Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Arindelle]

24/192 Music Downloads ...and why they make no sense http://people.xiph.org/~xiphmont/demo/neil-young.html

An engineer friend of mine were talking and arguing about "audiophile" myths and your basic 3rd-whiskey-discussion about "scientific" measurements and tests vs listening "perception" -- another one of those debunk the myths conversation.

This started out because my DAC is limited to 24/96k, and am thinking of changing it. In a very polite and round about way, he was inferring that I didn't have a clue as to what I was talking about, and he said if I wanted to argue points about bit-depth and oversampling, I had to understand basic sampling better, and the Nyquist  theorem, and eventually aliasing.

Long story short he told me to read this article published on the website of the Xiph.Org Foundation -- a non for profit organization which if I understood correctly is now the developing body for FLAC authored by Jack Coalson.

Now I am an expert skeptic .. so many (too many) "experts" out there on the internet But the explanations in this article are very convincing -- to the point that somethings I took for granted (like oversampling, bit depth, and hi-rez audio) could fall right into the "myth/placebo/perception" category.

Anybody find the authors arguments wrong or misleading?

Otherwise I'm think of changing my DAC to one that will not upsample at all and I will certainly not be jumping on the DSD bandwagon any time soon.

[6233638]

16/44 sources being sufficient, and oversampling DACs are different things.
NOS DACs perform much worse than oversampling DACs - there's a reason that oversampling has become standard.
If you look at these measurements from a Teac DAC which can switch between oversampling and non-oversampling, you see that NOS means you do have a very aliased "stairstep" representation of the signal. (a bad thing)

With a DAC that can handle 24/96, I would not be thinking of replacing it just to have access to higher sample rates or native DSD playback. Those things alone shouldn't have much if any impact on sound quality.
A new DAC may still sound better than your old one for other reasons, but if you're happy with how it sounds when playing CD quality audio, it might not be worth replacing it.

[gappie]

great article Arindelle.. I just think some people are to much focussed on that stuff.. they are not listening to the music, but just imagining how great their speakers and cables and stuff are. decent is good enough, then just enjoy the music.

 
gab

[mwillems]

That article is a favorite of mine.  Here's another long thread on this subject discussing the pros and cons of that very article (starting about halfway down): http://yabb.jriver.com/interact/index.php?topic=78219.0

[InflatableMouse]

With a DAC that can handle 24/96, I would not be thinking of replacing it just to have access to higher sample rates or native DSD playback. Those things alone shouldn't have much if any impact on sound quality.

Have a look at this article as well.

I had a couple of chats with Archimago before he took those measurements and although subjectively, we both "felt" DSD upsampling seemed less fatiquing and overal more pleasurable to listen to.

[InflatableMouse]

Anybody find the authors arguments wrong or misleading?

Otherwise I'm think of changing my DAC to one that will not upsample at all and I will certainly not be jumping on the DSD bandwagon any time soon.

I think that article is straight and true but we have to understand the purpose or the scope of that article. It mainly argues that storing high frequency information beyond what we can hear is pointless and could even have a negative effect (intermodulation). As long as you keep that in mind its not misleading but if you are led to believe that that's all there is to it, than I believe you do risk being misled.

See my reply to 62 and read that article I linked. Upsampling to DSD introduces some distortion but apparently it's experienced as pleasurable.

And while I agree with 62 about NOS DAC's, there seem to be legions of people swearing by them.

Different devices behave differently. Some DAC's have a sweet spot for 24/96, others might sound better with DSD or 16/48. Upsampling a CD to 24/96 doesn't magically turn that CD into high definition, but it could be that with certain DAC's it simply sounds better upsampled than playing it natively at 16/44, simply because that device behaves differently. One could argue that the DAC's design is flawed or whatever, I don't know. I simply state that this situation exists (and is measurable!) and is not something we should ignore.

[Wungun]

I just read that article...and I think the author is a little biased himself to prove his points...

I dunno about you, but how much of your media is below 20Hz or above 20Khz? And if it is, is it for sustained periods? No, and no...
Harmonic distortion as shown on a scope is obvious whether from a pure analogue or digital standpoint, and for the most part, is inaudible.

My media is played digitally thru my HTPC, and so I have a myriad of sampling rates available to choose from...AND on the fly thru my soundcard. And it's EASILY apparent the difference from switching between 16/44 and 24/192. The higher sample/bit rate is the clear winner.

Then the author goes on the claim that a SACD uses a different master than the same recording for a regular CD...? Really?!
So what, the artist and producers re-perform and record the same music in the studio to make a second master?? Of course they don't. lol
They call it a "master" for a reason.

And his point on the noise floor on higher sample rates...? The way I interpret it, the lower sample rate artificially "removes" or lowers the noise floor and you end up with a better signal/noise ratio. But if you're up sampling a 16/44Khz with a better signal/noise ratio than what was actually mastered at a higher sample rate in the studio, is this non-existent "noise" being over sampled as well, so that it is audible as distortion? I don't think so...
If the data isn't there to be converted from DA, then how could it be audible?

[Arindelle]

Thanks for the info/links guys and the link to the previous post ... have a bunch of questions, but have to run out now.

But just quickly NOS Dacs (non-oversampling dacs) -- I'm confused I googled it and landed on this

Oversampling adds additional interpolated data based on existing data points whereas up-sampling 'pads' the existing data to reach some arbitrary higher sample rate.

which confuses me even more ... like upsampling is ok but oversampling is bad (or vice versa)  ?

I thought a NOS was if it goes in at 44.1K it goes out at 44.1k, now I'm not sure if most sigma delta dacs oversample, up-sample, or both

[mwillems]

Thanks for the info/links guys and the link to the previous post ... have a bunch of questions, but have to run out now.

But just quickly NOS Dacs (non-oversampling dacs) -- I'm confused I googled it and landed on this
 which confuses me even more ... like upsampling is ok but oversampling is bad (or vice versa)  ?

I thought a NOS was if it goes in at 44.1K it goes out at 44.1k, now I'm not sure if most sigma delta dacs oversample, up-sample, or both

The vast majority of DAC chips oversample seamlessly (i.e. they don't require any configuration, they just do it and it can't be turned off), and oversampling is generally a good thing (that's why almost all DACs do it).  You have to actively look for a NOS DAC at this point: http://www.audiostream.com/content/non-oversampling-nos-dacs-list

Oversampling "smooths" your existing data by interpolation (it guesses what the samples between the existing samples would be to get a smoother result).  So, for example, 44.1KHz music wouldn't "acquire" any high frequency content through oversampling, you just get a smoother waveform.  Upsampling (in software before sending it to your DAC) adds padded data to the top of the frequency range, but does not necessarily do the kind of interpolation described above (does not provide all of the benefits of oversampling).  Upsampling in software is generally harmless (because the DAC will just oversample anyway), but in some DACs may actually prevent the DAC from oversampling.  Most DAC chips oversample at much higher rates than the highest sample rate they're rated for (i.e. at 384, 768 or more), but I've read that some DAC chips don't have that oversampling "overhead."  So by sending upsampled data (at the limits of the DAC) to such a DAC, you prevent it from oversampling and effectively turn it into a NOS (which may or may not be a good thing depending on how you feel about NOSs).  

Then the author goes on the claim that a SACD uses a different master than the same recording for a regular CD...? Really?!
So what, the artist and producers re-perform and record the same music in the studio to make a second master?? Of course they don't. lol
They call it a "master" for a reason.

@Wungun: that part of the article is true for the SACDs I've personally heard. Upon researching the SACDs I've heard, every one of them was either a remaster or a different original master (and it shows if you look at the spectral analysis below 20KHz).  Many early recordings had multiple sets of masters: an original master of the separate recorded parts, and a final "master" that was mixed and ready to press (King Crimson's Lizard, for example, had at least two masters, probably three).  If you research individual SACD releases (read the reviews, forum background, etc.), they're very often a different mix of the album than the redbook version.  Humorously, some SACD producers have even indicated that they do different mixes on SACD hybrid disks for the CD layer and the SACD layer, presumably to "confirm" the impression that the SACD format "sounds better," which seems like a really questionable practice. On the other side, there are also SACDs that are just upsampled 44 or 48 KHz PCM: http://archimago.blogspot.ca/2013/07/list-suspected-44-or-48khz-pcm.html (although I haven't personally encountered any).

If you want to try an experiment, try converting an SACD rip to 44.1/16 and then burn it to a CD and give it a listen.  I've never personally been able to tell the two apart, but some folks say they can.  But try it for yourself and see.

[Wungun]

"Remastered" I can understand...whereby remixing the individual tracks to make a new "master". Using a "Different" master, gives me the impression that the studio recording was reperformed, which obviously wouldn't be done.
I could see the temptation when remixing and producing a SACD for example, the engineer might be tempted to boost this frequency, reverb that instrument, pass it all thru a noise filter, etc.
thanks for bringing that up...
I'd think in this day of digital downloading, there isn't more hi-Rez material available for download, i.e. 24/96 studio/master version of the same CD that you can buy, not just a ploy to sell remixes by greedy record companies.

Okay, so JRiver... Does it up sample or over sample?! Lol

[Poodle]

I think most people don't really understand audio principles. The kinds of arguments you are to-ing and fro-ing are the typical nonsense one would expect from people who are more interested in IT and Computers, out of which some interest in audio has grown.

Here's some bait... those of you who have done enough mastering might want to weigh in on how the file in this screen grab "should" sound.

[InflatableMouse]

"Remastered" I can understand...whereby remixing the individual tracks to make a new "master". Using a "Different" master, gives me the impression that the studio recording was reperformed, which obviously wouldn't be done.

I think you are confused. The intial recordings are not called the master. In fact, many recordings are made initially as instruments are often performed and recorded individually, often many times over (different takes). These are not called "the master".

Mastering is a post-production process for which a subset of the initial tapes or recordings are used. The result is called a master, and is used for reproduction (ie, where all copies or pressings are made from).

To "remaster" literally means to take the initial studio recordings and redo the mastering process, creating a new master.

So yes, many albums have multiple masters. Each time (well, maybe not each time, but often) a remaster is done, a new master is created from which new pressings are made.

At least, this is how I always thought it worked.

[Poodle]

Close. It's usually a two stage process these days. The recorded 20-odd tracks are mixed down to two, left and right. The mix is then usually mastered, with things like analog compressors and EQs, if done properly. Remasters tend to be off the mix, not the original multi track reels/files. To confuse it even more, the original mix is usually referred to as the "master tape".

Of course the other end of the industry uses a glorified version of GarageBand to generate an m4a.

[kstuart]

There are plenty of rebuttals on the Web to the OP's xiph.org article.  Here is some solid information from those rebuttals.

Concerning the idea that humans cannot hear above 22hz:

    Given the existence of musical-instrument energy above 20 kilohertz, it is natural to ask whether the energy matters to human perception or music recording. The common view is that energy above 20 kHz does not matter, but AES preprint 3207 by Oohashi et al. claims that reproduced sound above 26 kHz "induces activation of alpha-EEG (electroencephalogram) rhythms that persist in the absence of high frequency stimulation, and can affect perception of sound quality." [4]
    Oohashi and his colleagues recorded gamelan to a bandwidth of 60 kHz, and played back the recording to listeners through a speaker system with an extra tweeter for the range above 26 kHz. This tweeter was driven by its own amplifier, and the 26 kHz electronic crossover before the amplifier used steep filters. The experimenters found that the listeners' EEGs and their subjective ratings of the sound quality were affected by whether this "ultra-tweeter" was on or off, even though the listeners explicitly denied that the reproduced sound was affected by the ultra-tweeter, and also denied, when presented with the ultrasonics alone, that any sound at all was being played.

    In a paper published in Science, Lenhardt et al. report that "bone-conducted ultrasonic hearing has been found capable of supporting frequency discrimination and speech detection in normal, older hearing-impaired, and profoundly deaf human subjects." [5] They speculate that the saccule may be involved, this being "an otolithic organ that responds to acceleration and gravity and may be responsible for transduction of sound after destruction of the cochlea," and they further point out that the saccule has neural cross-connections with the cochlea. [6]

http://www.cco.caltech.edu/~boyk/spectra/spectra.htm

and

Does High Sampling Frequency Improve Perceptual Time-Axis Resolution of Digital Audio Signal?

The effect of frequency bandwidth on perceptual time-axis resolution of a digital audio signal was studied experimentally. Four kinds of pulse train having constant intervals were used for the test signal. To make the test signal, the upper frequencies were limited using an FIR low-pass filter. Cutoff frequencies of the FIR low-pass filter were 40 kHz and 20 kHz. The experiment was performed in a room with a volume of 54.7 m3, reverberation time of 0.36 s (at 1 kHz), and noise level of NC-25. As the reference signals, pulse trains having designated intervals 1, 0.5, 0.25, and 0.125 s were used. On the other hand, pulse train having slightly different intervals from the reference signals (maximum+20% to -20% were used as the test signals. Reference signals and test signals were presented alternately to the subjects. The subjects were asked whether the test signal is equal to the reference signal or not by hearing. The subjects were 11 males, age 22 to 24, having normal hearing. Purpose of this experiment is to confirm the following hypothesis. Since the waveform of a pulse signal reproduced by the 40-kHz bandwidth system is more similar to the original signal than by the 20-kHz bandwidth system. We can assume that the subjects would be able to identify the interval of the test signal more easily and more correctly using the 40-kHz system thab using the 20-kHz system. When the interval of the test signal is short, identification becomes easier, but when it becomes long, identification becomes rather difficult. As a result of the experiment, we found that widening the frequency band improves perceptual time-axis resolution of a digital audio signal. Adoption of higher sampling frequency will be helpful from the viewpoint of improving the time-axis resolution.

Paper Number: 4562 Convention: 103 (September 1997)
Authors: Yoshikawa, Shokichiro; Noge, Satoru; Yamamoto, Takeo; Saito, Keishi
Affiliations: IMedia Laboratory S, Mitaka, Tokyo, Japan ; Kanagawa Institute of Technology, Atsugi, Kanagawa, Japan ; Pioneer Electronic Corporation, Meguro-lm, Tokyo Japan (See document for exact affiliation information.)

and

Anti-alias and anti-image filtering:
The benefits of 96kHz sampling rate formats for those who
cannot hear above 20kHz.
JulianDunn
NanophonLimited
Cambridge, UK
Email: Jdunn@iee.org
Web: http://www.nanophon.com
ABSTRACT
Reports that 96kHz sampled digital audio systems have greater
transparency than those sampling at 44.1kHz apparently conflict
with knowledge of the capability of human hearing. The band
limiting filters required are examined for a role in producing these
differences. Possible mechanisms are presented for these filters
to produce audible artefacts and filter designs avoiding these
artefacts are illustrated.

http://www.nanophon.com/audio/antialia.pdf

[kstuart]

Something in plainer English:

Here's Robert Harley's rebuttal to a reader touting the Monty theory in the letters column of the March 2013 issue of The Absolute Sound:

I frequently receive e-mails from people who have read some anti-audiophile propaganda on the Internet that asserts 44.1kHz/16-bit audio is perfect, or that all amplifiers or cables sound the same, or other such nonsense. An excerpt from the article you mention: "Unfortunately, there is no point to distributing music in 24-bit/192kHz format. Its playback fidelity is slightly inferior to 16/44.1 or 16/48, and it takes up 6 times the space." The article appears on the surface to be technically credible but the author is naïve at best.​
​
The resolution of a digital audio system isn't determined by the number of bits available in each sample, but rather by the number of bits being used at any given moment. For example, during very low-level passages the signal might be encoded at -80dBFS (80dB below full scale). With 16-bit quantization, a signal at -80dBFS would be encoded with three bits (eight possible quantization steps). With 24-bit quantization, that same -80dB signal would be encoded with 11 bits (2048 possible quantization steps).​

The article also claims that high sampling rates offer no sonic or technical advantage over 44.1kHz because in theory 44.1kHz sampling of a 20kHz bandwidth signal is perfectly lossless. The latter part of that statement is true, but it fails to address the well-documented problem of the time-domain distortion introduced by steep filters with cutoff frequencies close to the audio band. Faster sampling rates allow filters with gentler slopes that have cutoff frequencies well away from the audio band, and thus do less damage to the signal.​
​
Finally, anyone who is actually listened to 44.1kHz/16-bit and 176.4kHz/24-bit versions of the same recording can come to no other conclusion than hi-res digital sounds significantly better.​

[JezQ]

I suspect that most people, like me, don't have sufficient technical knowledge to argue the point in terms of measurements and numbers. But I do remember similar articles in the late 70s full of graphs and tables of figures that 'proved' that competently designed amplifiers couldn't possibly sound different to each other, that it was against the laws of physics, and anyone who thought they did was a fool. I was one of the fools ;-) Still am. Given a great recording to start from, the hi-rez versions sound much better to me, at least they do at 24/96, I can't honestly tell the difference between them and ones at 192, but I don't know if thats the format, the limited tracks I have in both formats, limitations in my equipment, or my old ears. Which is not to say every hi-rez download will sound better than the CD version. Most do, some don't. So non-tech folks relax and enjoy the music, if it sounds better, it is better.