yes you did. I thought it had enough info to add here. esp regarding 24/96K. We dont know which remastering had dithering or if a compressor was used.
But already, even this subject is getting far behind the curve -- a new Standard of codex is about to be unleashed on us by 90% of the mfr:
DTS:X - Raising the Bar in Immersive Sound | DTS
To be launched next month. Now what will This lead to? Doesnt sound like a HiRes attempt. An expander?
THx-RNMarsh
Last edited:
Richard,
I posed the question days ago about using an expander to restore some of the dynamic range removed by over zealous use of compression, I got no replies to that inquiry. The problem is if there is enough differential in the digital domain to actually accomplish that. If everything is brought up to a very narrow dynamic range then it would seem a fools errand if the range was very narrow. I would think you would need some minimum dynamic range to be able to expand the dynamic rnage without causing some very weird artifacts.
I posed the question days ago about using an expander to restore some of the dynamic range removed by over zealous use of compression, I got no replies to that inquiry. The problem is if there is enough differential in the digital domain to actually accomplish that. If everything is brought up to a very narrow dynamic range then it would seem a fools errand if the range was very narrow. I would think you would need some minimum dynamic range to be able to expand the dynamic rnage without causing some very weird artifacts.
I never saw it. But I just wonder if this is what the new codec is going to be? Leaving the compressed data for radio etc and in home expand it back to something close to normal again. But, I still like the straight-ahead 24/96K HiRes download.
THx-RNMarsh
This can be a somewhat straightforward, or quite complex exercise. Relatively simple if the final mix was where the major compression occurred; far more involved if the sound elements were individually compressed, using different parameters. There would never be a problem of there not being enough differential, the real challenge is to guesstimate the parameters used to do the compression at the time - a type of reverse engineering has to be done.
Arbitrarily expanding will not work - it will sound weird, and wrong, even if the original compression was relatively mild, with poor expansion parameters selected.
If all the original tracks were separately compressed then it gets a lot, lot harder - the sound elements would need to be separated, a largely unsolved challenge still, and then individually expanded to the appropriate settings; and finally the mix reconstituted.
This will all certainly happen down the track - just requires someone or a group who are sufficiently motivated ...
We do not live in the world of primitive analog compressors/expanders anymore. Once digital dynamic and lossy compression is applied to pop music, the content is lost.
You guys need to listen to classical music in case you want to speak about good sound, good recordings and hi-fi. For today's pop music production and re-masters, even 8 bits + dither is more than enough.
You guys need to listen to classical music in case you want to speak about good sound, good recordings and hi-fi. For today's pop music production and re-masters, even 8 bits + dither is more than enough.
But why other than the odd greybeards like Pink Floyd is no one really embracing 5.1 or the up coming atmos. I would have though the pot jockeys would love the idea of being able to accurately place a sound in a 3D space knowing there is more than a fighting chance of reply returning that sound. Or has the LCD of the ear buds killed that until someone comes up with an easy way of creating binaural from mono tracks in the studio?
Lossy compression, yes, dynamic compression, no. Information is not being discarded in the latter, it's being distorted, in shape. If enough clues are analysed then the distortion algorithm can be deduced, and untwisting of the twisted can be applied. It may not be a perfect reconstruction, but the "shape" can be greatly improved over the original.
If the compression is done to all signal, it is possible to decompress it to a certain degree. But because it is not so, I would prefer to say that the content was "lost" during the compression. You don't know which one is a real 90dB and which one is a result of compression.
As a handwaving explanation, dynamic compression of audio is like taking a sloping straight line, selecting a point somewhere near the top of the slope, and bending the line at that point - the line has a kink, or "knee" in it. If you can work out, after the fact, where that knee was inserted, and how much it was bent, then you can straighten out the line, virtually perfectly. Yes, it's almost guaranteed that some slight increase in noise levels will occur, but the actual gains in restoring close to the original will far outweigh those slight losses.
I did an interesting exercise some years ago, where I manually closely examined the waveform of a piece of compressed pop, which was highly objectionable because of its shouty, compressed nature. It was possible to see the compression algorithm "working" on the signal, and by a bit of deductive backtracking make some intelligent guesses as to the parameters used. A couple of iterations of more precisely adjusting those parameters, and a very promising, expanded waveform popped out. And when played back, yes, it was chalk and cheese - the restored track sounded like ordinary musicians playing in a studio, versus the hyped up, ranting intensity of the original.
Now, this was done by eye, and a bit of intuition, and trial and error - but it would only be an extension of what I did, to begin to automate that analysis, and reversal of compression. No significant content was lost, just the "squashing" was reversed.
Yes, it's theoretically possible to go completely nuts with compression, doing it in very complex ways - but I would estimate that it should be possible to undo most of it, provided the motivation was there to put the effort in.
I did an interesting exercise some years ago, where I manually closely examined the waveform of a piece of compressed pop, which was highly objectionable because of its shouty, compressed nature. It was possible to see the compression algorithm "working" on the signal, and by a bit of deductive backtracking make some intelligent guesses as to the parameters used. A couple of iterations of more precisely adjusting those parameters, and a very promising, expanded waveform popped out. And when played back, yes, it was chalk and cheese - the restored track sounded like ordinary musicians playing in a studio, versus the hyped up, ranting intensity of the original.
Now, this was done by eye, and a bit of intuition, and trial and error - but it would only be an extension of what I did, to begin to automate that analysis, and reversal of compression. No significant content was lost, just the "squashing" was reversed.
Yes, it's theoretically possible to go completely nuts with compression, doing it in very complex ways - but I would estimate that it should be possible to undo most of it, provided the motivation was there to put the effort in.
Last edited:
I wonder in what state should be the 45 years old original master tapes ?
With the consequences of demagnetization, chemical changes of the support (loss of flexibility, loss and migration of magnetic particles, lubrication of the magnetic surface), pre and post echoes etc...
We don't have to look elsewhere if we found some digital re-mastering to sound bad. Certainly not the fault of the digital process.
Oh, really ?
Last edited:
Look, it was never as easy as now, in the era of DSP's, to manipulate with sound - noise reduction, filtering of all kinds, pops and clicks suppression, amplification, level equalization, soft clipping, hard clipping - you name it, tens of other effects. It is a temptation to manipulate with sound, and new releases of old recordings are usually worse than originals and they are prepared for use in subway train with your iPhone and headphone sets. Possibilities are great, but the resulting sound is not. Remember how you can hear filter ringing when noise reduction is applied to jazz files from thirties or forties, just one example from many.
Good recordings of classical music are the only exception and they profit from high resolution, low noise and low distortion available.
Good recordings of classical music are the only exception and they profit from high resolution, low noise and low distortion available.
I believe producers and record companies have a minimal sense of the historical and artistic value of their catalogs. So, if they decide to restore an old recording, i'm sure they will 'manipulate' as little as possible.
I own some old original masters from the seventies, and i can tell-you that, when i decided (a decade ago) to save them on Hard disk, it was a real nightmare, listening to them, to discover how they were damaged by the years.
And i had a lot of work in order to make them at least 'audible'.
Not necessarily. It depends on the intended audience. If you remaster a 1950-ies recording for re-issue in 2015 you should at least make sure that the 2015 audience really likes it and buys it, otherwise the whole exercise is irrelevant.
If you want to re-issue historical events like, say, early speeches from Mr. Stalin, you better manipulate it so that the intended audience at least understands what's being said.
Remastering /reissuing in and of itself is useless - there must be an audience for which you do it. And that audience determines the 'kind' of remastering.
Jan
- Status
- Not open for further replies.