[Archimago]

Just wanted to invite everyone who might not have seen the 24-bit vs. 16-bit audio test I have on the blog post.

3 musical pieces, 2 samples - either 24-bit or dithered down to 16-bit. Have a listen. Can you tell the difference? Think about how easy/hard it was to differentiate... Then complete my survey

I know many here have mad audio technical skillz. Please try to listen 1st before running file through the audio editor!

Collecting data until June then I'll publish a summary analysis. You might have seen this test announcement already on a few of the other sites. So far, I have >50 results already but would love to expand it even further and I know many of you guys out here have excellent gear to evaluate these kinds of things.

http://archimago.blogspot.ca/2014/04/internet-test-24-bit-vs-16-bit-audio.html

All the best,
Archimago

[RDavidson]

I like the photo example at the top of the page. Though I know it is there just for visual reference for the test, I can immediately see the difference in resolution. It's obvious when you know what to look for.

I imagine the same can be said for audio. Once you know what to listen for, loss of resolution becomes glaringly obvious, but, scientifically speaking our hearing is not nearly as acute as our eye sight. In order of importance, our vision is at the top or very near the top of our sensory system.

[Russtafarian]

I'll check it out.

[Russtafarian]

I'd like some clarification on the provenance of the source files.  I understand that one sample for each piece of music is native 24 bit and the other is 16 bit upsampled to 24 bit.  Did you source the 16 bit files as 16 bit files, then upsample them to 24 bit?  Or did you source them as 24 bit files, downsample them to dithered 16 bit, then upsample them back to 24 bit?

I'm not questioning your methodology.  I just want to make sure I understand it.

Russ

[Russtafarian]

O.K., I’ve listened to the samples on a modest $100 USB DAC + headphone combo.  The DAC does 24/96 so no issue there.  I’ll save my file choices for the survey, but I will say the differences I heard were subtle.  I did not hear much difference between a native 24 bit file and the same file dithered to 16 bit and upsampled back to 24 bit.  At least for these examples, you made your point.

If you are trying to make the point that 16 bits is enough, this test doesn't get us there.  We've had nearly 15 years (since the first DAD discs came out) to compare 16 bit files from CDs to 24 bit files from whatever format.  The results are all over the place, but mostly positive.  Here's what I think this means:

1. There are enough examples of 24 bit (or DSD) files that sound better than equivalent 16 bit files to validate the production and distribution of hi-rez audio formats.

2. There are enough examples of 16 bit recordings delivered in 24 bit (or DSD) formats and sold at a premium without offering any increase in sound quality to make music consumers skeptical of hi-rez audio formats.

3. Point 2 does not invalidate point 1.

4. The success of Pono and/or Itunes will accelerate both points 1 & 2.

5. Our role as informed consumers and/or industry professionals is to champion point 1 and mitigate point 2 as much as possible.

Russ

[RDavidson]

I'm curious if one performed the test on a home system (built of quality components) if they'd hear spatial information in the hi res samples that doesn't exist or is less "resolved" in the lower res. With that said, I also agree that with some types of music it may be easier to hear the differences than others.....which may even come down to how well the music was recorded and tinkered with in the studio to begin with.........but, that's a whole other can of worms.

In essence, I think it is entirely possible for a CD to sound as good as a high res download in some cases. And that's where we consumers can get screwed if we trust a number applied to a file (the stated resolution) more than our ears. Unfortunately, not everyone is willing to buy both the CD and high res file of all music they buy to make a comparison. Honestly, someone needs to step in and regulate what constitutes a true high res file and a non high res file..........or always offer both the lower and higher res files with a music purchase. This is similar to what Amazon does. If you buy a CD, you also are given the mp3 for free.

[Davey]

Is the premise here that 24 bits provides "higher resolution" than 16 bits?  If so, I don't agree.  Wider dynamic range window yes, but not higher resolution.

You could make an argument that an increase in sampling rate might provide "higher resolution", but even that argument is a difficult one to make. 

Cheers,

Dave.

[RDavidson]

Is the premise here that 24 bits provides "higher resolution" than 16 bits?  If so, I don't agree.  Wider dynamic range window yes, but not higher resolution.

You could make an argument that an increase in sampling rate might provide "higher resolution", but even that argument is a difficult one to make. 

Cheers,

Dave.

Dave, I'm not sure what's to agree or disagree about here. We're talking science. Not a lot of room for subjectivity in the actual technicalities of the subject matter.

A wider dynamic range IS a large part of what is supposed to make high res music high res. It captures the shimmer of a cymbal way out beyond the range of human hearing, but if you have sensitive ears you may still sense the sound or some type of "pressure" from it, similar to bass frequencies below 20 hz. Consequently, it may also capture more subtleties throughout the entire range, including the midrange.

An increased sample rate, is just that. An increased sample rate. But there's no sense in an increased sample rate on a lower res file. You can't "create" information that doesn't exist in the bits to begin with, just by upsampling. That is COMPLETE nonsense, and if someone argued otherwise, well they have reading to do. With 24 bit, the information IS there and thus a higher sample rate is necessary to capture and retain it during playback.

[Davey]

I think you missed my point.  I'm not talking about subjective, I'm talking about objective.

The dynamic range is in the performance/recording not in the digitization method.  Assuming the performance can be captured within the 16-bit (96.3db) range window, then what objective improvement could be gained with eight more bits of depth?

Your comment regarding sample rate is a strawman.  The "test" presented here by Archimago is regarding bit depth not sample rate, yes?  The sampling rate would have to be the same in either case for a valid test.

What I'm saying here is the whole premise of this particular test is flawed.
If there are audible differences it will be the result of other variables and not the bit depth difference.

Since this thread is regarding a subjective evaluation, a more valid subjective test (IMO) of the robustness of the 16-bit recording method is this:

http://www.aes.org/e-lib/browse.cfm?elib=14195  (good reading.)

Cheers,

Dave.

[RDavidson]

I understand what you're saying and appreciate the explanation. Bringing up a case example isn't really saying much about the entire musical landscape, but you're correct, and this is good food for thought. It may even be a good explanation regarding why some may say their CD sounds just as good or better than their SACD or HR download of the same music.

I see what you're saying regarding sample rate, but let's assume the sample rate is equally high (196 kHz) for the sake of argument for both files AND the music covers the full dynamic range beyond human hearing AND was recorded / mastered perfectly. The 24 bit file WILL have more information than the 16 bit file. I don't think there's any argument there, right?

Thanks for your insight. You definitely bring up very valid points. If the test is flawed, it's no better than those examples used at Best Buy promoting their HD set-up service vs doing it yourself.

And at the end of the day, I'd rather all purchasable music be hi res, made directly from the master tape regardless of dynamic range. At least then you'd know (or at least feel fairly confident) that all the music in your library is the best possible playback resolution/quality possible. This is a major pipe dream given how horrible mastering and processing can be these days, even with all the advanced tools at the disposal of the studios.

[*Scotty*]

Maybe this should viewed as a teachable moment. In a digital recording system the lowest possible THD figure is seen when all bits are present. The number of bits present is a function of signal level, at 0dB all bits are used and the lowest possible THD is observed. In a 16bit system by the time you reach -60dB you have a little over 3% THD.
 In a 24bit system with a DAC operating at about 20bits which is currently the common hardware imposed limit, the THD at -60 is about .05% THD.
A PCM1794 and PCM 1792 both show a .05% THD at -60dB. This is about as good a spec as can be currently found for a 24 bit DAC from online sources
This should be an audible difference in many systems.
As far as the test is concerned, I think it may be flawed. Apparently one is listening for differences between a true 24/96 source and a 16/44.1 source upsampled to 24/96. The approximately 3%THD at -60 db should still be there even when 16/44.1 is upsampled to 24/96. That doesn't mean that the upsampled 16/44.1 source won't sound better to some people than straight 16/44.1. for reasons possibly related to the reconstruction filter design used at higher sample rates.
Obviously I am in the "we need every source to be 24/96" for purely technical reasons related to what I consider a reasonable distortion floor. The 3%THD at -60dB limit of 16/44.1 isn't it.
Scotty

[JerryM]

Obviously I am in the "we need every source to be 24/96" for purely technical reasons related to what I consider a reasonable distortion floor. The 3%THD at -60dB limit of 16/44.1 isn't it.
Scotty

Hear, hear!!! 
 
I'll take the test and complete the survey this weekend. 

[Davey]

I see what you're saying regarding sample rate, but let's assume the sample rate is equally high (196 kHz) for the sake of argument for both files AND the music covers the full dynamic range beyond human hearing AND was recorded / mastered perfectly. The 24 bit file WILL have more information than the 16 bit file. I don't think there's any argument there, right?

A 24-bit file with your "conditions" will have more "information"....obviously.    But it's another strawman.  Do you see that? 

Certainly some good reasons to utilize 24-bit quantization in the recording end of things since it provides greater leeway in signal processing, non-optimum gain structures, etc, etc.
For those of us here in the playback end of things there are some advantages also.  Example:  Adding 8 bits allows digital volume controls to operate at attenuations of -48db without loss of resolution.
However, here we're focused like a laser beam on a different aspect, yes? 

Scotty, THD is a bit of a tangent.  You're aware there are experts out there who claim audibility of harmonic distortion is a meaningless concept?    Have you read some of the work of Earl Geddes?

Cheers,

Dave.

[RDavidson]

A 24-bit file with your "conditions" will have more "information"....obviously.    But it's another strawman.  Do you see that? 

Dave.

I guess I'm missing something here? I don't think I've formed any more straw-men than you have. In fact, pretty much everything I've questioned or added further discussion to is based on your straw-men (ie regarding sample rate, and the somewhat specific case you brought up regarding the dynamic range window).
If there's more information, ie musical information even if it is beyond human hearing, this doesn't mean it doesn't exist (presumably on the master tape) or shouldn't be included in the music we buy. Please enlighten me as you either disagree or I'm missing something.........or you disagree because I'm missing something.......or........

[wushuliu]

Lol, there's a lot of stuff between a downloaded sound file and your ears. I'd guess the chances of anyone telling the difference with a computer and headphones is pretty slim and not many are going to make an effort to play these files through their main rig. But then, the people who think there's no difference don't believe any of that matters anyway so round and round we go.

[Davey]

I guess I'm missing something here? I don't think I've formed any more straw-men than you have. In fact, pretty much everything I've questioned or added further discussion to is based on your straw-men (ie regarding sample rate, and the somewhat specific case you brought up regarding the dynamic range window).
If there's more information, ie musical information even if it is beyond human hearing, this doesn't mean it doesn't exist (presumably on the master tape) or shouldn't be included in the music we buy. Please enlighten me as you either disagree or I'm missing something.........or you disagree because I'm missing something.......or........

How are the sample rate and dynamic range comments I made a strawman?  You noted the conditions I placed on both of those with my statements, did you not?

Do these test tracks from Archimago have a dynamic range greater than 96db?  After you've performed you listening evaluation take a look at the Norwegian Armed Forces track.  It's interesting.

Cheers,

Dave.

[G Georgopoulos]

cd as original specs is very good (16/44)
most artists produce cd's
cd was ahead of its time and alot
of great engineers had contributed
personally i dont like mp3's but they
play well despite compression

all i say cd is very good...

[RDavidson]

How are the sample rate and dynamic range comments I made a strawman?  You noted the conditions I placed on both of those with my statements, did you not?

Do these test tracks from Archimago have a dynamic range greater than 96db?  After you've performed you listening evaluation take a look at the Norwegian Armed Forces track.  It's interesting.

Cheers,

Dave.

I don't know. I'm just pointing out that you say my comments are a strawman, when they are no more of such than your own original comments regarding dynamic range and sample rate, with ZERO explanation as to what you're talking about. Only when prodded, you elaborated. So..........how are my comments a strawman and your's aren't?

Yes, I agree more info is necessary regarding the source material to understand if the test is truly valid.....which I think was your main point. So, let's just move on, shall we?

[Archimago]

One more month to go folks!

Many very thoughtful responses to the survey already so if you have gear capable of high-res, please give this a try!

BTW: Some might be suspicious of whether I have some nasty trick up my sleeve... No, not really, I am indeed just testing 16-bit vs. 24-bits with the same recording!

[maty]

Hi-Rez Audio Distinguished in Blind Testing, by Jim Austen Dec 16, 2019

https://www.stereophile.com/content/hi-rez-audio-distinguished-blind-testing

For me it is a matter settled for years. True, it is only on my two stereo systems. In others, much more expensive, I was unable to differentiate even MP3 LAME 320 kbps from FLAC 16/44 (from my USB memory). As I have been eliminating bottlenecks in my second system the differences have increased.

You can already have the best system in the world, with there being an important bottleneck somewhere this will be the limiting factor.