Forum Home Forum Home > Progressive Music Lounges > Prog Recommendations/Featured albums
  New Posts New Posts RSS Feed - Steven Wilson Insurgentes 2016 remaster
  FAQ FAQ  Forum Search   Events   Register Register  Login Login

Topic ClosedSteven Wilson Insurgentes 2016 remaster

 Post Reply Post Reply Page  <12
Author
Message
JediJoker7169 View Drop Down
Forum Senior Member
Forum Senior Member
Avatar

Joined: May 05 2009
Location: West Coast, NA
Status: Offline
Points: 195
Direct Link To This Post Posted: November 04 2016 at 20:53
Originally posted by Catcher10 Catcher10 wrote:

What probably needs to be scrutinized is the measuring process of this database, maybe that is what Dean is pointing out....dunno.

These DR numbers were generated using the Tischmeyer Technology (TT) Dynamic Range Meter offline algorithm, either in its standalone application version or as a plugin for some other piece of software (foobar2000, for instance). Friedemann Tischmeyer, who developed the algorithm and software with Algorithmix, is a mastering engineer and dynamic range advocate. There are now more standardized measures of loudness using the EBU R128 standard, but the TT Meter's readings are highly comparable to the loudness (LU) range reported by software using the R128 algorithm. This is different from the integrated LKFS/LUFS reading, which is a measure of the average loudness (similar to RMS, reported in negative) against the digital limit of 0dBFS (for fixed-point bit-depth). For both DR and LU (any variant), each unit is equivalent to 1dB. Both use perceptual models of human hearing (read: scientifically derived) to determine the loudness we actually experience while listening.
Back to Top
SteveG View Drop Down
Forum Senior Member
Forum Senior Member
Avatar

Joined: April 11 2014
Location: Kyiv In Spirit
Status: Offline
Points: 20245
Direct Link To This Post Posted: November 05 2016 at 03:00
^I don't believe that one needs a meter to determine loudness and compression. If it's there, most audiophiles will know. How much is irrelevant in my opinion.
This message was brought to you by a proud supporter of the Deep State.
Back to Top
Dean View Drop Down
Special Collaborator
Special Collaborator
Avatar
Retired Admin and Amateur Layabout

Joined: May 13 2007
Location: Europe
Status: Offline
Points: 37575
Direct Link To This Post Posted: November 05 2016 at 03:21
Originally posted by JediJoker7169 JediJoker7169 wrote:

Originally posted by Dean Dean wrote:

This is over-hyped pseudo-scientific bollocks.
No, it isn't.  I'm an audio engineer with a degree in psychology and I can tell you point-blank that it is not pseudo-science, it is not over-hyped, and it is not bollocks.
Oh! We're playing credentials top trumps Big smile ... okay, my go: I'm an Aries with a clean driving licence, I like cats and going for long walks by the river, my favourite colours are black and purple and I don't like pasta and my drink of choice is tea. I'm sorry I've left all my academic and vocational qualifications in the pocket my lab coat along with my employment record but not wanting to get into bragging contest to see which of us can piss the highest up a wall, I can assure you yours didn't scare me much. And DR numbers are still over-hyped pseudo-scientific bollocks.
Originally posted by JediJoker7169 JediJoker7169 wrote:

Originally posted by Catcher10 Catcher10 wrote:

What probably needs to be scrutinized is the measuring process of this database, maybe that is what Dean is pointing out....dunno.

These DR numbers were generated using the Tischmeyer Technology (TT) Dynamic Range Meter offline algorithm, either in its standalone application version or as a plugin for some other piece of software (foobar2000, for instance). Friedemann Tischmeyer, who developed the algorithm and software with Algorithmix, is a mastering engineer and dynamic range advocate. There are now more standardized measures of loudness using the EBU R128 standard, but the TT Meter's readings are highly comparable to the loudness (LU) range reported by software using the R128 algorithm. This is different from the integrated LKFS/LUFS reading, which is a measure of the average loudness (similar to RMS, reported in negative) against the digital limit of 0dBFS (for fixed-point bit-depth). For both DR and LU (any variant), each unit is equivalent to 1dB. Both use perceptual models of human hearing (read: scientifically derived) to determine the loudness we actually experience while listening.
I have no issue with the measurement process, (though again, the credentials of the participants fail to impress me, even so-called experts get it wrong), DR numbers are relative measurements and my only real issue with them is they are by that very virtue relatively meaningless and quoting them to 2 d.p. is farcical. Dynamic Range measurement (by whatever method) attempts to normalise the complex non-repetitive, non-cyclic waveforms of music, (that not only differ from track to track and artist to artist but from mix to mix), into an empirically quantifiable single value and then use that as a subjective measure of quality. The algorithms used to calculate this magic number are getting (for want of a more apt word) better and they are far more involved than simply calculating the crest factor (too complex to go into here) of a signal by dividing the peak by the rms (again, too complex to go into here), however they are still attempting to quantify values that are continuously varying through out a piece of music into a single comparative value - the "loudness" (once again, too complex to go into here but "loudness" is not a wholly empirical measurement) of any piece of music continuously changes in time and the DR value is an attempt to "average" (not the actual mean value) that out across the measurement duration time into a measure of goodness.



Edited by Dean - November 05 2016 at 03:28
What?
Back to Top
JediJoker7169 View Drop Down
Forum Senior Member
Forum Senior Member
Avatar

Joined: May 05 2009
Location: West Coast, NA
Status: Offline
Points: 195
Direct Link To This Post Posted: November 05 2016 at 04:08
Well, Dean, it sounds to me like you should know better and simply don't want to believe the empirical work done by the audio engineering and broadcast societies the world over. Fortunately, it's still true whether you believe it or not. Integrated loudness and loudness range (which are well-defined in the R128 white paper) has a profound psychological and physiological effect on music listening. Differences in level of one or even one half dB are easily heard by even an untrained ear in a controlled environment. Does a complex musical signal and a different type of difference—one of loudness range rather than sound pressure level—make it more difficult to both detect and study? Yes, of course. Nevertheless, the research is fairly clear that a loudness difference of a couple dB (not at two decimal points; where were you getting that from, the peak values?) will affect human hearing and music listening enjoyment. You can argue that enjoyment is subjective, and I would argue that subjectivity is irrelevant given a large enough sample size.
Back to Top
SteveG View Drop Down
Forum Senior Member
Forum Senior Member
Avatar

Joined: April 11 2014
Location: Kyiv In Spirit
Status: Offline
Points: 20245
Direct Link To This Post Posted: November 05 2016 at 04:31
I'm a retired tomato farmer that helped to invent a hybrid zucchini.
This message was brought to you by a proud supporter of the Deep State.
Back to Top
Dean View Drop Down
Special Collaborator
Special Collaborator
Avatar
Retired Admin and Amateur Layabout

Joined: May 13 2007
Location: Europe
Status: Offline
Points: 37575
Direct Link To This Post Posted: November 05 2016 at 04:50
Originally posted by JediJoker7169 JediJoker7169 wrote:

Well, Dean, it sounds to me like you should know better and simply don't want to believe the empirical work done by the audio engineering and broadcast societies the world over. Fortunately, it's still true whether you believe it or not. Integrated loudness and loudness range (which are well-defined in the R128 white paper) has a profound psychological and physiological effect on music listening. Differences in level of one or even one half dB are easily heard by even an untrained ear in a controlled environment. Does a complex musical signal and a different type of difference—one of loudness range rather than sound pressure level—make it more difficult to both detect and study? Yes, of course. Nevertheless, the research is fairly clear that a loudness difference of a couple dB (not at two decimal points; where were you getting that from, the peak values?) will affect human hearing and music listening enjoyment. You can argue that enjoyment is subjective, and I would argue that subjectivity is irrelevant given a large enough sample size.
Okay, I made a small joke at your expense but let's not get too personal here, I am permitted to have a difference of opinion that has nothing to do with belief or knowing better and more to do with knowledge and experience. Not every engineer has bought into the "loudness wars" - yes we can tell when something is over-compressed and yes the mastering process will involve some level of compression but creating an empirical measure for that is not quite as simple as it seems and using that as a comparative is fraught with issues and problems. We engineers can produce a wealth of numbers based upon all manner of metrics and these are by definition empirical, it is their use and misuse that determines whether they are meaningful practical tools or meaningless marketing tools and once that threshold has been crossed their empirical value is lost. It's not about the numbers - it is about how it actually sounds to the listener (which is the point Steve rightly made).

While the threshold of human hearing is generally stated at 3dB this is a rounded average so sure, some can discern a couple of dB difference under very controlled conditions but no one can really do much better than that. When I see any dB value (peak, RMS or whatever) expressed to 2 d.p. I reserve the right to regard that with an air of caution. 

What?
Back to Top
JediJoker7169 View Drop Down
Forum Senior Member
Forum Senior Member
Avatar

Joined: May 05 2009
Location: West Coast, NA
Status: Offline
Points: 195
Direct Link To This Post Posted: November 06 2016 at 01:25
The peak and RMS numbers are not important to this discussion except in how they are used to calculate the DR number, which, you'll note, is always an integer. LU, on the other hand, are usually measured to two decimal points in the interest of precision. Will a human hear a loudness difference of 0.01 or even 0.09 LU? No, but no one is or was or has been arguing that. But a loudness difference of 1-5 dB? You betcha.

However, if a track has a sample peak value approaching 0dBFS, then it's quite possible that track may have intersample peaks which will clip, causing distortion on playback. So, in that way, a peak value of -0.01dBFS is a useful piece of information in the form of a red flag. Will a human hear a difference between peak values varying by less than a few dB? No. But will a human hear distortion caused by intersample peaks above 0dBFS? Absolutely.

Edited by JediJoker7169 - November 06 2016 at 01:55
Back to Top
JediJoker7169 View Drop Down
Forum Senior Member
Forum Senior Member
Avatar

Joined: May 05 2009
Location: West Coast, NA
Status: Offline
Points: 195
Direct Link To This Post Posted: November 06 2016 at 01:28
Unfortunately, modern mastering practices have lead to many CDs having intersample peak clips, often with noticeable distortion. I do believe the peak meter used in the DR software is true peak aware, because it will report "over" on occasion (though not specifically by how much). And how would it be able to report anything other than a sample peak of 0dBFS otherwise?

I'm aware peak values have no bearing on dynamic range (which is operationalized here as peak-to-loudness ratio or crest factor, in a sense) except as I alluded to in my previous post.

Edited by JediJoker7169 - November 07 2016 at 06:04
Back to Top
Dean View Drop Down
Special Collaborator
Special Collaborator
Avatar
Retired Admin and Amateur Layabout

Joined: May 13 2007
Location: Europe
Status: Offline
Points: 37575
Direct Link To This Post Posted: November 06 2016 at 01:49
Originally posted by JediJoker7169 JediJoker7169 wrote:


However, if a track has a sample peak value approaching 0dBFS, then it's quite possible that track may have intersample peaks which will clip, causing distortion on playback. So, in that way, a peak value of -0.01dB is a useful piece of information in the form of a red flag. Will a human hear a difference between peak values varying by less than a few dB? No. But will a human hear distortion caused by intersample peaks above 0dBFS? Absolutely.
Therein lies the problem with sampling and putting all one's trust in metering, but peak clipping has nothing to do with dynamic range as such. If the signal clips at any time and it is audibly noticeable then that is the the fault of the engineer (who frankly should stop calling himself an engineer because he's obviously nothing more than a bloody technician if he clips the waveform and fails to notice it).
 



Edited by Dean - November 06 2016 at 01:52
What?
Back to Top
 Post Reply Post Reply Page  <12

Forum Jump Forum Permissions View Drop Down



This page was generated in 0.328 seconds.
Donate monthly and keep PA fast-loading and ad-free forever.