atici wrote:The author of the article mentions about "severely clipped transients" and the waveform of the CD layer having squared-off shape at the peaks (which is a result of peak limiting). I'd like to hear your opinions about how harmful this is to the fidelity.
It is a nice article. Now could someone explain why "peak limiting" process damages the sound and how much? I use the Waves L2 Ultramaximizer plug-in at home (which is essentially a peak limiter and a dithering & noise shaping tool). I wrote about this here before. And when I read about what it does in its manual it seemed to me some kind of local replaygaining using some sort of look-ahead to figure out the peaks that should not affect the average sound volume. Therefore you can set the avergae volume to be at your regular listening level without worrying about the peaks. Of course how agressive it would be is up to you because you deterimine when those peaks are reached. The author of the article mentions about "severely clipped transients" and the waveform of the CD layer having squared-off shape at the peaks (which is a result of peak limiting). I'd like to hear your opinions about how harmful this is to the fidelity.
Remind me why we need 24 bits of resolution on DVD-A....
And since dynamics compression brings everything up to a normal volume, there will be no need for 24-bit's added dynamic range anyway. And the reduced quantisation noise will be drowned out compared to the distortion that extreme compression makes.
Hmm but in 24 bit audio after peak limiting there will be less problems related to the cutoff because the range is higher so the peak limiter does not need to be that agressive. Isn't it the case? (Assuming you listen at the same volume) Because after all the reason one uses peak limiter is the limited dynamic range to prevent clipping, otherwise a simple change in volume would solve it.
Is it indeed possible to ABX in a professional test setting between 24bit/96kHz and 16bit/44.1kHz downsample of the same song? It should depend on the sample but I think most members in HA think 24bit/96kHz is overkill and could not be ABXed.
The range of a 24-bit recording is no "louder" than a 16-bit recording, it is just more accurate. As I said above, the peak value on a CD should be no louder than the peak value on a DVD-A.You're assuming each amplitude level (65,536 for 16 bits, 16,777,216 for 24 bits) represents the same change in volume, which it doesn't.
It's not a format change that's nessesary to solve the overcompression problem, it's the whole cycle that needs to stop.
Peak limiting has nothing to do with resolution.
Loudness just depends on how you set your amp volume level, nothing to do with the nº of bits.
With 16 bits, noise floor can be as low as around -94 dB below full scale. With 24 bits, noise floor can be as low as -144 dB below full scale, but in practice quite less due to real world electronics.
QuotePeak limiting has nothing to do with resolution.I meant the effective resolution, i.e. it sounds like... Squezzing the most of our bits.
QuoteLoudness just depends on how you set your amp volume level, nothing to do with the nº of bits.So doesn't a given CD have intrinsic sound level? That contradicts same recordings having different volumes (take this album for instance).
QuoteWith 16 bits, noise floor can be as low as around -94 dB below full scale. With 24 bits, noise floor can be as low as -144 dB below full scale, but in practice quite less due to real world electronics.This is due to the inherent thermodynamic noise in the DACs I guess, right?
atici wrote:but that's why most professional plugins use >16 bit internal processing techniques: mainly to prevent clipping withing the plug-in processing) therefore the volume of SACD should have been higher which is confusing me
16-bit 24-bitmaximum - - - - - - - - - - - - 0 - -
I've compared the "Shine On" boxed set version with the standalone version (still sold in stores today) [...]
Now given any analog wave we aim to quantize it in x-bits. What volume should we set? If we set the volume low then the during quantization we'd lose more information then we otherwise would because our ADC wouldn't be able to distinguish the difference as accurately as it could if we had increased the volume. However we don't want to increase the volume so high so that our signal goes beyound the boundaries of the amplitude range we represent with our x-bits. So the most conservative solution is to increase it so that the highest peak in the signal (during a single mastering cycle) sits at the highest amplitude our x-bit system could represent. Naturally this is not the best solution, because there would be (and there's with most of the music I listen to) some peaks that does not reflect the general flow of the signal.
"The _Shine On_/EMI remasters" ...More recently, all these remastered albums were made available separately; and additionally, most other Floyd albums have been remastered in the same manner. The only exceptions are the compilation album Works and the most recent albums which have no need for touching up (DSoT, TDB, and p.u.l.s.e). These remasters are based on the original master tapes, and were done by Doug Sax (supervised by James Guthrie) at the Mastering Lab, in Los Angeles. They generally represent a higher level of quality than the previous Harvest discs (which in turn were generally superior to the Capitol and CBS discs sold in the US). In addition to the heightened sound quality, the remastered editions feature (in almost all cases) expanded booklets with new artwork and lyrics (even on the early albums!); the discs themselves are all picture discs. NOTE: There has been some disagreement over whether the new EMI discs that have _Shine On_ counterparts are or are not identical. The general consensus is that they are; and if they are not, then they were at least done by the same people, at the same location, with the same equipment, at the same time, and for the same company.
Jon Iverson Adds Some CommentsI was puzzled when reviewing the DSotM disc: the CD layer sounded more aggressive than the hybrid's SACD tracks. Not having access to the test equipment JA has on hand, I chalked the differences up to varying characteristics of the two analog-to-digital converters (one PCM-based, the other DSD) used for each layer and the more laid-back qualities of SACD sound.
Uh? I always thought the "standalone" version and the Shine On version would have be identical, because the standalone versions of the Pink Floyd Remasters say "Digital Remasters (P) 1992", which is the same year the Shine On set was released AFAIK.
That's one thing I don't quite understand either (probably because I lack basic knowledge about the concepts of digital audio). If the maximum volume is the same in 24-bit and 16-bit audio, how can a sample that would be clipped in 16-bit audio not clip in 24-bit? What I'm imagining is something like this (grossly oversimplified):[...]Im thinking that 24-bit audio just allows for more precise sample values, which is correct. But there's something I'm missing - if anybody could enlighten me, I'd be grateful.