4ZFnyBJohn
|
|