Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
The 24 bit debate
#1
There is much information and debate regarding the improvements that can be found by going from 16 to 24 bit music files. Much of what you find on line tends to suggest that 16 bit can perform to the limits of human hearing, making 24 bit rather pointless. I was talking to someone recently who seems to know a lot more about this subject than I do. He suggests that broadly speaking, 16 bit is indeed more than good enough. However, there was one interesting point he described. When music is relayed via 16 bit, at the higher volume sections of the track 16 bit replay is more or less distortion free. (assuming all is done correctly). Counterintuitively, very quiet sections of the track can indeed suffer distortion. Putting this crudely, (I am not an expert on sampling theories) if silence is considered as all zeros, then a very quiet section needs to be conveyed using very few "ones" in the code. Hence, very quiet sections of songs, notes, or whatever can suffer distortion via 16 bit. Of course, this is by definition not that noticeable, because the distortion is only occurring very quietly. But the idea is that this does feature within human perception, so although the music may sound subjectively perfect, there is something at the limits of perception that makes one aware that the music is in some way digitally rendered. With the vastly higher level of data and resolution offered by 24 bit, this issue disappears into what is definitively imperceptible by human hearing.

Now I have no idea if this stands up to scientific scrutiny, but I will say that it does fit with my own observations of 24 bit. In no way can I say that what I have listened to something via 24 bit is a quantum leap in sound quality vs well mastered 16 bit, but there is something I like about 24 bit, its is kind of subconsciously or subjectively (horrible words both in hi fi land) analogue. So maybe 24 bit is taking something away that with 16 bit is there and distracting from the ultimate listening pleasure. Does anyone know the facts?
Reply
#2
I've heard this argument before. The theory is that when playing at very low volumes, you are as it were "throwing away" bits and hence reducing resolution. i think this may be based on a misunderstanding of sampling theory.

I've had mixed experiences with testing red book vs hi-res files. (I'm assuming here that the red book version is a down-sampled version of an original hi-res master; there is of course no point in comparing red book and hi-res files that have been mastered differently. I'm also assuming the tests are double-blind ABX: again, there's no point in doing any other sort of comparison.) In one test of red book vs 24/44.1 I was unable to hear a difference. In another test of red book vs 24/192 I successfully differentiated the files.

There is a school of thought that says high sample rates can cause distortion in playback equipment. Most playback equipment is only designed to work up to about 22kHz.

EDIT Here are the results of the red book vs 24/44.1 test that I referred to:
http://archimago.blogspot.co.uk/2014_06_01_archive.html

Sonos Connect (W4S) > DSpeaker Antimode 2.0 > Sanders Magtech > Martin Logan Montis
Sonos Connect (W4S) > Devialet 200 > Vivid V1.5
Silver Phantoms (just the two)
London
Reply
#3
24 bit is easier for recording, 16 bit is fine for distributing music. With properly dithered and noise shaped systems 16 bit can give more dynamic range that any domestic hifi that I am aware of.
One problem is that most recordings are done at 48, 96 or 192 kHz sampling rates. This means converting to 44 kHz red book standard requires computation. Not all the algorithms used are audibly transparent, so a file converted from 48/16, 48/24, 96/24 or 192/24 which are the pro standards commonly used to 44/16 may or may not sound the same as the original, depending on the quality of the algorithm used in the conversion.
I can hear the differences produced by such different downsampling calculations, though using the best one I could not hear the difference between the downsampled 44/16 and the 96/24 original.
It was most thought provoking.
Actually a signal to noise ratio of 16 bits is pretty impressive for most parts of the analogue chain, and 20/21 bit absolute state of the art, so post DAC very little will be better than 16 bit and nothing better than 21-bit.
FWIW
Devialet Original d'Atelier 44 Core, Job Pre/225, Goldmund PH2, Goldmund Reference/T3f /Ortofon A90, Goldmund Mimesis 36+ & Chord Blu, iMac/Air, Lynx Theta, Tune Audio Anima, Goldmund Epilog 1&2, REL Studio. Dialog, Silver Phantoms, Branch stands, copper cables (mainly).
Oxfordshire

Reply
#4
There's a long and interesting (though I gave up reading new posts a long time ago!) on WBF, here:
http://www.whatsbestforum.com/showthread...-different
Reply
#5
(13-Sep-2014, 18:54)Rufus McDufus Wrote: There's a long and interesting (though I gave up reading new posts a long time ago!) on WBF, here:
http://www.whatsbestforum.com/showthread...-different

Yes but the lack of understanding of the majority of posters on the WBF forum makes me lose the will to live. The noise to signal ratio is pitiful.

Reminds me of the phrase "produces more heat than light" on almost every thread I try to read Sad
Devialet Original d'Atelier 44 Core, Job Pre/225, Goldmund PH2, Goldmund Reference/T3f /Ortofon A90, Goldmund Mimesis 36+ & Chord Blu, iMac/Air, Lynx Theta, Tune Audio Anima, Goldmund Epilog 1&2, REL Studio. Dialog, Silver Phantoms, Branch stands, copper cables (mainly).
Oxfordshire

Reply
#6
I think for me it's quite simple. If I'm buying something new, and it's available in 24bit, then why not? It's not going to sound worse, and I have stacks of hard disk space available. (Matt's comment re high sample rates noted) However, rebuying stuff I already have in 24 bit? No, I'd rather spend my cash on something else.
Reply


Forum Jump:


Users browsing this thread: 1 Guest(s)