• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). There are daily reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

CD vs hi-res

SIY

Master Contributor
Technical Expert
Joined
Apr 6, 2018
Messages
8,398
Likes
18,897
Location
Alfred, NY
Simple soundstage with speakers is where you can mostly hear differences. You will not so much hear sonic differences as you can hear differences in how the soundstage changes. Also, since many if use computers as a player, I am not sure how much that affects the reproduction but that is an argument for another day.
Evidence?
 

ThatM1key

Addicted to Fun and Learning
Forum Donor
Joined
Mar 27, 2020
Messages
825
Likes
642
Location
USA
I would argue, that other than different mastering methods, headphones may not give you enough information. Where I am going with this? Simple soundstage with speakers is where you can mostly hear differences. You will not so much hear sonic differences as you can hear differences in how the soundstage changes. Also, since many if use computers as a player, I am not sure how much that affects the reproduction but that is an argument for another day.
If were talking about PC's and not computer-based electronics like CD players and if this was the 1980s and 1990s, Sure those PC's couldn't handle and replay CD quality content but these days were lucky. You can easily build/own a PC that beats all 1980s CD players by using a good DAC (That are super affordable), good software & WASAPI/ASIO output. Heck I could even Rip the CD and never have to worry about disc rot.
 

anotherhobby

Active Member
Forum Donor
Joined
Dec 17, 2021
Messages
252
Likes
468
I've read several pieces from people in the recording industry on why I should not be able to tell the difference between 16/44 and high res, and I generally believe them. Of course there are also many articles saying the same about 360mp3/256AAC vs 16/44, but I can totally tell that difference on quality speakers. I decided to try and test 16/44 vs hi-res the other day and I could hear a difference. I'm assuming it's my test that's flawed, but I'm not sure why (other than it's not blind). On a Mac, using Audio MIDI Setup, you can change your audio output format on the fly between 16/44 and 24/192 (and other formats). It'll drop audio for about 1/2 sec while it switches. That's what I used to down sample.

I have the 75th anniversary Blue Note collection, and playing Kind of Blue 24/192 thru a NAD 2030 V2 into Revel M105s near field, if I switch from 24/192 down to 16/44, I can tell a very very subtle degradation. It's so subtle that I can only hear it when going down in resolution. I cannot tell when I switch it back up to 24/192. I also don't think I could tell if it was a 2 second gap instead of 1/2 second. But I do hear it in this test, just every so barely, in the sound stage. I'd think that if I hear can it when going down, then I should hear also it get better when I flip back up, but I can't.

What I do know is that the change is so insanely subtle that I can't reliably tell with just a 1/2 gap, while focusing my ears as hard as I can, listening to an album I know inside and out, looking for the change, and knowing it's going to happen (not blind). It convinced me that any differences there may be are so infinitely subtle as to be not meaningful in any way. I have don't wonder if I'm somehow missing out on anything by listening to 16/44 content most of the time.

I've been thinking that any difference I was hearing could be due to the how the Mac was down sampling it on the fly, but I have no idea. That's the most curious part to me, I really wonder why I can even hear the change at all?
 

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
14,686
Likes
23,831
I've read several pieces from people in the recording industry on why I should not be able to tell the difference between 16/44 and high res, and I generally believe them. Of course there are also many articles saying the same about 360mp3/256AAC vs 16/44, but I can totally tell that difference on quality speakers. I decided to try and test 16/44 vs hi-res the other day and I could hear a difference. I'm assuming it's my test that's flawed, but I'm not sure why (other than it's not blind). On a Mac, using Audio MIDI Setup, you can change your audio output format on the fly between 16/44 and 24/192 (and other formats). It'll drop audio for about 1/2 sec while it switches. That's what I used to down sample.

I have the 75th anniversary Blue Note collection, and playing Kind of Blue 24/192 thru a NAD 2030 V2 into Revel M105s near field, if I switch from 24/192 down to 16/44, I can tell a very very subtle degradation. It's so subtle that I can only hear it when going down in resolution. I cannot tell when I switch it back up to 24/192. I also don't think I could tell if it was a 2 second gap instead of 1/2 second. But I do hear it in this test, just every so barely, in the sound stage. I'd think that if I hear can it when going down, then I should hear also it get better when I flip back up, but I can't.

What I do know is that the change is so insanely subtle that I can't reliably tell with just a 1/2 gap, while focusing my ears as hard as I can, listening to an album I know inside and out, looking for the change, and knowing it's going to happen (not blind). It convinced me that any differences there may be are so infinitely subtle as to be not meaningful in any way. I have don't wonder if I'm somehow missing out on anything by listening to 16/44 content most of the time.

I've been thinking that any difference I was hearing could be due to the how the Mac was down sampling it on the fly, but I have no idea. That's the most curious part to me, I really wonder why I can even hear the change at all?
Could you supply one track to me. I'd like to see how Deltawave compares the files. I have the CD for Kind of Blue. If I had your 192 version Deltawave can make the comparison to 44.1 khz.
 

BeerBear

Active Member
Joined
Mar 9, 2020
Messages
106
Likes
98
I've read several pieces from people in the recording industry on why I should not be able to tell the difference between 16/44 and high res, and I generally believe them. Of course there are also many articles saying the same about 360mp3/256AAC vs 16/44, but I can totally tell that difference on quality speakers. I decided to try and test 16/44 vs hi-res the other day and I could hear a difference. I'm assuming it's my test that's flawed, but I'm not sure why (other than it's not blind). On a Mac, using Audio MIDI Setup, you can change your audio output format on the fly between 16/44 and 24/192 (and other formats). It'll drop audio for about 1/2 sec while it switches. That's what I used to down sample.

I have the 75th anniversary Blue Note collection, and playing Kind of Blue 24/192 thru a NAD 2030 V2 into Revel M105s near field, if I switch from 24/192 down to 16/44, I can tell a very very subtle degradation. It's so subtle that I can only hear it when going down in resolution. I cannot tell when I switch it back up to 24/192. I also don't think I could tell if it was a 2 second gap instead of 1/2 second. But I do hear it in this test, just every so barely, in the sound stage. I'd think that if I hear can it when going down, then I should hear also it get better when I flip back up, but I can't.

What I do know is that the change is so insanely subtle that I can't reliably tell with just a 1/2 gap, while focusing my ears as hard as I can, listening to an album I know inside and out, looking for the change, and knowing it's going to happen (not blind). It convinced me that any differences there may be are so infinitely subtle as to be not meaningful in any way. I have don't wonder if I'm somehow missing out on anything by listening to 16/44 content most of the time.

I've been thinking that any difference I was hearing could be due to the how the Mac was down sampling it on the fly, but I have no idea. That's the most curious part to me, I really wonder why I can even hear the change at all?
Your testing method is flawed, indeed, for several reasons.
First of all, of course, because it's sighted (not blind).
Second, because you have too many variables that can affect the result.
For instance, it's a good idea to convert the 192/24 to 44/16 and then back to 192/24 again (with a good SRC/dither), and then compare that file to the original 192/24. This way you eliminate the artifacts/pauses that happen when changing the audio settings. An audio device can perform differently at different sample rates, too.
You also need to carefully level match the files you're comparing, to make sure they're equally loud. A good SRC alone shouldn't affect the loudness significantly, but something like an MP3 or AAC conversion could. Lossy encoding/decoding could even add some silence at the start or end of the file.
These are just some basic things that you need to pay attention to.

Also, I don't think an analog recording from the 1950s makes much sense for "high res" testing/listening...
 
Last edited:

SIY

Master Contributor
Technical Expert
Joined
Apr 6, 2018
Messages
8,398
Likes
18,897
Location
Alfred, NY
Also, I don't think an analog recording from the 1950s makes much sense for "high res" testing/listening...
Indeed. When I was the guinea pig for an MP3-vs-WAV test, I scored perfectly on a modern music piece, and totally random on Take Five.
 

anotherhobby

Active Member
Forum Donor
Joined
Dec 17, 2021
Messages
252
Likes
468
Your testing method is flawed, indeed, for several reasons.
First of all, of course, because it's sighted (not blind).
Second, because you have too many variables that can affect the result.
For instance, it's a good idea to convert the 192/24 to 44/16 and then back to 192/24 again (with a good SRC/dither), and then compare that file to the original 192/24. This way you eliminate the artifacts/pauses that happen when changing the audio settings. An audio device can perform differently at different sample rates, too.
You also need to carefully level match the files you're comparing, to make sure they're equally loud. A good SRC alone shouldn't affect the loudness significantly, but something like an MP3 or AAC conversion could. Lossy encoding/decoding could even add some silence at the start or end of the file.
These are just some basic things that you need to pay attention to.

Also, I don't think an analog recording from the 1950s makes much sense for "high res" testing/listening...
I'm aware of the blind aspect, which I mentioned. To me, if I can't tell or can only barely tell and I know I'm not confident, then blind is just going to confirm that I can't really tell. It's just not necessary for negative results, because if I can't tell when I know, how on earth would I tell if I didn't know? With the test I did above with Kind of Blue, I am convinced a blind test would show the same lack of confidence (or likely more) that I have when I know. If I felt I could determine it confidently, then a blind test could confirm if it is or is not placebo.

For down sampling the bitrate, I was thinking that it's possible that the Mac doing the work on the fly could be tainting it, but I noticed that in Audio MIDI Setup the clock source is the Topping's internal clock, and the Topping is actually doing the conversion, not the Mac. That makes me no longer suspicious of the conversion. I also don't think level matching matters here when it's the same file being down sampled on the fly. The volume sounds absolutely identical between them, but I'm open to a good technical explanation if somebody can explain otherwise.

On your last point about Kind of Blue being an older recording, I do agree as I was also wondering the same thing while I was writing my post, even if the recording very well regarded. I have plenty of other hi-res files so I did the same testing on some newer albums, and here are my results:
  • Arcade Fire - Everything Now @ 24/96 - switching between 16/44 is more obvious to me for sure, and I feel I can tell when switching either direction
  • Gregory Porter - Liquid Spirit @ 24/192 - can also tell either way, but more a little more subtle than Arcade Fire and more content dependent
  • Gary Clark Jr - Blak and Blu @ 24/96 - cannot tell at all on any song (also, recording does not sound as good/open/wide as the above)
  • Madeleine Peyroux - Standing on the Rooftop @ 24/96 - cannot tell at all
I have many more, but stopped there. Since I feel confident I can tell with the Arcade Fire album when I know I'm switching it, I'm tempted to try and write some code to control the flipping process so I can blind test myself on it. I don't currently feel confident saying I actually can detect the difference unless it's blind, but I think I might be able to. It should be pretty trivial to randomize it and do fake flips where I just drop audio for the same time window it takes to switch bitrates. I'd have my code record what it did and then compare that to what I write down.
 

anotherhobby

Active Member
Forum Donor
Joined
Dec 17, 2021
Messages
252
Likes
468
I don't currently feel confident saying I actually can detect the difference unless it's blind, but I think I might be able to. It should be pretty trivial to randomize it and do fake flips where I just drop audio for the same time window it takes to switch bitrates. I'd have my code record what it did and then compare that to what I write down.
It was pretty quick to whip up a script to do a blind test on myself using Arcade Fire's Everything Now album at 24/96. I can only switch bits between 24 and 36, so I can't test for 16 vs 24 bits, but I can test for sample rate. The code sleeps for 10 seconds, then using a random number generator for choice, it'll either flip the DAC between 44.1 and 96 kHz or it won't. With either choice there is the same 1/2 second silence to cue me to make a guess. then it sleeps for 10 seconds and repeats until I stop it. I wrote down on paper if I thought the sample rate stayed the same as the previous 10s, or if I thought it went up or down. I did two different tests of 25 passes. In one test I allowed myself to see if I was correct after each guess, and on the other one I didn't look until I was done. I used 3 different tracks during the test and scored 18 out of 25 correct on both tests, so 72% accuracy over 50 total passes. Of note, on the test where allowed myself to check and see if the DAC display switched after each guess, my misses were strongly correlated to when the 1/2 second silence occurred during more dramatic changes in the music.

So at least on this album, 72% of the time I can determine a difference between 44.1 and 96 kHz, and tell you which is which.
 

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
14,686
Likes
23,831
Well I was able to compare the 192 khz Kind of Blue with my Sony Super Bit-mapped CD of Kind of Blue. Unfortunately I couldn't get the usual analysis from Deltawave as the timing varies too much between them. Not unexpected as they likely played back the tapes on different RTR tape machines and they are going to vary in speed too much. Obviously these two recordings have different masters.

The CD I have drops steeply in response below 50 hz. It is a couple db hotter between 200 hz and about 5 khz. Then the 192 khz version is hotter above 5 khz by about 3 or 4 db.

None of this explains why you would hear a difference if you downsample your own 192 khz copy of the album. Sorry I couldn't tell you more.
 

BeerBear

Active Member
Joined
Mar 9, 2020
Messages
106
Likes
98
It was pretty quick to whip up a script to do a blind test on myself using Arcade Fire's Everything Now album at 24/96. I can only switch bits between 24 and 36, so I can't test for 16 vs 24 bits, but I can test for sample rate. The code sleeps for 10 seconds, then using a random number generator for choice, it'll either flip the DAC between 44.1 and 96 kHz or it won't. With either choice there is the same 1/2 second silence to cue me to make a guess. then it sleeps for 10 seconds and repeats until I stop it. I wrote down on paper if I thought the sample rate stayed the same as the previous 10s, or if I thought it went up or down. I did two different tests of 25 passes. In one test I allowed myself to see if I was correct after each guess, and on the other one I didn't look until I was done. I used 3 different tracks during the test and scored 18 out of 25 correct on both tests, so 72% accuracy over 50 total passes. Of note, on the test where allowed myself to check and see if the DAC display switched after each guess, my misses were strongly correlated to when the 1/2 second silence occurred during more dramatic changes in the music.

So at least on this album, 72% of the time I can determine a difference between 44.1 and 96 kHz, and tell you which is which.
The percentage of correct guesses alone isn't enough, because it depends a lot on how many tests you perform. A more useful number is the p-value aka the probability of getting this result by random guessing. In your case (18 out of 25) p=0.0216, which is better than the minimum standard required (lower than 0.05), but it's not very convincing. If you got just one less right (17 out of 25) it would not be enough.
But I can believe that you are hearing a difference and not just guessing. It's just that I would still recommend that you test with converted files first, reducing the influence of audio device switching and sample rate settings... And also because it's overall easier and there are already file ABX testing tools out there, like foobar2000 (for Windows, but I'm sure there's something for macOS too).
 

anotherhobby

Active Member
Forum Donor
Joined
Dec 17, 2021
Messages
252
Likes
468
The percentage of correct guesses alone isn't enough, because it depends a lot on how many tests you perform. A more useful number is the p-value aka the probability of getting this result by random guessing. In your case (18 out of 25) p=0.0216, which is better than the minimum standard required (lower than 0.05), but it's not very convincing. If you got just one less right (17 out of 25) it would not be enough.
But I can believe that you are hearing a difference and not just guessing. It's just that I would still recommend that you test with converted files first, reducing the influence of audio device switching and sample rate settings... And also because it's overall easier and there are already file ABX testing tools out there, like foobar2000 (for Windows, but I'm sure there's something for macOS too).
Good info on the p-value, and your explanation aligns with what I felt the result showed. I did not take the 72% accuracy to be evidence that I can reliably tell the difference, especially knowing how much I had to focus and concentrate. To the contrary, I thought it was a pretty low detection score and I took this as evidence that 44.1 kHz is absolutely good enough for me, and I can basically ignore the notion of higher sample rates. The difference is so infinitely small to me that there is no meaningful difference. That makes me happy since that's the sample rate of most available lossless content.
 

makinao

Member
Joined
May 11, 2019
Messages
52
Likes
64
I just want the unadulterated final master copy. I want hi-res as long as it's about skipping the downsampling step of converting the original final master to cd spec. So for me it's about removing an unnecessary step in the chain. Not about whether or not I can hear a difference. So I guess provence is more important to me. Like I want something that authenticates that I'm receiving master quality...
I did that once on an EP I produced back in 2014. We originally planned to make the mp3 and 16/44.1 versions available on bandcamp. But when I found out the engineer recorded in 24/88.2, I told the artists manager why not offer that too in case anyone wanted an exact same copy of the recording? So three versions were put up

Subjectively, I thought the 24/88.2 version had more transparent highs. But I didnt do any extensive analysis of the waves, so I can‘t confirm any measurable difference.
 

Offler

Senior Member
Joined
May 27, 2021
Messages
322
Likes
264
I can only provide subjective experience on the topic.

a) Compression vs uncompressed/lossless
I prefer uncompressed/lossless formats, as the original content can be recreated bit-by-bit. Compression produce certain artifacts, which may become annoying once you know how they sound.

b) 16 or 24 bits.
In general I cant tell any difference. It can be measured over oscilloscope, but nothing audible.

c) 44.1, 48, 88.2, 96 or 192KHz.
Resampling between 44.1 / 48 unfortunately can be heard a bit, when using Windows. Avoiding it or using 192KHz as an output does solve it.

If CD quality is being played "as is" over Wasapi Exclusive and any "higher res" in same manner, I would not be able to tell difference if both media are lossless.
 

Robin L

Major Contributor
Joined
Sep 2, 2019
Messages
4,409
Likes
6,205
Location
1 mile east of Sleater Kinney Rd
I saw "Dr. Mark" and thought someone was talkin' about me... oops. ;)
Solipsism
is a terrible condition. I should start a fund to help stamp it out in my lifetime, eh? ;):cool::facepalm:
R.jpg
 

BN1

Active Member
Joined
Feb 9, 2020
Messages
115
Likes
59
Quick story: w/old ears, damaged hearing I typically "fail" the on-line tests about sound quality (at best, guessing). We listen mostly to background music via distributed audio into ceiling speakers. I moved from Pandora free to Spotify free (160 kbps) and eventually got tired of Sportify's intrusive ads and recently got a trial 3-month Amazon Music HD (CD/96/192). I was surprised by the sound quality on my decent 5.1.2 system but couldn't tell much/any difference via distributed audio (downscaled to CD). Perhaps some bias playing in on primary system or maybe only louder, who knows ?
 
Top Bottom