PDA

View Full Version : 24-bit audio (new article)



mortslim
2011-02-23, 13:16
Gizmodo has published a new article as to why 24-bit audio is not appropriate for end users:

http://gizmodo.com/#!5768446/why-24+bit-audio-will-be-bad-for-users

ShutterShock
2011-02-23, 13:27
Am I the only one who didn't get the author's point? 24 bit = larger files. Okay... But in terms of the listening experience, I'm not clear on the downside.

maggior
2011-02-23, 13:39
The point is that there really is no advantage for the end listener with 24 bit audio files. This is especially true with the current state of things where albums are being produced with the dynamic range squashed (i.e. the loudness war).

What I would love to see is CD quality (16 bit 44.1 kHz) lossless audio made available for purchase on a wider scale.

There will be those that insist that 24 bit is better than 16 bit. I've read forum threads where people insisted recordings of vinyl made with 24 bit samples sounds better than 16 bit samples. People will hear what they want to hear.

If 24 bit files are the path forward for readily available lossless audio, then so be it. However, it would be overkill IMHO.

maggior
2011-02-23, 13:40
Am I the only one who didn't get the author's point? 24 bit = larger files. Okay... But in terms of the listening experience, I'm not clear on the downside.


There isn't really a downside, it's just that there is no upside, hence the "con".

amey01
2011-02-23, 16:35
The true advantage comes from higher sampling rates. But people seem obsessed with bit-depth as well.

Soulkeeper
2011-02-24, 04:14
People are simply obsessed with numbers. Simple numbers. Which should be as high as technically possible. E.g. the widespread Megapixel obsession when it comes to digital cameras. The Megapixel Myth (http://www.kenrockwell.com/tech/mpmyth.htm)
One dimension is better than two dimensions, so the cameramakers hit gold when they figured they could square the resolution parameters, and print only this number in their ads (ignoring the bit depth, various other digital parameters, and myriads of analog ones).

When the audio equipment makers figure out a way to combine bit depth and sampling rate in one number, we can expect people to start using that number as the ultimate measure of audio quality.

adamdea
2011-02-24, 07:58
"Even CDs are 16-bit, and the sonic quality of a CD is an accepted definition of consumer-worthy HD quality."
That sentence is basically the entire argument

rayman1701
2011-02-24, 08:27
Yeah, I understand the overall limited downside, especially with how the internet in the US is right now, with wildly different speeds and accessibility, it is not in any way practical at the moment or the near future. It can take a couple of hours to download 96/24 files on a decent DSL, and while they were talking about 24bit they didn't have any details on sample rate, so who knows what would actually happen. So it might not work mass market, let alone how the supposed benefits, which I can hear on carefully mastered material, won't make any difference in today's loudness wars quality except maybe they can compress more before clipping. Oh the joy of that thought. I'm not one of those who thinks there is no good new music, I love a lot of new music out there today, I just wish it could be mastered to sound better, not just louder. And I doubt getting iTunes store to go to 24bit would change that in any way. So in the end no, if it happens, there will be no benefit.

However, with all that said, he lost me with the "the noise floor is much lower, and you'll hear a hum in your amp when you turn up the volume" argument. I'm sorry if your amp hums when you turn it up, I say get a working amp not less resolving files. So after that really weak logic, I just found it hard to take him seriously. I mean really, too much dynamic range, too low a noise floor, really that's the best you could come up with for an argument against it?

sebp
2011-02-24, 08:49
The true advantage comes from higher sampling rates. But people seem obsessed with bit-depth as well.
Oh, really?

Technically speaking, 24 bit-depth allows to reproduce the sound of a fly farting, immediately followed by the sound of a thunderclap, both being at relative realistic levels.

Higher sampling rates just allow the reproduction of higher frequencies, which would lead to give these questions an answer :
- Does your audio system is able to reproduce frequencies above 20 kHz?
- Are you really able to hear frequencies above 20 kHz?

My amp and loudspeakers are supposed to be able to reproduce frequencies up to 40 kHz, but I cannot hear a 16 kHz test tone.
So, what's the use?

Pneumonic
2011-02-24, 09:07
I mean really, too much dynamic range, too low a noise floor, really that's the best you could come up with for an argument against it?
I think the guy's point (which admittedly he doesn't seem to explain well enough) is that lowering the noise floor and going beyond the 96 dB dynamic range that redbook now affords, is pointless. At least if one listens to music with today's gear, in real world listening environments, since neither stress the limits of 16 bit resolution.

Pneumonic
2011-02-24, 09:10
Oh, really?

Technically speaking, 24 bit-depth allows to reproduce the sound of a fly farting, immediately followed by the sound of a thunderclap, both being at relative realistic levels.

Higher sampling rates just allow the reproduction of higher frequencies, which would lead to give these questions an answer :
- Does your audio system is able to reproduce frequencies above 20 kHz?
- Are you really able to hear frequencies above 20 kHz?

My amp and loudspeakers are supposed to be able to reproduce frequencies up to 40 kHz, but I cannot hear a 16 kHz test tone.
So, what's the use?
The argument for increased sampling rates comes from the ability to then do away with (supposed sonic degradation) Nyquist filtering issues. Note the inclusion of the word supposed since not everyone has enjoyed 16 bit audio with the same quality and cailbre anti imaging filter designs.

mdconnelly
2011-02-24, 09:21
There is no question that sound quality corresponds to recording quality and attention to detail by the recording engineer and that many 44.1/16 CDs can sound amazing (think XRCD) and that far too many totally suck due to loudness wars and compression.

Be that as it may, when you have an excellent recording, getting a 88/24 or 96/24 version of it and playing it back on a high quality system (and I'm including Squeezeboxen in this), can sound truly exceptional and definitely better than just CD quality.

I have purchased such recordings from a number of sites and I have to say that I've been very impressed. The ones I'm most familiar with include...

http://www.hdtracks.com
http://www.linnrecords.com
http://www.bowers-wilkins.co.uk/Society_of_Sound

Hey, it's not for everyone and you certainly won't hear a difference on computer speakers or cheap playback equipment. And you do need a way to play it back (which, of course, your Touch will certainly accomodate). But given a good recording and a great playback system, there's no going back.

TerryS
2011-02-24, 09:59
Claiming that 16 bits provides 96 dB of dynamic range ignores the fact that the distortion rises as the number of bits decreases. For an analog system, it is OK to define the dynamic range as the ratio of the loudest signal that can be produced to the quietest (or the noise floor). But I have a problem when they apply that same definition to a sampled system. Sure 16 bits implies 96 dB: 20*log(2^16) = 96dB). But that means you are willing to accept absurd amounts of distortion (100%) as a reasonable "signal". The distortion (due to quantization error) increases as the number of bits decreases. With 16 bits, the distortion is very small (0.0015%), but each bit that is removed doubles the distortion. So a signal sampled at 6dB lower than the peak level will have double the distortion. How much distortion are you willing to accept? If you draw the line at 1% distortion (pretty high in my opinion), then that requires 7 bits. That means the dynamic range of a 16 bit sampled system is only 54 dB: 20*log(2^(16-7))= 54dB if you limit the distortion allowed to less than 1%. About as good an old cassette deck. Not my definition of "Hi-Fi". 16 bits was a compromise based on the available technology of the day. Many people still claim it is good enough. Many others don't agree.

Terry

Phil Leigh
2011-02-24, 11:26
"Even CDs are 16-bit, and the sonic quality of a CD is an accepted definition of consumer-worthy HD quality."
That sentence is basically the entire argument

it's also twaddle...

TerryS
2011-02-24, 11:38
it's also twaddle...

Well said....
I couldn't agree more.

Phil Leigh
2011-02-24, 11:48
Claiming that 16 bits provides 96 dB of dynamic range ignores the fact that the distortion rises as the number of bits decreases. For an analog system, it is OK to define the dynamic range as the ratio of the loudest signal that can be produced to the quietest (or the noise floor). But I have a problem when they apply that same definition to a sampled system. Sure 16 bits implies 96 dB: 20*log(2^16) = 96dB). But that means you are willing to accept absurd amounts of distortion (100%) as a reasonable "signal". The distortion (due to quantization error) increases as the number of bits decreases. With 16 bits, the distortion is very small (0.0015%), but each bit that is removed doubles the distortion. So a signal sampled at 6dB lower than the peak level will have double the distortion. How much distortion are you willing to accept? If you draw the line at 1% distortion (pretty high in my opinion), then that requires 7 bits. That means the dynamic range of a 16 bit sampled system is only 54 dB: 20*log(2^(16-7))= 54dB if you limit the distortion allowed to less than 1%. About as good an old cassette deck. Not my definition of "Hi-Fi". 16 bits was a compromise based on the available technology of the day. Many people still claim it is good enough. Many others don't agree.

Terry

Except it doesn't work like that in practice because we don't record at 16 bits or less, we record at 24 (OK let's call it an effective 21 in reality) and dither down to 16, effectively masking (decorrelating) the quantization distortion.

Thus the effect of quantization error on low-level signals is nowhere near as bad as it would be if we were still (1980's) recording at 16 or even 14 bit!


so yes, the 96dB on playback is achievable in theory - even if in practice it is impossible to properly record anything much quieter than (say) -70dB anyway because of ambient and electronic noise ... :-)

TerryS
2011-02-24, 13:28
Except it doesn't work like that in practice because we don't record at 16 bits or less, we record at 24 (OK let's call it an effective 21 in reality) and dither down to 16, effectively masking (decorrelating) the quantization distortion.

Thus the effect of quantization error on low-level signals is nowhere near as bad as it would be if we were still (1980's) recording at 16 or even 14 bit!


so yes, the 96dB on playback is achievable in theory - even if in practice it is impossible to properly record anything much quieter than (say) -70dB anyway because of ambient and electronic noise ... :-)

I have a problem claiming decorrelating when working on random (music) signals. I understand the principle when talking about sinusoidal signals that repeat for some period of time. But I can't see how decorrelation benefits a one time transient signal. Quantization error prevents you from perfectly reconstructing such a signal with any more precision than the quantization step size. Maybe a mental block on my part...

Terry

sebp
2011-02-24, 14:56
There is no question that sound quality corresponds to recording quality and attention to detail by the recording engineer and that many 44.1/16 CDs can sound amazing (think XRCD) and that far too many totally suck due to loudness wars and compression.
I definitely agree.


Be that as it may, when you have an excellent recording, getting a 88/24 or 96/24 version of it and playing it back on a high quality system (and I'm including Squeezeboxen in this), can sound truly exceptional and definitely better than just CD quality.

I have purchased such recordings from a number of sites and I have to say that I've been very impressed. The ones I'm most familiar with include...

http://www.hdtracks.com
http://www.linnrecords.com
http://www.bowers-wilkins.co.uk/Society_of_Sound

Hey, it's not for everyone and you certainly won't hear a difference on computer speakers or cheap playback equipment. And you do need a way to play it back (which, of course, your Touch will certainly accomodate). But given a good recording and a great playback system, there's no going back.
I mostly disagree, now.

My audio system consists of a Transporter, a NuForce IA-7 amp and KEF iQ9 loudspeakers.
Not something I'd consider "cheap playback equipement".

Thanks to B&W's SoS, I downloaded my first 24/48 recording two years ago, a Portico Quartet album, and when I listened to it for the first time, I couldn't help thinking : "Wow, so HD files sound like that? How great!".
But then, out of curiosity, I also downloaded the 16/44.1 version of the same album, listened to it very carefully, and was really surprised I honestly couldn't tell that the 24/48 version sounded better in any way than the 16/44.1 did.

I since have downloaded many other albums from B&W's SoS, going from classical to world music, and each time I compared CD quality vs 24/48, I simply failed to tell that one version sounded better than the other.

I also tried recently to challenge downloaded 24/96 files versus regular CD 16/44.1 rips, with no more success.

In my humble opinion, people who can tell for sure they're able to distinguish 24/96 from 16/44.1 either have not compared versions of the same recording, or are fooled by their brain.

usbethjim
2011-02-24, 18:43
it's also twaddle...

Marvelous word - it's been awhile since I've seen it used. I spewed coffee on the keyboard...

Very amusing
Thanks
Jim

Griffin
2011-02-25, 02:24
In my humble opinion, people who can tell for sure they're able to distinguish 24/96 from 16/44.1 either have not compared versions of the same recording, or are fooled by their brain.

I can tell you the difference can be heard in some very specific cases. But it is so subtle that, when just listening en enjoying the music, the difference can be considered void.

For example, listen to Stevie Ray Vaughan's "Tim Pan Alley". At one point he sings "I heard a pistol shoot - yeah, it was a .44". You can here the different in the instrumental crescendo there if you listen very carefully. The 24/96 has a bit more bite there than it's downsampled 16/44 variant as converted for a SB receiver. But you'ld have to listen very, very carefully to be able to spot the difference. When just enjoying the music instead of searching for these things, thus for all practical purposes, the differences are nil.

To be honest, I wouldn't be able to tell you which version is playing unless I hear the same 10 sec snippet from both version immediately after each other. The difference thus can be heard in an ABX - but it makes no difference in real life.

NB: hardware I checked this on consists of a SB touch & SB duet receiver + XTZ Class A 100 D3 amplifier + Elac FS-247 SE speakers, using digital inputs of the receiver. YMMV.

sebp
2011-02-25, 03:32
The 24/96 has a bit more bite there than it's downsampled 16/44 variant as converted for a SB receiver.
Which could well be related to either the noise shaping algorithm used by SoX for downsampling, or by differences in the way the Receiver and the Touch handle S/PDIF.


The difference thus can be heard in an ABX - but it makes no difference in real life.
I read a very well documented report on a french forum where the tester was able to successfully ABX 16/44.1 and 24/192 on digital silence... until he used proper dithering and noise shaping algorithms to downsample the 24bit/352.8kHz DXD file he was using as a reference.

However, the point is, as you say: it makes no difference in real life.

Mnyb
2011-02-26, 13:11
In my humble opinion, people who can tell for sure they're able to distinguish 24/96 from 16/44.1 either have not compared versions of the same recording, or are fooled by their brain.

Hmm out of my 100 of so dvda i have one where the CD master is very likely the same ?
I think I can here a small improvemement in favor of 24/96 ;) so it is not easy to compare.

Also lets blow the myth that there is an abundance of recordings that can be " saved" by 24 bit , note that we are talking playback now recording should be done at very high sample and bitrates, post processing done with floating piont math and the consumer product can be a 24/96 pcm or 16/44.1 both made with proper dithering etc.
So a good remaster/ mix from multitrack can be a real benefit for all music lovers, but has nothing to do whether it is presented as CD or hirez to us consumers.
So rematered at 24 bit is also very good for the CD produced from it :)

But the intrisinic sound quality of most recordings ever done is not better then the humble redbook CD ?

But personally i act on the small benefit of a doubt I have experienced very good 24 bit. Playback and have a lot of files and DVDA's in the format .

To be awesome a CD master must really be perfect, 24/96 leaves a healty slush margin for errors that in itself can be argument enough to have them.

Some sound engineers may be more of skilled artist than having total awareness of whats really happens inside the shiny tools he uses and can thus be forgiven for minor technical errors that would not be fatal
with a format like 24/96 , but rendeer a CD rather unispiring.

Also DVDA had another thing discrete 24/96 multichannel that is awesome, had to be heard.
It may rise from the dead as dolby true HD for bluray . Dolby true HD is MLP ( Meridian Lossles Packaging aka DVDA ) in a new guisse for bluray and licensed trough dolby

guidof
2011-02-26, 15:38
Hmm out of my 100 of so dvda i have one where the CD master is very likely the same ?
I think I can here a small improvemement in favor of 24/96 ;) so it is not easy to compare.


I have two 24/96 recordings for which I also have the CD version. Of course I have no idea whether each was mastered from the same tape.

I also think that I hear a very small improvement in favor of 24/96, but would find it very difficult, probably impossible, to reliably distinguish it from the CD version in a blind comparison.

Of course, one cannot generalize from a sample of two, but this suggests to me that the sonic benefits of 24/96 recording may be, at best, secondary to the quality of the entire recording set-up (mic positioning and quality, cables, electronics, venue, etc.).

Guido F.

sebp
2011-02-26, 17:20
I think I can here a small improvemement in favor of 24/96 ;)


I also think that I hear a very small improvement in favor of 24/96

You shouldn't be thinking that much, guys.
Beware psycho-acoustics... ;)

sebp
2011-02-26, 17:36
Just to be clear:

I have made some experiments on my system with my ears, where I just could not hear any difference between standard and hires files.
I wish I could, but I honestly couldn't.

This doesn't mean other people won't hear differences.
My point is just that the human brain is a curious thing, and as maggior said already, people will probably hear what they want to hear.

Brain's even such a curious thing that, even if I'm firmly convinced I cannot distinguish 24/48 from 16/44.1, I'm still downloading 24/48 from B&W's SoS.
Go figure... :D

guidof
2011-02-26, 17:55
Brain's even such a curious thing that, even if I'm firmly convinced I cannot distinguish 24/48 from 16/44.1, I'm still downloading 24/48 from B&W's SoS.
Go figure... :D

Perhaps you are not that firmly convinced? Deep, deep down, maybe there is a smidgeon of doubt?

I know there is in my mind . . .

So, I too keep downloading the hi-rez stuff.

Guido F.

sebp
2011-02-26, 18:22
Perhaps you are not that firmly convinced? Deep, deep down, maybe there is a smidgeon of doubt?
No doubt, really, just getting what I've paid for.

Mnyb
2011-02-26, 23:38
Post processing is another factor. And the major factor for me

I do room correction and uses my tone controlls subwoofer fillter and sometimes i use multichannel modes on stereo material ( this works 5.1 can help the stereo illusion imensly, this is another way of figthing your acoustics and win).all this is done digitally in my hifi.
And finaly my speaker xover is digital .

So if all those algorithms can work with higher resolution they will sound better, i think thats contribute to the good sound i get on 24/96 files .

Griffin
2011-02-27, 05:51
I have made some experiments on my system with my ears, where I just could not hear any difference between standard and hires files.
I wish I could, but I honestly couldn't.

I think it's pretty clear the differences are anything but earth-shaking. From my experience, the difference is not there for 99% of the music and in that 1% where it is audible, the difference is so small it's easily missed. Think of it as the need for a car having 10 gears. In most situations it really doesn't make a difference from a standard car having just 5 gears but in some limited very specific cases it may be the possibility of having an extra gear in between the normal 4 and 5 may bring a very small advantage to the table.

jimzak
2011-02-27, 12:02
Apple's sudden interest in 24-bit sound is likely in preparation for new hardware, and where Apple goes, folks tend to follow.

I like the idea of 24-bit files, but my equipment and ears cannot distinguish it from 16-bit.

Perhaps Apple can lure a few young folks away from white earbuds to audiophile equipment. I see that as a good thing.

I'll stick with SB though.

Wombat
2011-02-27, 12:57
Perhaps Apple can lure a few young folks away from white earbuds to audiophile equipment.

I bet they saw services like HDTracks taking some of the typical iTunes users away, thats all.

aubuti
2011-02-27, 15:14
Perhaps Apple can lure a few young folks away from white earbuds to audiophile equipment. I see that as a good thing.
Me too, but I'm a bit puzzled by the leap to 24-bit. I mean, does iTunes store even offer anything lossless yet, even in 16-bit?

Mnyb
2011-02-27, 15:40
Me too, but I'm a bit puzzled by the leap to 24-bit. I mean, does iTunes store even offer anything lossless yet, even in 16-bit?

Worst case scenario is that they are developing some mongrel hirez aac format ? 24/96 but done with perceptual coding so the filesize is at CD size or something.

That would surely sound good enough, but leave you with all other problems a propriatary lossy format has such as cascade coding artefacts when transcoding etc, problems when migrating the data in the future.

If they offer exactly whats on the CD people are going to compare with CD prices.

Wombat
2011-02-27, 15:52
Worst case scenario is that they are developing some mongrel hirez aac format ? 24/96 but done with perceptual coding so the filesize is at CD size or something.

That would surely sound good enough, but leave you with all other problems a propriatary lossy format has such as cascade coding artefacts when transcoding etc, problems when migrating the data in the future.

If they offer exactly whats on the CD people are going to compare with CD prices.

Recent lossy codecs are already floating point and have no bit depth in a way PCM wavs have. So if you feed them with 24bit files and no higher sampling rate i doubt the files will be different much cause the psycho-acoustic will leave the music content below the noise floor of 16bit away already most of the time.

Mnyb
2011-02-27, 16:17
Recent lossy codecs are already floating point and have no bit depth in a way PCM wavs have. So if you feed them with 24bit files and no higher sampling rate i doubt the files will be different much cause the psycho-acoustic will leave the music content below the noise floor of 16bit away already most of the time.

Very likely so but I wonder if one gets better result compressing to mp3 from 24/96 or 16/44.1 ? An exception must be another mongrel format DTS 24/96 (not to confuse with DTS-HD Master Audio wich is for bluray and lossles )

As many others have done you can wonder what psycho-acoustic model DTS uses above 20khz

Wombat
2011-02-27, 16:27
Very likely so but I wonder if one gets better result compressing to mp3 from 24/96 or 16/44.1 ?
I doubt that. The developers of mp3 and their test sample arsenal no and never had a 24bit sample that only was transparent when restoring more then 16bit. If so i am really curious and must have missed that in all the years i followed the lame encoder developement.



As many others have done you can wonder what psycho-acoustic model DTS uses above 20khz
I bet they had an audiophile beneeth the developement team that had mental contact to his cat...

mortslim
2011-02-27, 17:11
Several other news sources have also reported that iTunes will probably be offering 24-bit audio soon.

http://edition.cnn.com/2011/TECH/web/02/22/24.bit.music/

http://www.geekistry.com/2011/02/22/24-bit-music-coming-to-itunes/

http://www.geek.com/articles/apple/apple-wants-24-bit-audio-for-itunes-music-downloads-20110222/

http://arstechnica.com/apple/news/2011/02/itunes-may-upgrade-to-24-bit-files-but-why-bother.ars

http://www.businessinsider.com/apple-could-sell-higher-quality-music-files-but-why-would-it-2011-2

This will be like iTunes and Amazon Video on Demand offering movie rentals in standard definition at $3.99 and the same movies in high definition for rental for a dollar more.

Apparently Apple wants to cover the entire audio market too, irrespective of whether 24-bit audio will be worth the extra cost to the consumer.

toby10
2011-02-28, 05:36
Apple needs to find ways to get their customers to:
a. buy more music
b. re-buy their same music
c. update their hardware

.... what better way to accomplish this than to offer higher bit rate and lossless music.

Soulkeeper
2011-02-28, 06:51
By inventing a brand new number called iTune iNdex, which is sample rate in kHz multiplied by bit depth. That will enable their users to appreciate the enormous quality improvement from 16/44.1 (iNdex 705.6, or i700 for simplicity's sake), to 24/48 (iNdex 1152, or i1200 for simplicity's sake), to 24/96 (iNdex 2304, or i2300 for simplicity's sake), to 24/192 (iNdex 4608, or i4600 for simplicity's sake), all expressed in one single, beuatifully simple, delicious iNumber.

(Yes I know, the question was intended rhetorically, and I also know that if someone from Apple reads this, I should get ready to sue them for stealing my idea, but I probably can't afford going to court against them anyway.)

sebp
2011-02-28, 06:58
That will enable their users (...)
You mean iDiophiles? ;)

dsdreamer
2011-02-28, 21:13
I like the linked Gigaom article's tone and approach.

http://gigaom.com/apple/24-bit-itunes-music-would-be-a-step-in-the-right-direction/

I'd personally only interested if it were 24 bits@96kHz. I care more for the 96kHz than the 24 bits, but I'd certainly be glad to have both. I'd feel a bit sorry for the likes of Linn Records and HDTracks if this were to happen, though.

Mnyb
2011-02-28, 22:29
I like the linked Gigaom article's tone and approach.

http://gigaom.com/apple/24-bit-itunes-music-would-be-a-step-in-the-right-direction/

I'd personally only interested if it were 24 bits@96kHz. I care more for the 96kHz than the 24 bits, but I'd certainly be glad to have both. I'd feel a bit sorry for the likes of Linn Records and HDTracks if this were to happen, though.

? I was under the impression that majority of the eventual improvement is in the increased bith depth ? As this would increase the signal content that we stand a chance to hear.

I know what a brick wall fillter is, but this is nowdays only a factor during recording then the frequency must be high to avoid that,.
But for playback at home most modern DAC's have massive oversampling so for playback this is not really a problem anymore.

But 24/96 would be a good consumer standard if it came to that, as maybe 44.1 is a little to low but it don't need to be 96 and maybe more than 16bit is needed, but again not as much as 24bit . So this would be all you can eat with some error margin built in .

trivia, a part off the DVDA standard I never seen in practical use was that you could have different bith deth and sample rate for different channels so theoretically if you did an 5.1 dvda and wanted to save some disc space you could have 20/48 for the rear channels or what ever, 192k was only allowed for pure 2ch otherwise you could mix between all rates.

Even if like the 24/96 format it would be hard to motivate 24/96 for the LFE channel ?

If they want hifi why not reintroduce multichannel music again ? as dvda an d sacd had. This works very well it's not a gimmick (unless the producers of the record make it soo ), but as when stereo was introduced they where Luddites claiming that mono was better too.

Soulkeeper
2011-03-01, 03:31
Bit depth is not all about quiet parts, and sampling frequency is not all about high frequencies. The point is that the resolution increases, AFAIU.

Pneumonic
2011-03-01, 10:45
So called "hi rez" offerings beyond 16/44 are nothing but marketing ploys concocted by audio firms as they look to penetrate another entirely new (read profitable) market segment, extracting hard earned money from gullible consumers in the process.

Why shouldn't Apple be able to profit similarly?

Phil Leigh
2011-03-01, 10:52
Bit depth is not all about quiet parts, and sampling frequency is not all about high frequencies. The point is that the resolution increases, AFAIU.

For bit depth yes, but not for sampling frequency. You can capture higher frequencies, however the lower frequencies are captured with the same precision they ever were...

dsdreamer
2011-03-01, 22:23
? I was under the impression that majority of the eventual improvement is in the increased bith depth ? As this would increase the signal content that we stand a chance to hear.

I know what a brick wall fillter is, but this is nowdays only a factor during recording then the frequency must be high to avoid that,.
But for playback at home most modern DAC's have massive oversampling so for playback this is not really a problem anymore.



For any kind of oversampling to occur without gross errors, there needs to be a very sharp low pass filter (usually done in the digital domain). There is nothing quite like this high order, low pass filter in natural acoustic spaces, which is why low sample rate PCM recordings always sound like something's wrong with them, IMO. The counter argument is that our ears have the bandwidth limitation, so that is doesn't matter if the reproduction system has a compatible bandwidth limitation. The reason I think otherwise is that we humans have a very non-linear perception of audio amplitude, and we can detect wrong timing of transient events even though we may not have the corresponding ability to detect continuous high frequency tones. For a purely linear system, this would be nonsense, since bandwidth and timing resolution are inextricably linked as reciprocal quantities in linear signal processing. Nevertheless, it is my conjecture that a higher than 20kHz bandwidth audio channel is necessary to avoid that our sensitive, non-linear ears pick up disconcerting, not-found-in-nature timing cues that cause us to identify the music source as electronic, unnatural and perhaps even fatiguing for long-term listening.

I guess I am setting myself up for several flames for writing this post. However, I am just sharing an opinion.

Wombat
2011-03-01, 22:32
I guess I am setting myself up for several flames for writing this post. However, I am just sharing an opinion.

No, of cause not. We are on "Slimdevices General" forum. Let your feelings flow, no evidense needed! :)

Mnyb
2011-03-01, 22:47
May be soo ? But our old analog playback chain is/ was also bandwith limited and suffer similar problems ? Can these problems be interly described as a " digital evil" .

The digital filtering in downsampling they are not nearly as bad as the analogue counterpart if you record at 44.1 ? Hence my pow, as sugested by people that have tried downsampling the material themself.
Many claim that they hear no difference.

I think Phil uploaded some samples in another tread, I shall dl and try them in the afternoon if I have time and if they are still there.

My software collection is a bit shoody audacity is buggy I don't trust it completely.

Wombat
2011-03-01, 22:55
No matter what these samples bring, there are some that claim it is better even when no one can prove to hear differences or the other way around. No way to prove anything in here in any way.

If you do they wonder for a fraction of a second about your deafness and go on writing in the next thread or different forum...

Mnyb
2011-03-01, 23:07
I stole this link from you Wombat..

If it is ok

http://hlloyge.hl.funpic.de/wp-content/uploads/2007/09/audibility-of-a-cd-standard-ada-loop-inserted.pdf

It is a good experiment.

Read the whole article, it is apparently soo that very good master are made avaible to us via hirez channels so given that this is the way we are going to get them...

dsdreamer
2011-03-01, 23:17
No, of cause not. We are on "Slimdevices General" forum. Let your feelings flow, no evidense needed! :)

It is not a feeling, but a reasoned conjecture. Difficult to measure per se without sticking electrodes in people's brains. The non-linear and time variant nature of the human ear's response to sound pressure waves is, however, well-established.

The following is not a scientific paper, either, but you may find the following discussion interesting.

http://www.ayre.com/pdf/Ayre_MP_White_Paper.pdf

All I am saying is that from my point of view this seems plausible.

Mnyb
2011-03-01, 23:25
Well given the choice between 256k AAC or 24/96 PCM what to do .
And better soo if it is sourced from a different less compressed master .

So let them market this, if it ends with us getting better sound evfen if it would be possible to cram into a CD .

Specifically markett you effort at audiphiles gives some dgrees of freedoom it does not have to sound god on AM radio in a car..

So if this for business reasons (nobody believes in good old cd) , must be done on a new format so be it .

If it's not to outrageously expensive I'll buy

Cape11
2011-03-02, 04:27
I agree with TerryS on this (noting PL's point about dither). I don't think the evidence for the roughness of 16 bit CD sound is a subtle audio point. Listen to any recording of a string quartet when they play a quiet passage and the roughness of the sound is perfectly obvious. On some recordings, where they have recorded at a rather lower level than usual, the sound can be quite dreadful. A 'good' example of this is the Caliope recordings of the Beethoven Quartets played by the Talisch. The LPs sound fine - good, warm string tone; the CDs are so bad that you might have difficulty identifying the source as a string quartet. So I'm all for 24 bit delivered to the end user.

I'm also for higher sampling rates. One of the other failings of CD, compared to LP, is the poor quality of transients. Without over-sampling, CD delivers a frequency response flat to ~22 kHz and then a cliff edge. Imagine the filter you'd design to achieve that, and then consider the phase/frequency distortion that would occur. I doubt over-sampling removes these phase artefacts.

Phil Leigh
2011-03-02, 05:47
I agree with TerryS on this (noting PL's point about dither). I don't think the evidence for the roughness of 16 bit CD sound is a subtle audio point. Listen to any recording of a string quartet when they play a quiet passage and the roughness of the sound is perfectly obvious. On some recordings, where they have recorded at a rather lower level than usual, the sound can be quite dreadful. A 'good' example of this is the Caliope recordings of the Beethoven Quartets played by the Talisch. The LPs sound fine - good, warm string tone; the CDs are so bad that you might have difficulty identifying the source as a string quartet. So I'm all for 24 bit delivered to the end user.

I'm also for higher sampling rates. One of the other failings of CD, compared to LP, is the poor quality of transients. Without over-sampling, CD delivers a frequency response flat to ~22 kHz and then a cliff edge. Imagine the filter you'd design to achieve that, and then consider the phase/frequency distortion that would occur. I doubt over-sampling removes these phase artefacts.

That whole filter thing is indeed history, thanks to oversampling which moves the filter way out of harms way...unless you think 384kHz and higher is still an issue?


The other issues you describe are addressed by recording and mastering things properly (including the use of high sample rate and bit depths). Get this right and the final playback in 16/44.1 is NOT a problem, as eloquently explained - and proven - in the Meyer/Moran paper. The problem is most people seem to judge 16/44.1 PLAYBACK on the basis of ropey masters. This isn't helped by the fact that the best masters are often only available in DVD-A/SACD/Hi-res downloads... this makes a casual apples-to-apples comparison difficult

adamdea
2011-03-02, 06:06
I agree with TerryS on this (noting PL's point about dither). I don't think the evidence for the roughness of 16 bit CD sound is a subtle audio point. Listen to any recording of a string quartet when they play a quiet passage and the roughness of the sound is perfectly obvious. On some recordings, where they have recorded at a rather lower level than usual, the sound can be quite dreadful. A 'good' example of this is the Caliope recordings of the Beethoven Quartets played by the Talisch. The LPs sound fine - good, warm string tone; the CDs are so bad that you might have difficulty identifying the source as a string quartet. So I'm all for 24 bit delivered to the end user.


I have the Calliope "Integrale des Quatuors" by the Talich Qt but have not listened to it recently. I have not in the past noticed any issue with the sound quality (other than noting that it is not a great recording to start off with). Can you point me to where you find the sound so rough (3rd movement of the 135?).
I note that the Gramophone reviewer commented on the first CD issue of this set (March 1987)http://www.gramophone.net/Issue/Page/March%201987/69/815793/BEETHOVEN.+STRING+QUARTETS.+Talich++Quartet+(Petr+ Messiereur,+Jan+Kvapil,+vns+Jan#header-logo

"The recordings which, like those of the Vêgh, seemed a bit bottom heavy, are much less aggressive than the DG for Melos and want the bloom and richness of the EMI set. All the same, the sound is more firmly defined and better-focused than on LP and eminently natural. The sound does not call attention to itself, the instruments are well placed and the timbre truthful: in fact, one quickly forgets about it and loses one's self in the music."

adamdea
2011-03-02, 06:30
see also http://www.classicstoday.com/review.asp?ReviewNum=2550


"The Talich traversal of the Beethovens originally came out in the dying days of analogue. I loved them then and I love them even more in their CD incarnation, which adds transparency and bite to sound that always was clear and impactful."

BTW I am all for 24 bit recording and playback I am just puzzled by the prevalence of the view in audiophile circles that properly done 16 bit is noticeably inferior to vinyl, when the opposite view appears to have been held by most music reviewers (see the digital vs analog thread)

Cape11
2011-03-02, 08:01
Well, I simply speak as I find.

Equipment:
Loudspeakers: ATC SCM100 asl
Preamp: Chord CPA2800
DAC Chord DSC1500E
Deck SME20.2 with series 5 arm
Cartridge: Lyra.
Squeezebox.

(Expensive, yes, but then there's little reason to change the system, and it's already done 15 years of good service).

Sample any part of the Talisch CDs esp. when quiet and it sounds rough - opening of 131 will do. Go to the LPs and the sound is much better - they really are string instruments!

Cape11
2011-03-02, 08:13
That whole filter thing is indeed history, thanks to oversampling which moves the filter way out of harms way...unless you think 384kHz and higher is still an issue?Well is it? Over sampling means there's no need for a filter, but my point was that the frequency response is shaped like a cliff edge, and over-ssampling can do nothing to remedy that under-lying condition. I remain concerned about possible phase distortion into the audio band which would have a deleterious effect on the leading edge of percusiion instruments. And that's what I miss on CD (the clink on the triangle when first hit, attack of piano notes) but get on LP.

I would think 192 ksps would suffice, which is 'sort of' the sampling equivalent of SACD. If all these effects have been nailed, why is there almost universal agreement that SACD is better than CD?

Phil Leigh
2011-03-02, 08:38
Well is it? Over sampling means there's no need for a filter, but my point was that the frequency response is shaped like a cliff edge, and over-ssampling can do nothing to remedy that under-lying condition. I remain concerned about possible phase distortion into the audio band which would have a deleterious effect on the leading edge of percusiion instruments. And that's what I miss on CD (the clink on the triangle when first hit, attack of piano notes) but get on LP.

I would think 192 ksps would suffice, which is 'sort of' the sampling equivalent of SACD. If all these effects have been nailed, why is there almost universal agreement that SACD is better than CD?

There is almost universal misunderstanding. I'll leave to one side the fact that large parts of the upper spectrum of SACD's are full of noise... thankfully we can't hear it :-)

As I keep saying, SACD playback is not inherently superior to redbook... in terms of what we can hear. That's the whole point of the Meyer/Moran paper. They inserted a redbook adc-dac chain in the midst of an SACD playback chain. Nobody could hear it! (this isn't the basis for my argument, which I've been developing for years - it just neatly backs it up).

What CAN BE better is the mastering job done to produce the SACD (sometimes - there are some pretty ropey SACD's out there too...).

Don't be fooled into comparing the redbook and sacd layers of a hybrid disk - they often don't use the same master, rendering the comparison pointless. Of course, far be it for me to suggest that record companies would deliberately make the expensive SACD layer sound better :-)


To reiterate, it's NOT the playback technology, it's the mastering. Redbook can sound identical to SACD.


By the way, that frequency response cliff exists on many SACD's too...

Cape11
2011-03-02, 08:53
Well we're both arguing by repetition, so we'll have to agree to differ.

Wombat
2011-03-02, 10:22
It is not a feeling, but a reasoned conjecture. Difficult to measure per se without sticking electrodes in people's brains. The non-linear and time variant nature of the human ear's response to sound pressure waves is, however, well-established.

The following is not a scientific paper, either, but you may find the following discussion interesting.
http://www.ayre.com/pdf/Ayre_MP_White_Paper.pdf


It is not well established people can hear any of the problems that Ayre marketing paper talks about.
These funny impulses that need the magic Apodizing show pre or post-echo in music material and ONLY happen above 20kHz. So these fancy pics tell it is pre-echo but leave out the fact it is happening at low volume and high frequency.
Now when you create a 44.1 version of your hires material with resampling that doesn´t produce pre-ringing and you magic filter it afterwards you even harm the signal. These filters are non-linear also.

Benchmark Media is another audio manufacturer that clearly states this pre-rining is a non-issue to them.

Of caus these fancy graphs look shocking!

bobkoure
2011-03-02, 10:46
That's the whole point of the Meyer/Moran paper. They inserted a redbook adc-dac chain in the midst of an SACD playback chain. Nobody could hear it!
Hi Phil - got a link to that paper? I googled for it, and it's apparently generated loads of controversy (my google search (http://www.google.com/search?num=50&hl=en&lr=&q=Meyer+Moran+CD+audio&btnG=Search&aq=f&aqi=&aql=&oq=)), which IMO is all to the good, but I'd be curious to read the paper itself.

BTW, as I remember, the CD audio format was chosen, in part, to be able to fit an entire LP onto a single data tape. No, I can't find a reference, but I vaguely remember a good discussion of the process to determine/decide on this new digital format in one of the IEEE mags in the early '80s. Do you remember this being one of the factors? Or am I totally misremembering?

Phil Leigh
2011-03-02, 11:11
Hi Phil - got a link to that paper? I googled for it, and it's apparently generated loads of controversy (my google search (http://www.google.com/search?num=50&hl=en&lr=&q=Meyer+Moran+CD+audio&btnG=Search&aq=f&aqi=&aql=&oq=)), which IMO is all to the good, but I'd be curious to read the paper itself.

BTW, as I remember, the CD audio format was chosen, in part, to be able to fit an entire LP onto a single data tape. No, I can't find a reference, but I vaguely remember a good discussion of the process to determine/decide on this new digital format in one of the IEEE mags in the early '80s. Do you remember this being one of the factors? Or am I totally misremembering?

Courtesy of Wombat: http://hlloyge.hl.funpic.de/wp-conte...p-inserted.pdf

wasn't the cd supposed to be long enough to hold a particular Beethoven symphony? (this might be a myth)

Wombat
2011-03-02, 11:38
Courtesy of Wombat: http://hlloyge.hl.funpic.de/wp-conte...p-inserted.pdf


I have that link to that paper from Hydrogenaudio.org
On there are similar discussions running, there no one can do esotheric claims without backing it up btw. It is violating the rules then :)

adamdea
2011-03-02, 13:17
Courtesy of Wombat: http://hlloyge.hl.funpic.de/wp-conte...p-inserted.pdf

wasn't the cd supposed to be long enough to hold a particular Beethoven symphony? (this might be a myth)
Yes the ninth. I have no idea whether it's true, but I have heard various versions. The most entertaining (and improbable) has it that the stipulation was made by "Mr Sony's" wife.

maggior
2011-03-02, 13:25
I recall reading an article in recent years celebrating an anniversary of the CD format. In it they discussed the facts and myths surrounding the format's development, including why 74 minutes and so forth. If I remember correctly, they even spoke with members of the team or the key developer.

I'll have to see if I can jog the appropriate memory cells and see if I can locate the article.

sebp
2011-03-02, 14:11
If I remember correctly, they even spoke with members of the team or the key developer.

I'll have to see if I can jog the appropriate memory cells and see if I can locate the article.
It would be really nice if it's confirmed by somebody at Sony or Philips.

In the meantime, here's another interesting version:
http://www.wired.com/thisdayintech/tag/norio-ohga/

I love this:

That was theoretically long enough for Furtwängler’s Ninth, but in reality it wasn’t. The real limit for CDs started at 72 minutes, the maximum length of the U-Matic videotapes then used for audio masters. So the Furtwängler performance couldn’t be released on a single CD until new digital audio technology made that possible in 1997.

dsdreamer
2011-03-02, 21:45
Most of the tests were done using a pair of highly regarded, smooth-measuring full-range loudspeakers in a rural listening room with an ambient noise floor of about 19 dBA SPL, all electronics on (see Fig. 2).

This vagueness about the equipment being used is a bit of a concern, but I managed to find out from another source that:


The playback equipment in this system consisted of an Adcom GTP-450 preamp and a Carver M1.5t power amplifier. Speaker cables were 8 feet of generic 12-gauge stranded wire; the line-level connecting cables were garden-variety. Three different players were used: a Pioneer DV-563A universal player, a Sony XA777ES SACD model, and a Yamaha DVD-S1500. The loudspeakers were a pair of Snell C5s.

It seems as if the equipment used to evaluate the audibility of the 16/44.1 “bottleneck” was variable and never itself truly verified for audio bandwidth and dynamic range. If this were my measurement campaign I would insist on a frequency sweep of the entire equipment chain from source to sound waves captured with a calibrated microphone at the listening position. As it is, I am not convinced that the introduced 16/44.1 bottleneck was truly the dominant system bottle neck. I don't have any worries about the Carver amp, but the Adcom pre has a 0.5dB bandwidth of 20Hz to 20kHz, and the speakers top out at 22kHz.

It's 2011 and its still "well past time to settle the matter scientifically."

Mnyb
2011-03-02, 23:20
If you read the paper they are open for that you can not prove nonexistance of something scientifically.
So there can be outliers where detections is possible.
Also they used more than one setup but these are not specified ?
Also some studio monitors where involved in some unspecified cases ?

This paper is most interesting if you see what they tried to do, they used a range of typical audiophile stereos and a bunch of mostly typical audiphiles.

So they build a very strong case for that the normal situation is that sligthly worse than CD quality is tranparent to most audiphiles in most cases especially at normal listening levels.
The oposite is a rare exception but not ruled out as completely impossible.

Their other conclusion is just as interesting, that most of the dvd-a and sacd they used had very good sq.
Good masters .

If you specifically wants to find the rare exceptions one can set it up like a challange.
Where interested parties can bring in equipment they think is capable.
A good modern fully digital recording that stand a chance for detection I don't think remasters of 30 yo stuff cuts it ( not DSD either )..
Listener(s) they believe are qualified.

Failing on your own terms is much more convincing .

And you most likely will find exceptions, people that with some special recordings on very good systems that can hear this.
But it can not be expanded to a general case where most audiophiles can safely assume that they are this
Special case.

But if it makes you feel better you can secretly assume that you are that exception, I wont tell ;)

But hey this is not what hirez audio is hyped up to be.
The labels make the hype becomme true by providing better masters.

An another note what is the "stuart paper" they challange is it an AES paper by Bob Stuart of Meridian perhaps ? I once had a white paper from Meridian where Bob stated that 20 bit 55 kHz sample rate was
the ultimate limit .
And they did sell a lot of dvda players ( can you see my sig ).

firedog
2011-03-03, 04:27
quote from the original article:
"but the loudness war has resulted in music squashed to within a few decibels of its life."

Yes, and hi-res releases seem to generally be mastered without that. That alone makes them much more listenable. The recent uncompressed hi-res remaster of Band of the Run is a good example. I can definitely hear more detail and a warmer tone on the hi-res. Because of it not being overly compressed there's also a lot of dynamics. It invites you to turn it up and listen loudly, b/c it sounds better that way. Not true of the standard CD. This may all be do to better mastering and a lower noise floor, and not some inherent "superiority" of the format, but I don't care. I can hear the music better and it's more enjoyable to listen to. So I don't see how I'm being "fooled" by the hi-res release.

I've got some jazz and classical releases that sound wonderful on CD. Can't really say the same for much redbook pop/rock. So the hi-res versions often sound better to me.

I have no argument if you want to tell me that standard CD is all I need and can hear IF DONE PROPERLY. But apparently it isn't done properly most of the time.

And I have no illusions that the music conglomerates are going to start putting out extremely high quality standard CDs. So the fact that CD's can sound as good as hi-res is irrelevant in practical terms, b/c those CD's don't exist, for the most part.

Even the Boston Audio Society paper referred to acknowledges that
"one trend became obvious very quickly and held up throughout our testing: virtually all of the SACD and DVD-A recordings sounded better than most CDs—
sometimes much better."

So I'll keep buying hi-res. But before you call me silly and a fool, just change the name from "hi-res" to "properly mastered" and then we won't have anything to argue about.

Mnyb
2011-03-03, 05:16
I also buy hirez in some quantities. >1500 tracks so far.

But I try not have to many illusion about it .

But I won't fall in to HD tracks newest trapp, charching more for 176 or 192 khz than the pedestrian 88/96 that I would be happy with.

Unless... these to are off different masters :-/
HD Tracks does not have stellar records in keeping track of the pedigree of all their releases.
It would be better for all if it where properly documented which master and how the transfer was done.
In the urge to differentiate the products it is way to tempting to *really* make sure that they are different.

Guess I have to take the bait and try some of that 176kHz... d*mn

Phil Leigh
2011-03-03, 05:17
...So I'll keep buying hi-res. But before you call me silly and a fool, just change the name from "hi-res" to "properly mastered" and then we won't have anything to argue about.

That's what I do too!

bobkoure
2011-03-03, 06:50
Thanks for the link!

wasn't the cd supposed to be long enough to hold a particular Beethoven symphony? (this might be a myth)
I remember something like this too (not definitely Beethoven, but some particular long work that would just fit on both sides of an LP). Of course, I may have just heard the myth long ago.

firedog
2011-03-03, 11:41
I brought this topic up over at computeraudiophile.com

Got a thoughtful answer from music producer Barry Diament, who records in 24/192 and produces hi-res discs. He has an explanation for why 24bit does matter.

http://www.computeraudiophile.com/content/debate-24-bit-it-just-audiophoolery#comment-72930

In forums he claims he can consistently pick out hi-res recordings on good systems, and he seems to have evidence to back up his claims. So maybe he is like the tested people in the study who scored 8/10 and 7/10. A small group with trained listening skills that can hear the difference.

Wombat
2011-03-03, 12:06
cumputeraudiophile!? Serious? This is the last place i will ever look for any evidence.

Phil Leigh
2011-03-03, 13:25
It's easy enough to do your own tests... you already have SOX which is capable of high quality downsampling/word reduction.

mortslim
2011-03-07, 23:37
Here's a new article:

"Whatever Happened To The Audiophile?"

http://www.npr.org/2011/03/05/134256592/whatever-happened-to-the-audiophile

sebp
2011-03-08, 01:59
It's easy enough to do your own tests... you already have SOX which is capable of high quality downsampling/word reduction.
Hi Phil,

I had another challenging session last sunday, and I surprisingly had the impression I could hear some differences this time.
I couldn't really tell I heard anything missing from the CD version, but the HD file just sounded better to my ears.

As I don't trust them, I used Audacity's spectrum analyser with both files, and looking at average gain values for some frequencies, they always appeared to be a bit louder in the HD Version than in the CD version.
If you look at the screenshots, you'll see it's 4 dB louder for 100 Hz, for example (-32dB is for the HD file, -36dB is for the CD file).

I then computed replaygains using foobar2000, and got -6.40dB for the CD version and -6.02dB for the HD version.

So, downsampling 24/96 to 16/44.1 with SoX seems to lower the average gain of the 16/44.1 version a little bit.

Is an average 0.38dB volume difference sufficient to be heard?
I can't tell for sure, but thought it would be worth mentioning...

Phil Leigh
2011-03-08, 02:40
Hi Phil,

I had another challenging session last sunday, and I surprisingly had the impression I could hear some differences this time.
I couldn't really tell I heard anything missing from the CD version, but the HD file just sounded better to my ears.

As I don't trust them, I used Audacity's spectrum analyser with both files, and looking at average gain values for some frequencies, they always appeared to be a bit louder in the HD Version than in the CD version.
If you look at the screenshots, you'll see it's 4 dB louder for 100 Hz, for example (-32dB is for the HD file, -36dB is for the CD file).

I then computed replaygains using foobar2000, and got -6.40dB for the CD version and -6.02dB for the HD version.

So, downsampling 24/96 to 16/44.1 with SoX seems to lower the average gain of the 16/44.1 version a little bit.

Is an average 0.38dB volume difference sufficient to be heard?
I can't tell for sure, but thought it would be worth mentioning...

Hi.
Whether or not you can hear a 0.38dB difference is highly debatable - in my experience such things are barely audible... when mixing down, a 0.5dB lift/cut is barely noticeable and more often than not a 1-1.5dB change is requird o register as "different". Obviously this varies slightly from person to person.

In this comparison, how were the 2 versions created? - using SOX to downsample? - taking HD and CD versions from a commercial release doesn't work as there is no way of knowing how the masters were created.




By the way, the Audacity frequency spectrum analysis is highly suspect - it isn't accurate...

sebp
2011-03-08, 03:19
Obviously this varies slightly from person to person.
As said, I don't trust my ears that much, and am pretty sure I couldn't ABX the files.

The strange part is that I was challenging the files alone, and ignored two friends of mine were doing the same.
One of them, who is more accustomed to comparisons than me, also suspected his downsampled version was sounding lower.
I only discovered that later.


In this comparison, how were the 2 versions created? - using SOX to downsample? - taking HD and CD versions from a commercial release doesn't work as there is no way of knowing how the masters were created.
I first compared commercial HD/CD downloads, and then rolled my own CD version from the HD file.
In both cases this 0.38dB volume difference existed.

I'll try using r8brain to check whether the volume difference also exists with that tool.


By the way, the Audacity frequency spectrum analysis is highly suspect - it isn't accurate...
As I'm not an audio specialist, I wasn't sure my readings were right.
That's why I've computed replaygains to check.

Phil Leigh
2011-03-08, 03:29
As said, I don't trust my ears that much, and am pretty sure I couldn't ABX the files.

The strange part is that I was challenging the files alone, and ignored two friends of mine were doing the same.
One of them, who is more accustomed to comparisons than me, also suspected his downsampled version was sounding lower.
I only discovered that later.


I first compared commercial HD/CD downloads, and then rolled my own CD version from the HD file.
In both cases this 0.38dB volume difference existed.

I'll try using r8brain to check whether the volume difference also exists with that tool.


As I'm not an audio specialist, I wasn't sure my readings were right.
That's why I've computed replaygains to check.

How did you roll your own?

sebp
2011-03-08, 04:02
How did you roll your own?
I've used the example given in the FAQ:

sox any-file -b 16 outfile rate 44100 dither -s

Please note that I have not challenged my version with the downloaded master.
I only wanted to check whether there would be any volume difference, like with the downloaded CD version, or not.
As it was identically lowered, I must admit I have not bothered challenging it (yet).

My friends have not used the downloaded CD version to make their comparison, but used their own downsampled version (they used dbPoweramp).

Wombat
2011-03-08, 11:48
I've used the example given in the FAQ:

sox any-file -b 16 outfile rate 44100 dither -s

Please note that I have not challenged my version with the downloaded master.
I only wanted to check whether there would be any volume difference, like with the downloaded CD version, or not

This sox setting should be ok.
For the loudness thing i wonder. If you don´t tell sox any different it won´t change the loudness Replaygain calculates cause Replaygain only uses audible audio bands for calculation.
Since some resampling settings create short peaks while resampling sox can adjust the volume to prevent clipping then with using --norm additionaly.
The resulting file will be some 0.1 of dB more silent.
I remember a Linn sample that was discussed on Hydrogenaudio. The 44.1 kHz version was 0.5dB more silent as the 96khz version. Nobody knows it was done cause of giving a small advantage to the 96kHz version due to some positive reaction for louder sounds or Linn simply adjusted the level to prevent clipping.
Most likely the first one cause even many so called studio masters have clipping samples.

sebp
2011-03-08, 13:19
Most likely the first one cause even many so called studio masters have clipping samples.
Well, I also thought at first the CD version could have been made sounding quieter intentionally, but given the album I've tested comes from a very small label, and the fact that my downsampled version exactly matches the average gain of the downloaded commercial version, it looks like conspiracy theory to me.

More likely, they've also used SoX to downsample their 24/96 master to 16/44.1.

firedog
2011-03-21, 07:32
From a thread on another forum designed to show the differences between standard CD quality and hi-res recordings (same recording at different sampling rates and bit depth).

To my ears, the differences are clearly there, just as described. If you can't discern them, then either your system isn't up to snuff or your ears need training.

IMO, the vast majority of listeners, if played these tracks, would say they sound "the same". They all sound good -they're from the same master. It's just that when you know what to listen for, you can hear the difference.


from Bruce Brown's (Puget Sound) thread at http://www.whatsbestforum.com/showthread.php?2938-The-Art-of-Listening-Hi-rez-music

One thread (the one with the link here) lets you compare redbook to higher res versions of the same music. You can clearly hear the differences, as Bruce says below in the thread:

"Here is a file that we recorded at a native 24/352.8 It starts off with piano/drum. Do not listen to the music! Instead, concentrate on the sticks hitting the cymbal. Listen to the attack/transient of the initial hit. Listen to the tone and then the decay. Focus on just this one element.
Listen to the sound of the cymbals in the room. How big is the room? What kind of space is this drummer in? Listen to the tone of the cymbal.
Now... as you go down to 176.4, then 88.2 and finally 16/44.1, listen to these elements that I talked about above. Listen how the transient attack becomes more dull. Notice how the tone of the cymbal changes from crisp/pristine to dull and flat. Next, notice how the decay becomes shorter and shorter and the "room" becomes smaller and more dry with less reverb."

To my ears, clear proof that the improvement with hi -res is audible. You can hear the differences in transients and decay, and you can hear that the room "gets smaller" at lower res. To me, in spite of what others wrote on this thread, this makes a difference in "real life". The music sounds more realistic and differently situated in space. Simply "better" - closer to the real thing, as you'd hear it if you were in the room.

And please don't start telling me I imagined it all.

Phil Leigh
2011-03-21, 09:21
From a thread on another forum designed to show the differences between standard CD quality and hi-res recordings (same recording at different sampling rates and bit depth).

To my ears, the differences are clearly there, just as described. If you can't discern them, then either your system isn't up to snuff or your ears need training.

IMO, the vast majority of listeners, if played these tracks, would say they sound "the same". They all sound good -they're from the same master. It's just that when you know what to listen for, you can hear the difference.


from Bruce Brown's (Puget Sound) thread at http://www.whatsbestforum.com/showthread.php?2938-The-Art-of-Listening-Hi-rez-music

One thread (the one with the link here) lets you compare redbook to higher res versions of the same music. You can clearly hear the differences, as Bruce says below in the thread:

"Here is a file that we recorded at a native 24/352.8 It starts off with piano/drum. Do not listen to the music! Instead, concentrate on the sticks hitting the cymbal. Listen to the attack/transient of the initial hit. Listen to the tone and then the decay. Focus on just this one element.
Listen to the sound of the cymbals in the room. How big is the room? What kind of space is this drummer in? Listen to the tone of the cymbal.
Now... as you go down to 176.4, then 88.2 and finally 16/44.1, listen to these elements that I talked about above. Listen how the transient attack becomes more dull. Notice how the tone of the cymbal changes from crisp/pristine to dull and flat. Next, notice how the decay becomes shorter and shorter and the "room" becomes smaller and more dry with less reverb."

To my ears, clear proof that the improvement with hi -res is audible. You can hear the differences in transients and decay, and you can hear that the room "gets smaller" at lower res. To me, in spite of what others wrote on this thread, this makes a difference in "real life". The music sounds more realistic and differently situated in space. Simply "better" - closer to the real thing, as you'd hear it if you were in the room.

And please don't start telling me I imagined it all.

I don't have a 352.8 DAC... and I have no way of knowing how the downsampling from 352-176-88-44 was done...

I'll try a few things...

Wombat
2011-03-21, 11:20
from Bruce Brown's (Puget Sound) thread at http://www.whatsbestforum.com/showthread.php?2938-The-Art-of-Listening-Hi-rez-music

One thread (the one with the link here) lets you compare redbook to higher res versions of the same music. You can clearly hear the differences, as Bruce says below in the thread:

Can you give us a short description how you tested the different recordings against and which of them? How you switched? From a very short look it doesn´t seem the recording uses more then 13bits.
You may join this thread at Hydrogenaudio.org or ask the poster of these samples to go there. This thread could be interesting from there on especially:
http://www.hydrogenaudio.org/forums/index.php?showtopic=86649&st=300

Edit: you may use that 16bit version in your next test setup that someone uploaded here: Test_soxed/File size: 3.49 MB (http://www.fileserve.com/file/djrZaEH)

In the thread you linked to the suggestion you hear tones above your hearing the same way you feel a dinosaur foot stomp of 12Hz is just nonsense..

Phil Leigh
2011-03-21, 12:04
I know you can't trust Audacity spectrum plots, but...

Mnyb
2011-03-21, 12:44
I know you can't trust Audacity spectrum plots, but...

Aaargh :-/ is that how DSD noise looks like ? please filter before putting trough your amps level is not obscenenely high but there is a risk that the amp sounds "different" and gets hot too

sebp
2011-03-21, 16:51
I know you can't trust Audacity spectrum plots, but...
Could you please explain what you see there, for us mere mortals? :)

sebp
2011-03-21, 17:02
Could you please explain what you see there, for us mere mortals? :)
Oh yes, I can see it now, I had lost one zero...

firedog
2011-03-22, 05:45
If this helps anyone

Wombat
2011-03-22, 18:06
Could you please explain what you see there, for us mere mortals? :)

You can see it has indeed tons of HF noise from somewhere and i think Mnyb describes it correctly as dsd noise. Also you can see a very rapid lowerage in HF energy on the music especially on the Audacity pic. A psycho-acoustic codec will lowpass it pretty agressive cause most of the HF in the music is masked already.
Besides that the recording sounds superb.

Mnyb
2011-03-22, 23:05
You can see it has indeed tons of HF noise from somewhere and i think Mnyb describes it correctly as dsd noise. Also you can see a very rapid lowerage in HF energy on the music especially on the Audacity pic. A psycho-acoustic codec will lowpass it pretty agressive cause most of the HF in the music is masked already.
Besides that the recording sounds superb.

It may be a superb recordning :) is the music good ? but the music content seems to end < 40kHz
Aplying a step filter here and use 88.2kHz as sample rate seem to be the rigth thing to do.
This would preserve all content thats originated from the musical performance, the DSD quatisation noise is not music.
In fact the 88.2 version must have this done to it if not your in a world of aliasing artifacts ?

That parabolic shape is the typical DSD HF noise tail it is meant to be filtered away by the SACD player for example, preserving it in an 192kHz file is an technical error so that file is broken and can not be used to judge the sq of this recording.