How good are your ears?

U

unhappy_mage

[H]ard|DCer of the Month - October 2005
Joined
Jun 29, 2004
Messages
11,455
In the GenMay discussion of the Sonos player, some people asserted that merely digitizing music destroyed some of the musicality present in the original analog work. I think they're being ridiculous. But maybe their equipment is just that much better than mine. I'd like to find out. My theory is that nobody can tell the difference between two digital files at different sample depths, let alone analog versus digital. I realize this is somewhat dodging the issue, but it's a little easier to ship around digital files than records or whatever format of analog recording would be best.

I did a test a while back comparing (at 44 kHz) 12 bits of precision to 16, and nobody could tell the difference. I did hear complaints about the source, though, so let's step up the ante. Anyone who has professional recording equipment and a good mic, I'd appreciate your help here. Capture a sample at (preferrably) 96/24 of anything - guitar, cymbal crash, whatever. Email it to me; keep it under 10 megs, please. That's pretty short; 15 seconds or so, assuming it's stereo. But I'm not too terribly concerned with the format here. All I ask is untouched capture - no reverb, no compression, not even normalization.

What I'll do is take this audio and run it through a program that makes the low 4 bits all zero. All this does is reduce the precision in the vertical direction. Then people can use the ABX comparator with Foobar to determine which track they think is which.

I'm curious to see how this will turn out. I have a sample if nobody steps up, but I'd like to hear some more professional equipment.

 
sounds like an interesting experiment...I'm game, but unfortunately I don't have any professional-caliber equipment to work with. All I have are some studio recordings from high school of my concert/jazz/marching bands...but I don't think I have any untouched material. I think it's all become mp3's by this point.
 
I've done tests with a few CD's (Enya, Dire Straights, some Bluegrass, and a few others), from the source material to 192Kbps/256 had no audio differences. If I listen real close, I can tell a difference with 192 sometimes, but at 256 there is no difference. I'm using an M&K speaker setup with a Velodyne 10" sub powered by a Denon receiver from a SB SB X-Fi sound card, Foobar, LAME ripped. The ripping process probably wasn't the best, but it gave great results.

I have my entire CD collection ripped to MP3's... That took a while.

Amanda ;)
 
Amanda said:
I've done tests with a few CD's (Enya, Dire Straights, some Bluegrass, and a few others), from the source material to 192Kbps/256 had no audio differences.
Click to expand...
I don't think anyone is going to assert that different sample rates sound better or worse. Mage is referring to the sample depth aka second half of 44/16, 96/24 etc.
 
Amanda said:
I've done tests with a few CD's (Enya, Dire Straights, some Bluegrass, and a few others), from the source material to 192Kbps/256 had no audio differences. If I listen real close, I can tell a difference with 192 sometimes, but at 256 there is no difference. I'm using an M&K speaker setup with a Velodyne 10" sub powered by a Denon receiver from a SB SB X-Fi sound card, Foobar, LAME ripped. The ripping process probably wasn't the best, but it gave great results.

I have my entire CD collection ripped to MP3's... That took a while.

Amanda ;)
Click to expand...
Yuck, I can't stand mp3's for CD backups. If I'm forced to use mp3's, i'll use VBR mp3 and put it at the highest quality I can. It does give good compression that way. I use WMA Lossless because it's quick and easy along with everything supporting it. Can't go wrong with that.
 
well, you can definitely tell the difference between a 320kbps MP3 and a CD.

just listen to the cymbals, they're the giveaway first if you don't have very good speakers. if you do have a good sound system, the sound imaging and soundstage get screwed up.

you can also tell the difference between 24 bit 192khz signal and 16 bit 44.1khz signal.

my hearing is ok. i can hear down to pretty much anything, and up to about 19khz.
it's quite obvious if a file is an mp3 or a wav to me. analog recordings are better IMO. my dad has a reel to reel and i recorded from a CD onto it and played it back. it sounded better than the CD. i don't know how that happened. its a Pioneer RT-707 if you're interested in that kind of thing.

if you sat me down in front of a good hifi system, and played me a song, one version 24 bit and one version 16 bit, i'm sure i could tell which one was 24 bit. but if i just heard a digital recording, i wouldn't know the bit depth of it off the bat.

i can hear the difference between 14 bit sampling, and 16 bit sampling. there's more noise in the background, and its just not as clear
 
with analog recordings there is a "warmth" associate with them because the sound is run through so many pieces of equipment where as digital is so perfect it has no "soul". I did a lot of test of this stuff in high school, and i would take an analog recording transfered to cd simply because it sounds better.

Also on high high end systems the differences in recordings become to great that it is really easy to tell the difference in them. But again who here has a cool $25,000 - $100,000 to spend on a sound system. in those setups you can hear scratched on cds, which is absolutely insane.

cheers
 
Yeah the better your sound system is the more difference you're going to hear between high quality sources. If you are amplifying something loud enough small differences in the source material can be readily evident. Of course if you are using your computer speakers no you probably can't tell the difference.
 
Well, since noone's stepped up yet, I'm gonna start the tests soon, using the 'Modern' sample from the Universal Audio webzine. I've emailed the author asking permission; when I get back from class tonight I'll probably be able to post original and chopped versions. Just gotta write the program to do the chopping first :rolleyes:

 
unhappy_mage said:
In the GenMay discussion of the Sonos player, some people asserted that merely digitizing music destroyed some of the musicality present in the original analog work. I think they're being ridiculous. But maybe their equipment is just that much better than mine. I'd like to find out. My theory is that nobody can tell the difference between two digital files at different sample depths, let alone analog versus digital.
Click to expand...

well here is the thing

the same EXACT thing is happening in the the DISPLAY world. Everything going to digital there. Yet to this very day the Digital display can not touch a CRT in color since most all digital displays have awful black levels which mean the entire gray scale is off. Some people don't care, some do. Not much different in the audio world. Analog v. digital. It is close but digital still has a bit of catching up to do in the audio world. remember as I stated digital is a DIGITAL representation of the actual wave form. Remember in the digital world the most complex object to handle is a circle and guess what the crest and the trough of the wave form is? :) No it is not square!! ;) hehe
 
figgie said:
Yet to this very day the Digital display can not touch a CRT in color since most all digital displays have awful black levels which mean the entire gray scale is off.
Click to expand...
Well, this is a little off the subject, but I'll bite; what about digital CRTs? Sure, it's mostly a gimmick (they have digital->analog converters inside), but all you can ever really do is move the DA conversion a little closer to the output. I think you're confusing LCD with digital.
figgie said:
Remember in the digital world the most complex object to handle is a circle and guess what the crest and the trough of the wave form is?
Click to expand...
Sure, but after a certain resolution you can't really tell the difference any more. Render a circle at 3000x3000px (this is a chunk, don't worry about memory usage), and it looks pretty dang circular. At some point your eyes aren't good enough to see the difference, or your ears aren't good enough to hear the difference.

This thread is my assertion that current high-res audio is good enough that people can't tell the difference between uber-high-res and slightly-less-high-res. In other words, it's good enough.

 
Considering it's been mathematically proven that there is 0 difference between 44.1khz and 92khz for playback, I'm going to call BS on this. Have you ever done ABX testing to verify this? If so I would like to see the test data. Even --V0 or --V1 to a CD is very hard if not impossible to tell.

>95% of people can't tell the difference between APS fast-vbr and a CD, let alone 320kbps and an actual CD. I want to see test data to back this up.


mike_6289 said:
well, you can definitely tell the difference between a 320kbps MP3 and a CD.

just listen to the cymbals, they're the giveaway first if you don't have very good speakers. if you do have a good sound system, the sound imaging and soundstage get screwed up.

you can also tell the difference between 24 bit 192khz signal and 16 bit 44.1khz signal.

my hearing is ok. i can hear down to pretty much anything, and up to about 19khz.
it's quite obvious if a file is an mp3 or a wav to me. analog recordings are better IMO. my dad has a reel to reel and i recorded from a CD onto it and played it back. it sounded better than the CD. i don't know how that happened. its a Pioneer RT-707 if you're interested in that kind of thing.

if you sat me down in front of a good hifi system, and played me a song, one version 24 bit and one version 16 bit, i'm sure i could tell which one was 24 bit. but if i just heard a digital recording, i wouldn't know the bit depth of it off the bat.

i can hear the difference between 14 bit sampling, and 16 bit sampling. there's more noise in the background, and its just not as clear
Click to expand...
 
Fryguy8 said:
Considering it's been mathematically proven that there is 0 difference between 44.1khz and 92khz for playback, I'm going to call BS on this. Have you ever done ABX testing to verify this? If so I would like to see the test data. Even --V0 or --V1 to a CD is very hard if not impossible to tell.

>95% of people can't tell the difference between APS fast-vbr and a CD, let alone 320kbps and an actual CD. I want to see test data to back this up.
Click to expand...

I'd like to see the mathematical proof showing that there is 0 difference between 44.1kHz and 96kHz playback.
 
I am 100% good to go according to my school's hearing test, but doing my own, my right ear is not as crystal clear as my left ear, although I can hear things as good as my left, it doesn't have that crisp to it. sigh... anyone know a way to make my left ear not as clear as my right??? LOL
 
aZn_plyR said:
I am 100% good to go according to my school's hearing test, but doing my own, my right ear is not as crystal clear as my left ear, although I can hear things as good as my left, it doesn't have that crisp to it. sigh... anyone know a way to make my left ear not as clear as my right??? LOL
Click to expand...

Rock concerts.

-----

I think the bottom line with audio is that there is no blanket "good enough." There is no universally "good enough" headphone, speaker, cable, amp, source, recording, etc. Everyone has different levels of hearing ability/pickiness, and certainly someone out there with some equipment out there can tell the difference between many different levels of digital source quality. The same is true for displays, computers, food, pens, and anything else with a vast range of quality.
 
jon67 said:
I'd like to see the mathematical proof showing that there is 0 difference between 44.1kHz and 96kHz playback.
Click to expand...

nyquist algorithm.
 
unhappy_mage said:
In the GenMay discussion of the Sonos player, some people asserted that merely digitizing music destroyed some of the musicality present in the original analog work. I think they're being ridiculous. But maybe their equipment is just that much better than mine. I'd like to find out. My theory is that nobody can tell the difference between two digital files at different sample depths, let alone analog versus digital. I realize this is somewhat dodging the issue, but it's a little easier to ship around digital files than records or whatever format of analog recording would be best.

I did a test a while back comparing (at 44 kHz) 12 bits of precision to 16, and nobody could tell the difference. I did hear complaints about the source, though, so let's step up the ante. Anyone who has professional recording equipment and a good mic, I'd appreciate your help here. Capture a sample at (preferrably) 96/24 of anything - guitar, cymbal crash, whatever. Email it to me; keep it under 10 megs, please. That's pretty short; 15 seconds or so, assuming it's stereo. But I'm not too terribly concerned with the format here. All I ask is untouched capture - no reverb, no compression, not even normalization.

What I'll do is take this audio and run it through a program that makes the low 4 bits all zero. All this does is reduce the precision in the vertical direction. Then people can use the ABX comparator with Foobar to determine which track they think is which.

I'm curious to see how this will turn out. I have a sample if nobody steps up, but I'd like to hear some more professional equipment.

Click to expand...


The problem is...you digitize.

Take for example a CD of Pink Floyd, and play it on a stereo hifi, while doing an A-B test with the analog record with a high quality pickup (Shure V5 quality or better).

It will only take a few minutes to realize which one sounds better. The CD will have less hiss and pop but the record will still sound better. Then again, this debate has been going on since 1984 when CD Players first hit the market (Probably even before that with the laserdisc video disc)
 
Hollow4 said:
with analog recordings there is a "warmth" associate with them because the sound is run through so many pieces of equipment where as digital is so perfect it has no "soul". I did a lot of test of this stuff in high school, and i would take an analog recording transfered to cd simply because it sounds better.
cheers
Click to expand...

The answer to it was the mechanical reproduction means produce an even order harmonic the brain percieves as natural sounding. This is mostly because nature doesn't produce odd order harmonics where digital signals and processing does.

There were some Carver CD Players that had converters that specifically produced even order harmonics and blended it in to give CD's the natural sound...and it worked. A few years ago, Via produced their 'Vinyl' codec to create the same sound, and it works as well. (which is why listening to music on a via vinyl sound card does in fact beat the sound of my X-Fi, which I experienced first hand repairing an old PC with a cheap motherboard with an integrated via vinyl codec)

Music is still one of those things that there is no absolute recording or reproduction of...just close proximations. Add to that the fact some music is designed WITH digital effects...well, who know what is the real sound. The way I have learned and can judge by is a live drum/guitar player (or orchestra) in front of me and then hearing the same performance from a recording.

Of course, a live to recorded performance comparison is almost useless because there is not an amplifier/speaker system in existance that can reproduce the full dynamic of a snare drum 3 or four feet away from you in real life, so you can only be judgemental to a certain degree.
 
aZn_plyR said:
I am 100% good to go according to my school's hearing test, but doing my own, my right ear is not as crystal clear as my left ear, although I can hear things as good as my left, it doesn't have that crisp to it. sigh... anyone know a way to make my left ear not as clear as my right??? LOL
Click to expand...


The last real hearing test I did was about 11 years ago in my exit examinations when I got out of the Navy. Compared to the hearing test when I first went in some 10 years before that, I had lost a few db sensitivity to certain midrange frequencies pretty much in both ears...
Some loss was attributable to natural aging and some due to loud machinery and loud music exposure, as well as gun range practice and hunting.
 
All the Nyquist theorem talks about is the sampling rate required to represent a signal. If I ever get this dang read/write thing working with 24 bit audio, I'll make a test signal to prove you can hear the difference.

Right now I've got it reading the file and writing one about two-thirds the size, which seems to indicate that the perl module I'm using doesn't understand 24-bit wav files. Maybe I'll submit a patch. :(

 
Yes, so if you can represent a signal in 44.1, what's the point in representing it in 96?

Hint: no point. For playback at least. It comes in handy when doing production/reproduction.
 
BBA said:
The problem is...you digitize.

Take for example a CD of Pink Floyd, and play it on a stereo hifi, while doing an A-B test with the analog record with a high quality pickup (Shure V5 quality or better).

It will only take a few minutes to realize which one sounds better. The CD will have less hiss and pop but the record will still sound better. Then again, this debate has been going on since 1984 when CD Players first hit the market (Probably even before that with the laserdisc video disc)
Click to expand...

I like your example, in fact this was my explaination posted in the GenMay forum where this conversation started:

BillR said:
Most of you are right and at the same time a bit wrong as well.

I’ll try to explain so bear with me please.

Many of you are not old enough to remember “records” maybe some of you are. Back in “the day” when standards were set for the “LP” there was an agreed upon limit of 5 HTZ for the bass or bottom end, but no one set limits for the highs or treble end of recordings.

Most record albums actually had real information recorded well above the 50 KHTZ range. No, your ears can’t hear that and that’s fact. Now here comes the “But” part.

Music is largely based on “harmonics” and “over tones” and many of those harmonics go way out to 100 KHTZ and beyond. Those harmonics bounce around and return to the audible spectrum and become part of the music. It is this very phenomenon that allows us to hear the difference between makes of pianos, guitars and most other instruments.

Harmon Kardon recognized this many years ago and began producing audio gear capable of 100 KHTZ and above and a major reason they used to be one of the really big sellers in the audio industry. Many HK dealers actually had a filter that was inserted between the turntable and the receiver or preamp with a three-position switch. One setting cut of any frequency above 20K, the next setting cut off at 50K and the last was straight through.

Believe me, I sold tons of HK audio because practically everyone who took the time to listen could hear the difference. All these demos were done on under $500.00 per pair speakers just to prove the point.

Now, along comes the CD. All music above 20k was instantly gone. Admittedly there were other problems (16 bits of data) as well but it took forever for CD to take a foothold in the audio market simply because they didn’t sound good. Most still don’t sound nearly as good as a good record on a good turntable.

Now along comes Sony with Super Audio CD, or SACD. Finally a true digital audio product. Frequency response out to 100 KHTZ and for good or bad, multi channel. Suddenly Pink Floyd Dark Side of the Moon actually sounded better then the $25.00 a copy 10mm thick specialty records of the old days (30 years ago). One step closer to the “Holy Grail” of the audiophile, recorded music almost as good as live.

So, from my perspective, no way is an iPod with its very limited frequency response going to reproduce what most reviewers call audiophile sound. The same holds true for any compressed format.

Sorry for being long winded but I thought a brief history of what’s happening today vs. yesterday might help.
Click to expand...

Most digital recording kills the overtones and even the best CD players fall off fast above 20K, thus further killing the overtones.

You don’t need an expensive system to hear this.
 
Fryguy8 said:
nyquist algorithm.
Click to expand...


The mathmatical proof is that this guy is an idiot.

http://en.wikipedia.org/wiki/DVD-Audio

From a purely technical standpoint, the sound resolution of a DVD-Audio recording (sampled at 24-bit/96 kHz) is substantially better than a standard CD recording (which is sampled at 16-bit/44.1 kHz). However, some people report that they cannot hear a difference between DVD-Audio and CD-Audio.

http://www.soundblaster.com/dvd-audio/

WHY DOES DVD-AUDIO SOUND SO GOOD?
The 24-bit/192kHz (stereo) or 96kHz (5.1) sound (compared against 16-bit/44 kHz for Audio-CD) exhibits sharp transient response, with clear reproduction of high frequency instruments such as cymbals and rich, authentic timbre in the mid and low frequencies. Rapid high-frequency passages show excellent definition of individual notes, while stereo imaging presents a precise sound stage with great depth and solidity. Also, multi-channel capability means you can experience the acoustics of the best concert hall or put yourself in the middle of an orchestra. The natural richness of the sound makes the performance come alive, delivering a convincing surround effect that immerses you in audio!

In other words, higher pitched sounds sound better at high sample rates.
 
Who said anything about DVD-audio?

Last I checked we were talking about stereo, and in stereo no difference can be heard using 96khz or 44.1khz, or 48khz, or whatever.
 
-_- that is dvd audio... last I checked, stereo can't exactly play 96 khz...

DVD audio is the new format that can play 96 khz, and 192 khz... for the longest tiem cds have been stuck with 44 khz... and stereos have used 48 khz... but if I'm wrong tell me.
 
Newsboys2004 said:
-_- that is dvd audio... last I checked, stereo can't exactly play 96 khz...

DVD audio is the new format that can play 96 khz, and 192 khz... for the longest tiem cds have been stuck with 44 khz... and stereos have used 48 khz... but if I'm wrong tell me.
Click to expand...

Do you know why cd is "stuck" at 44.1 khz? Because it makes no difference if it's anything higher.
 
Fryguy8 said:
Who said anything about DVD-audio?

Last I checked we were talking about stereo, and in stereo no difference can be heard using 96khz or 44.1khz, or 48khz, or whatever.
Click to expand...

Tell that to Denon and Sony ES and many other companies producing higher end (yet still affordable) “Stereo” playback using 96/24 DACS strictly for stereo playback. At the same time many audiophile-recording companies are recording 2 channel CDs also using 96/24 DA converters.

DVD audio and SACD audio are all using proprietary (and ever changing) algorithms to obtain their results and doing so with great success.

One of the big issues these days is time. People simply don’t have the time to sit and listen to music for the pure enjoyment of it.

Way too many people will “settle” for convince and thus use MP-3. ;)
 
I'd like to check my system's resolution...... I got over $10K here and if this cant resolve the difference between a 192 and a 256kb mp3 or vs a lossless recording, then it's hopeless.
 
For those who argue digital vs. analog, it's pretty simple the way I see it. For systems at a similar level, analog may sound better (preference), but digital will be more accurate.
 
BillR said:
Tell that to Denon and Sony ES and many other companies producing higher end (yet still affordable) “Stereo” playback using 96/24 DACS strictly for stereo playback. At the same time many audiophile-recording companies are recording 2 channel CDs also using 96/24 DA converters.

DVD audio and SACD audio are all using proprietary (and ever changing) algorithms to obtain their results and doing so with great success.

One of the big issues these days is time. People simply don’t have the time to sit and listen to music for the pure enjoyment of it.

Way too many people will “settle” for convince and thus use MP-3. ;)
Click to expand...

So because a product advertises it has it as a feature it has to be better because of that feature?

Might as well go buy opus speaker wire, since it has better technology in it as well...

here's some helpful reading from a thread I started about this on hydrogenaudio a while back, with some reasonable evidence in some of the linked threads inside:

http://www.hydrogenaudio.org/forums/index.php?showtopic=43548
 
I'm still working on the program to deal with the audio cutting; 24-bit is a significantly harder problem than 16-bit :eek: When I get that dealt with I'll post the samples. I only have one class today so I should be able to get it done.

Fryguy, try describing what happens to a (for example) 22049 Hz tone sampled at 44100. All Nyquist says is that you have to be *at least* twice the sampling frequency to get that frequency reproduced. Not exactly, at least. Like I said, once I get this program going I should be able to come up with a sample that makes the difference obvious.

 
mage, I understand this.

At least 44.1 means that 96 works too. I never said it doesn't, just that it doesn't do anything better.
 
Fryguy8 said:
mage, I understand this.

At least 44.1 means that 96 works too. I never said it doesn't, just that it doesn't do anything better.
Click to expand...

You are confusing the audible spectrum of the human ear to the audible spectrum that can be recorded. Two very different things entirely.

No one will argue that humans can’t hear much about 17 or 18k. The fact remains however that musical instruments produce sound out past 100k in the form of harmonics and those harmonics bounce around and return to the audible spectrum as part of any performance. Anyone who produces music knows this.

When you record at 44 you instantly lose virtually all of those harmonics whereas when you record at 96 and above they are preserved and as such are included in the end product.

Please stop confusing the two issues, they are not one and the same. ;)
 
Audio posted here. There are four files: the original (51_modern_mix.flac), and three test files. The number after the dash indicates how many bits of the original audio have been kept. Notice that the files with less bits are smaller; I've also verified that the program is working properly with a hex editor and a sound editor. The program is uploaded as "chopbits.pl". I think it works in the general case, up to 32-bit sample size - change the declaration of $mask on line 18 to change how much is kept.

Suggested test procedure: use Foobar2000's ABX comparator. Select the two files to be compared (I'd suggest starting with test-16 and the original), right-click->utils->ABX two tracks->uncheck all->ok. From there, it should be pretty obvious how to go about it. If anyone can hear the difference between the original @ 24 bit depth and the chopped at 16 (or even 12!) bits, please post ABX logs. I can't tell on my equipment, but it's less than $100 counting the amp I'm not using. I have an excuse, what's yours? ;)

Edit: but I can hear 8 bits *just fine*, thank you very much :p

 
I've gone through professional ear exams and my hearing scored some of the highest they've ever seen.
 
You must log in or register to reply here.
Back
Top