Digital Props Part II

[fas42]

First of all, I don’t believe you can uncompress a compressed file. The information that was on the waveforms that were compressed are “damaged” and without knowing what compression algorithm was used, your chances of “restoring” the dynamic range are slim to none. Even if you could prove that you could restore the dynamic range correctly and you wanted to use this forum to market your new found skills, I would think that you would need permission from Amir and Steve to do that.

I thought you were hard at work on building your 2.4kW amp. What happened to that effort?

First of all, I will have to disagree with yourself and Bruce, sorry! There are many ways to skin a cat, and I did say it would not be simple, it's not just a case of running it through a black box and out pops a completely pristine copy of an unadulterated version. Digitised music is just data, computer data in fact, and I've spent 30+ plus years playing with tools that allow one to pull data apart, work out exactly what's in there, and then put it back together again in some new combination of the elements.

I most certainly would not expect to use this forum for doing such an exercise, once working as a practical mechanism I would set up a completely independent website to implement the whole "business". The idea of doing such a thing only popped into my head a couple of weeks ago, and you just gave me the opportunity of mentioning it, to throw the idea out there.

Just to give some more clarity to the concept, if people offered a track to be sorted out, I would do a first cut of a fixup, get the owner's feedback by letting them play with a sample of the repair, do another round as a trial, and so on until the other person was happy. Only then would there be monetary consideration involved. And as a bonus I would include a group of settings for some readily available compression tool which you could apply to the restored track, which would spit out something very, very close to the original, compressed track.

The 2.4kW amp is the other thing that I'm doing at the moment, while I keep a lookout for intelligent postings here , I'm thrashing away running simulations on the design to date. A device like this is also not trivial, voltage spikes of over 400V, and current pulses of the order of 500A can be provoked very easily in the model, so the real thing needs to be just right, like a bowl of porridge ...


Frank

 

[mep]

Thanks Bruce. I appreciate the confirmation.

 

[mep]

First of all, I will have to disagree with yourself and Bruce, sorry! Digitised music is just data, computer data in fact, and I've spent 30+ plus years playing with tools that allow one to pull data apart, work out exactly what's in there, and then put it back together again in some new combination of the elements. I most certainly would not expect to use this forum for doing such an exercise, once working as a
Frank

It might be a "new combination of the elements," but it won't be what it was before the compression was added.

 

[Phelonious Ponk]

we are not talking about the theoretical dynamic range of CDs,

You keep repeating this as if saying it often enough will make it credible. The dynamic range of CDs is not theoretical. Neither is it limited by the current fashion of compressing the life out of popular recordings. It is limited to the dynamic range of those recordings on those recordings. That has nothing to do with it's capabilities. But even if you stop listening to compressed pop, you'll find few recordings that use the full dynamic range of CD. There just isn't much music that needs all of it. And that doesn't make it theoretical either. Or at least no more theoretical than the FR extension of your Krells beyond the limits of your hearing.

Tim

 

[mep]

You keep repeating this as if saying it often enough will make it credible. The dynamic range of CDs is not theoretical. Neither is it limited by the current fashion of compressing the life out of popular recordings. It is limited to the dynamic range of those recordings on those recordings. That has nothing to do with it's capabilities. But even if you stop listening to compressed pop, you'll find few recordings that use the full dynamic range of CD. There just isn't much music that needs all of it. And that doesn't make it theoretical either. Or at least no more theoretical than the FR extension of your Krells beyond the limits of your hearing.

Tim

I think we are more in agreement than disagreement Tim. When I use the word “theoretical,” I don’t want people to assume that because redbook CD is capable of 96dB of dynamic range that all CDs have 96dB of dynamic range. Some people are easily confused/mislead. The bottom line is that most recordings don’t come close to using half of the dynamic range that redbook CD is capable of.

 

[Ron Party]

Tim, it's a bit of both. Yes the theoretical dynamic range of CDs is a given, and yes it is considerably higher than vinyl or tape. Now having stated this, I'm not aware of even one CD that does employ that full theoretical dynamic range. ~85dB is the highest dynamic range on a commercial CD of which I've ever heard. Now, of course, play that CD at reference volume and you'll have crest factors which will test virtually any system.

As to the other issue discussed above, I believe some folks at hydrogenaudio have been playing with some software which, for lack of a better word, de-compresses (or un-compresses) compressed music. I haven't been following those discussions in detail, so I can't write to its effectiveness. To the extent it is doable it seemingly would not matter if the compressed music was in digital or analog format.

 

[fas42]

It might be a "new combination of the elements," but it won't be what it was before the compression was added.

I was using that language in an abstract sense, a restored track will certainly match up with the compressed one, from the point of view of the person listening to it. Remember, this is a customised version of a waveform which in one sense has never existed, so it is up to the end user to say whether it's correct or not, he's the one that's "paying" for it ...

I do note, Mark, that you are remarkably quick to proclaim that things are impossible: good thing you weren't around when Kennedy proclaimed something about getting to a lump of rock in the sky, otherwise the Ruskies would be the ones with bits of rubbish lying around up there ...

Frank

 

[fas42]

I think we are more in agreement than disagreement Tim. When I use the word “theoretical,” I don’t want people to assume that because redbook CD is capable of 96dB of dynamic range that all CDs have 96dB of dynamic range. Some people are easily confused/mislead. The bottom line is that most recordings don’t come close to using half of the dynamic range that redbook CD is capable of.

As I've mentioned before, I have a test CD with the same track of a classical snippet at 0dB, then attenuated 20, 40, and 60 dB. The last is reduced in volume a thousand times! With the volume on maximum, and my ears glued to the driver cone (where have I heard that before ...?) I can hear the last version, way, way, in the distance, and the music is all there! It doesn't sound like a distorted, digital mush, it sounds just like a piano, orchestra, cello, and just like the 0dB version ...

Frank

 

[Bruce B]

As to the other issue discussed above, I believe some folks at hydrogenaudio have been playing with some software which, for lack of a better word, de-compresses (or un-compresses) compressed music. I haven't been following those discussions in detail, so I can't write to its effectiveness. To the extent it is doable it seemingly would not matter if the compressed music was in digital or analog format.

Ron... if such an animal existed, every mastering engineer in the world would have at least 2! You can not "de-compress" music... either in the analog or digital domain. As Mark eluded to, it might be a "new combination of elements", but it would NEVER be as good or even close to the original file. I've been at this over 20yr, and can compare the original to the compressed. We have upward and downward compressors/expanders, gain changes and the best that money and research can buy. Not even close.

 

[Phelonious Ponk]

First of all, I will have to disagree with yourself and Bruce, sorry! There are many ways to skin a cat, and I did say it would not be simple, it's not just a case of running it through a black box and out pops a completely pristine copy of an unadulterated version. Digitised music is just data, computer data in fact, and I've spent 30+ plus years playing with tools that allow one to pull data apart, work out exactly what's in there, and then put it back together again in some new combination of the elements.

I most certainly would not expect to use this forum for doing such an exercise, once working as a practical mechanism I would set up a completely independent website to implement the whole "business". The idea of doing such a thing only popped into my head a couple of weeks ago, and you just gave me the opportunity of mentioning it, to throw the idea out there.

Just to give some more clarity to the concept, if people offered a track to be sorted out, I would do a first cut of a fixup, get the owner's feedback by letting them play with a sample of the repair, do another round as a trial, and so on until the other person was happy. Only then would there be monetary consideration involved. And as a bonus I would include a group of settings for some readily available compression tool which you could apply to the restored track, which would spit out something very, very close to the original, compressed track.

The 2.4kW amp is the other thing that I'm doing at the moment, while I keep a lookout for intelligent postings here , I'm thrashing away running simulations on the design to date. A device like this is also not trivial, voltage spikes of over 400V, and current pulses of the order of 500A can be provoked very easily in the model, so the real thing needs to be just right, like a bowl of porridge ...


Frank

"Theoretically," I think Frank is right. It's all data. It might take a billion Einstein monkeys typing code for 100 years, but it should be theoretically possible to reverse the data that compressed the master and restore the dynamic range. Do it, Frank. Forget the 2400 watt amp. That's just another amp. This is the greatest gift to audiophilia you could possibly give. Write this software. Write a Mac version. I'll beta test it for you.

Tim

 

[Orb]

Ron... if such an animal existed, every mastering engineer in the world would have at least 2! You can not "de-compress" music... either in the analog or digital domain. As Mark eluded to, it might be a "new combination of elements", but it would NEVER be as good or even close to the original file. I've been at this over 20yr, and can compare the original to the compressed. We have upward and downward compressors/expanders, gain changes and the best that money and research can buy. Not even close.

I have seen the conversations Ron is commenting about on that forum.
I have to say when it comes to this type of technology HydrogenAudio is a great forum as it has some very experienced DSP engineers, downside though is sifting through the real experts to those with new academic qualifications or engineers with hoby interest.
So it helps to have a very indepth knowledge in the 1st place and be able to categorise the responses.
That said there are some great posters over there.

Bruce, from what I can remember there are tools/ways to provide a better quality dynamics in terms of loudness, but as you say it is not to the audio standard-quality you could use for commercial studio releases.
But some/many might find the quality improvements in dynamics outweigh the fact it is not 100% accurate; perception and tolerance/threshold/expectation of quality for the listener the key whether this is satisfactory or not IMO.

Cheers
Orb

 

[fas42]

Thanks very much for that vote of confidence, Tim, it's exactly what I was looking for. It's enough to get me cracking, and as I said earlier, I've got a doozy of a track to work on.

As to what Bruce is alluding to, I agree that there is not a brilliant chance of getting a precise recovery, to the last dB of what may have existed at some point. No, what we are talking about is a RESCUE, not a perfect reversal: turning musical tracks which are obnoxious or overwhelming in their presentation into a version which is sane, has musical integrity and appropriate, human friendly dynamics for the style. We are assuming that no original, uncontaminated master exists, otherwise what would the point of the exercise be, so ultimately the version that is served up to someone is one that makes musical sense to them. As such, it would have to be an iterative process, playing with all the variables until a version that is good enough, without anything that was there at the beginning being lost, is served up.

What I have is "In the Summertime", by Thirsty Merc, used as theme music for a local TV show about beach rescues. Good, straight up and down garage rock, that wouldn't be out of place pumping out of a car radio in the 70's, some very punchy kick drumming in it. If someone has got something they think would be even better to start with, give me a yell.

I could post bits of the original -- what's the copyright situation on this? --- and decompressed versions as progress is made, for feedback from everyone. All thoughts and contributions appreciated ...

Cheers,
Frank

 

[Phelonious Ponk]

Thanks very much for that vote of confidence, Tim, it's exactly what I was looking for. It's enough to get me cracking, and as I said earlier, I've got a doozy of a track to work on.

As to what Bruce is alluding to, I agree that there is not a brilliant chance of getting a precise recovery, to the last dB of what may have existed at some point. No, what we are talking about is a RESCUE, not a perfect reversal: turning musical tracks which are obnoxious or overwhelming in their presentation into a version which is sane, has musical integrity and appropriate, human friendly dynamics for the style. We are assuming that no original, uncontaminated master exists, otherwise what would the point of the exercise be, so ultimately the version that is served up to someone is one that makes musical sense to them. As such, it would have to be an iterative process, playing with all the variables until a version that is good enough, without anything that was there at the beginning being lost, is served up.

What I have is "In the Summertime", by Thirsty Merc, used as theme music for a local TV show about beach rescues. Good, straight up and down garage rock, that wouldn't be out of place pumping out of a car radio in the 70's, some very punchy kick drumming in it. If someone has got something they think would be even better to start with, give me a yell.

I could post bits of the original -- what's the copyright situation on this? --- and decompressed versions as progress is made, for feedback from everyone. All thoughts and contributions appreciated ...

Cheers,
Frank

It wasn't a vote of confidence in anything more then the theoretical possibility, Frank. I don't, in my wildest dreams, believe you can pull it off. And of course Bruce is right;what could be "recovered?" The brick-walled master, FWIW, is the master. It is the artistic intent, or the resignation, at least. Anything done to reverse the compression would be "inaccurate" by definition. Of course that doesn't mean it wouldn't be better.

Tim

 

[JackD201]

It would be nice if you guys add "signal" or "data" everytime you guys use the term compression. It's getting confusing.

 

[FrantzM]

Hi

The discussion is moviing away from the OP and that at times is good. I must say I have heard the CD in question and the Boz Scag in particular is stunning. It would be one of those cuts that ptove how good digitlal is and how the MAstering/.Recording are responsible for the final sound. There is no lack of "body" or "density" on this CD... Anyone rejecting digital en masse would have an ear opener listening to this CD... It would sure tone down some exreme anti-digital positions , I have seen here (not even clost to have "body", "flat instruments", "lack of dimensionality" ...) ,, If such are your conclusions upon listening to this CD it may be time to acquire a new digital gear or a new system altogether.

 

[fas42]

Jack, it's all about the "signal" side of things, I'm not going to worry about "data" -- MP3 and similar lossy discarding of information -- that's nowhere near the problem the fiddling with waveform amplitudes causes for listeners.

Frank

 

[Phelonious Ponk]

It would be nice if you guys add "signal" or "data" everytime you guys use the term compression. It's getting confusing.

FWIW, Jack, even when I said data, I was talking about audio compression not codecs. But digital (signal) compression all happens at the data level; it should theoretically be reversible at that level.

Tim

 

[Bruce B]

As to what Bruce is alluding to, I agree that there is not a brilliant chance of getting a precise recovery, to the last dB of what may have existed at some point. No, what we are talking about is a RESCUE, not a perfect reversal: turning musical tracks which are obnoxious or overwhelming in their presentation into a version which is sane, has musical integrity and appropriate, human friendly dynamics for the style.

We "rescue" all the time. All you do is lower the gain of the file 3-6dB and use upward/downward compression/expansion of the Weiss DS1/DNA1 and the TC 6000
One of the plugins that can do it is the Flux Alchemist

 

[Ron Party]

Bruce, I like your use of the term *rescue*. IYO how effective are your rescue efforts? Just wondering because, in light of the fact that so much of what is released today is unduly compressed, if one could undertake these rescue efforts at home, well ... the benefit is obvious.

 

[Bruce B]

Bruce, I like your use of the term *rescue*. IYO how effective are your rescue efforts? Just wondering because, in light of the fact that so much of what is released today is unduly compressed, if one could undertake these rescue efforts at home, well ... the benefit is obvious.

We usually get good results. Most of the time it's used on remasters where the only copy you have of the album is the actual CD, Exabyte tape or MO disc. We put the file into the workstation, lower the gain 3-6dB and then run the tune through an ambience recovery system, then either the Weiss or TC and on to the analog console and capture at a higher resolution.