• The move to the new server is done. There are some software and database maintenance updates in process. This has us passing the hat around to help out. We appreciate any donations. Seriously, even a dollar helps. The payment page may be found here - https://www.audiokarma.org/support.html

Ego Deflation 101

Catswold

Catswold

It's psy-cat-delic!
Well . . . now I've done it. I just discovered that, for the most part, I can't tell the difference between a 320 bit rate mp3 and a 2800 bit rate flac from a 24/96 rip.

I used the ABX test from Foobar2000. Here are the dismal results:

Steely Dan Gaucho - Babylon Sisters -

File A: flac
File B: mp3 (done with Lame 3.98.4)

15:10:57 : Test started.
15:22:19 : 00/01 100.0%
15:26:02 : 00/02 100.0%
15:28:09 : 01/03 87.5%
15:31:00 : 01/04 93.8%
15:32:00 : 01/05 96.9%
15:32:59 : Test finished.

----------
Total: 1/5 (96.9%)

Dang! 97% probability that I was guessing . . . and you know what, I was.

Test 2:

Dire Straits - Brothers in Arms - Ride Across The River

File A: flac
File B: mp3

15:34:54 : Test started.
15:36:16 : 00/01 100.0%
15:36:55 : 00/02 100.0%
15:40:04 : 00/03 100.0%
15:40:47 : 00/04 100.0%
15:41:05 : Test finished.

----------
Total: 0/4 (100.0%)

Yikes! Even worse.

Test 3:

Bad Company - Bad Company - Bad Company

File A: flac
File B: mp3

15:44:33 : Test started.
15:46:35 : 01/01 50.0%
15:47:39 : 02/02 25.0%
15:49:50 : 03/03 12.5%
15:50:23 : 04/04 6.3%
15:51:32 : 04/05 18.8%
15:52:30 : 04/06 34.4%
15:53:30 : 05/07 22.7%
15:54:23 : 06/08 14.5%
15:54:42 : 07/09 9.0%
15:55:04 : 08/10 5.5%
15:55:16 : Test finished.

----------
Total: 8/10 (5.5%)

Yeah baby! I really could tell the difference between the two samples this time.

Aha! I thought. My ear is finally tuned into the differences now so I repeated the Dire Straits and Steely Dan tests but quit both after about 3 or 4 samples because I was getting the same results.

Babylon retest

A - Flac
B - mp3

15:59:13 : Test started.
15:59:58 : 01/01 50.0%
16:00:19 : 02/02 25.0%
16:00:42 : 02/03 50.0%
16:01:48 : 03/04 31.3%
16:02:43 : 03/05 50.0%
16:03:48 : 03/06 65.6%
16:05:04 : 04/07 50.0%
16:05:17 : 05/08 36.3%
16:05:34 : 05/09 50.0%
16:06:34 : 06/10 37.7%
16:07:10 : 06/11 50.0%
16:07:15 : Test finished.

----------
Total: 6/11 (50.0%)

Still 50% probablility that I was guessing. Oh well.

Amazing, the two cuts that I thought would be the most distinctive due to production values were (for me) indiscernable in their differences and the one that I thought would be the most difficult was no challenge at all.




Just call me The Tin Eared Tin Horn from now on.

I'll probably test some more of my high quality rips, but I don't hold out any hope that the results will be different. Seems 320 is good enough for most of my listening.:tears:
 
Last edited:
Register to hide this ad
A 320kb/s mp3 provides very good quality. I've often been surprised myself when I thought I was listening to HD content and it turned out to be an mp3 played off an iPod. There are differences, but they are subtle and somewhat depend on the quality of the original source material. If the original source material was good in the first place, I can usually tell right away.

There could also be a limiting factor in your playback setup- it could be that the playback system (from the soundcard on) may not be capable of revealing the tell-tale differences between good HD content and mp3's. This is certainly the case on my computer, but if I rip a CD with mixed high bit-rate mp3's, high-rez FLAC, and Apple Lossless, and play it back on my main system, I can always tell what was what.
 
You may be right, but I don't have any means of performing an ABX test outside of my computer. I am thinking of going to a better card.

I'll still hold on to my higher quality rips until I can do some more serious testing with better quality set up. Too much time invested to simply convert all the files to mp3, but you are right, 320 is pretty darn good.
 
It might have to do with the MP3 algorithm. I'd guess (guess is the operative word here) that the "denser" the music the more the algorithm has to step on it and mess up the sonics. The Steely Dan and Dire Straits cuts seem to me to be pretty open, with fairly sparse instrumentation and a dynamic, relatively uncompressed sound. The MP3 algorithm may not have to work so hard to reduce the file size. The Bad Company, on the other hand, has dense, distorted guitars, loud bass, heavy ringing chords, and a more compressed sound. More stuff for the MP3 algorithm to mess with perhaps.

:dunno: Just a guess.
 
Catswold said:
Well . . . now I've done it. I just discovered that, for the most part, I can't tell the difference between a 320 bit rate mp3 and a 2800 bit rate flac from a 24/96 rip.

I used the ABX test from Foobar2000. Here are the dismal results:

Steely Dan Gaucho - Babylon Sisters -

File A: flac
File B: mp3 (done with Lame 3.98.4)

15:10:57 : Test started.
15:22:19 : 00/01 100.0%
15:26:02 : 00/02 100.0%
15:28:09 : 01/03 87.5%
15:31:00 : 01/04 93.8%
15:32:00 : 01/05 96.9%
15:32:59 : Test finished.

----------
Total: 1/5 (96.9%)

Dang! 97% probability that I was guessing . . . and you know what, I was.

Test 2:

Dire Straits - Brothers in Arms - Ride Across The River

File A: flac
File B: mp3

15:34:54 : Test started.
15:36:16 : 00/01 100.0%
15:36:55 : 00/02 100.0%
15:40:04 : 00/03 100.0%
15:40:47 : 00/04 100.0%
15:41:05 : Test finished.

----------
Total: 0/4 (100.0%)

Yikes! Even worse.

Test 3:

Bad Company - Bad Company - Bad Company

File A: flac
File B: mp3

15:44:33 : Test started.
15:46:35 : 01/01 50.0%
15:47:39 : 02/02 25.0%
15:49:50 : 03/03 12.5%
15:50:23 : 04/04 6.3%
15:51:32 : 04/05 18.8%
15:52:30 : 04/06 34.4%
15:53:30 : 05/07 22.7%
15:54:23 : 06/08 14.5%
15:54:42 : 07/09 9.0%
15:55:04 : 08/10 5.5%
15:55:16 : Test finished.

----------
Total: 8/10 (5.5%)

Yeah baby! I really could tell the difference between the two samples this time.

Aha! I thought. My ear is finally tuned into the differences now so I repeated the Dire Straits and Steely Dan tests but quit both after about 3 or 4 samples because I was getting the same results.

Babylon retest

A - Flac
B - mp3

15:59:13 : Test started.
15:59:58 : 01/01 50.0%
16:00:19 : 02/02 25.0%
16:00:42 : 02/03 50.0%
16:01:48 : 03/04 31.3%
16:02:43 : 03/05 50.0%
16:03:48 : 03/06 65.6%
16:05:04 : 04/07 50.0%
16:05:17 : 05/08 36.3%
16:05:34 : 05/09 50.0%
16:06:34 : 06/10 37.7%
16:07:10 : 06/11 50.0%
16:07:15 : Test finished.

----------
Total: 6/11 (50.0%)

Still 50% probablility that I was guessing. Oh well.

Amazing, the two cuts that I thought would be the most distinctive due to production values were (for me) indiscernable in their differences and the one that I thought would be the most difficult was no challenge at all.




Just call me The Tin Eared Tin Horn from now on.

I'll probably test some more of my high quality rips, but I don't hold out any hope that the results will be different. Seems 320 is good enough for most of my listening.:tears:
Click to expand...

Thank you for taking a really fair stab at observing differences. Frankly, I am not too surprised at the outcome. Medium to high data rates with compressed audio are very hard to detect the compressed audio. The source material is really important in being able to eek out a difference. Personally, I find that highhat seems to show the quickst breakdown of the compression algorithm. Never low frequency or bass for me.
 
Also any continuous HF signal can give an mp3 encoder fits...a pipe organ with lots of upperwork/mixtures drawn can generate birdies at lower bitrates.
 
Jon_Logan said:
Thank you for taking a really fair stab at observing differences. Frankly, I am not too surprised at the outcome. Medium to high data rates with compressed audio are very hard to detect the compressed audio. The source material is really important in being able to eek out a difference. Personally, I find that highhat seems to show the quickst breakdown of the compression algorithm. Never low frequency or bass for me.
Click to expand...

For me it's room sound/reverb. It either sounds like a repeat or a smear/glare. I've also found that if I like the music I still like it on a cassette coming out of a boom box. And if I don't like it, the better the system the worse it gets.

But I sure as the dickens can tell the difference between a 128 or 160 mp3 and a FLAC--- then again, I can hear more of a difference lately between the hardware used to decode and play either one than the file format. A bad DAC or clock makes anything bad.

But a 320 variable? I'm not sure I could.:dunno:
 
From what I was told a few years back, the 320 bitrate is CD quality. They are converted to that so that you can assign an mp3 tag to it and edit the artist name, the album, etc.

I am not a big tech-head, so I don't know if it holds any merit, but I do know that I can't tell the difference between the 320 bitrate version, and the CD original.


Apparently I have tin-ears as well then. :tears:



-Jason
 
Tecnically it wouldn't be because any mp3 is lossy. However, to the heart of the matter, I doubt all but the most experienced and "in tune" could tell the difference.

Heck, I struggle at 192 VBR and up when using a good encoder like LAME.
 
Jon_Logan said:
Thank you for taking a really fair stab at observing differences. Frankly, I am not too surprised at the outcome. Medium to high data rates with compressed audio are very hard to detect the compressed audio. The source material is really important in being able to eek out a difference. Personally, I find that highhat seems to show the quickst breakdown of the compression algorithm. Never low frequency or bass for me.
Click to expand...

"Ay, there's the rub." I assumed that it would be the highhat, cymbals, chimes and the like, but-and I was wearing headphones-I found little to clue in on . . . very little difference (well, judging from my results, no difference :no:).

Doing a blind test like this is the only way I know of to accurately confirm whether or not you can actually discern any difference.

You may have noticed that the testing time for Bad Company was quite a bit shorter. I think the source material is definitely important.

What JonL said.:yes:

Anyway, it was one heck of an eye-opener:yikes:
 
I'm glad to hear you've been using LAME.

Unfortunately, I selected MP3Pro for a bunch of 320kbps rips back when I was using Nero... and learned that some quality CD/DVD players can't handle MP3Pro encoding. LAME seems to "own the joint" on compatibility.
 
If using a sound card and ear phones, another A/B test might be using an external DAC and speakers. My ears agree with yours on the 320kps vs Flac (and even CD) comparison, and I run it through a Grant Fidelity Dac-09 into Outlaw RR2150, then to Definitive Audio and Mirage speakers. I can't really tell the difference, and so stopped ripping to flac and now rip to 320kps mp3.

With terabyte drives at less than $100, the size doesn't really matter that much anymore. But I think we all look at the economy of ripping to the best size for the best audio quality. If there is no audible benefit beyond 320kps, why rip lossless?

I think the debate between mp3, Flac, CD, vinyl and tape benefits from testing like yours, and those who may have disparaged mp3 and other digital sources may want to take another look at its benefits. Yes, there are some bad mp3's but there's bad vinyl and tape, too. It's all about the music, and digital is just the latest incarnation of that.
 
wizargoz said:
... If there is no audible benefit beyond 320kps, why rip lossless?
Click to expand...

"If there is no audible difference...", therein lies the rub which makes it an individual decision. Some will make the decison, and I can't really fault them, on the fact that no matter how good a mp3 may sound, it is still lossy and we don't want to have any information missing.

As I've mentioned numerous times, I rip to WAV for masters/archive, then make 192 VBR copies of the WAV masters for portabililty. For me, 192 VBR is all I find necessary for the intended portable use.
 
Where is this test? I'd like to think I could tell the difference on my "downstairs" system. I'm likely wrong about that.
 
KeninDC said:
Where is this test? I'd like to think I could tell the difference on my "downstairs" system. I'm likely wrong about that.
Click to expand...

Maybe there is an easier way, but this is how I do it using Foobar 2000 player (free download).

Make a new playlist
Put the two tracks you want to compare on that playlist
Highlight/select both tracks the normal Windows way
With both tracks selected and the mouse pointer still hovering over one of them, do a right-click on the mouse to open up the menu
Go down to "Utils" in the menu, hover there, then you'll see the "ABX two tracks..." option in the submenu. Click that and follow the instructions from there.
 
wizargoz said:
If using a sound card and ear phones, another A/B test might be using an external DAC and speakers. My ears agree with yours on the 320kps vs Flac (and even CD) comparison, and I run it through a Grant Fidelity Dac-09 into Outlaw RR2150, then to Definitive Audio and Mirage speakers. I can't really tell the difference, and so stopped ripping to flac and now rip to 320kps mp3.

With terabyte drives at less than $100, the size doesn't really matter that much anymore. But I think we all look at the economy of ripping to the best size for the best audio quality. If there is no audible benefit beyond 320kps, why rip lossless?

I think the debate between mp3, Flac, CD, vinyl and tape benefits from testing like yours, and those who may have disparaged mp3 and other digital sources may want to take another look at its benefits. Yes, there are some bad mp3's but there's bad vinyl and tape, too. It's all about the music, and digital is just the latest incarnation of that.
Click to expand...

read kate wilhelms - where late the sweet birds sang if you are the one making the 1st generation mp3 i don't think there's much of an issue or difference in sound however once people start to transcode is when increased error rates come in when you clone a clone it just ain't the same
 
Lossy audio compression, when done properly, works very well. The problem is, it has been given a bad name by poor codecs and low bit rates. It's just like the difference between poor video compression (you tube), and good video compression (a good Blu-Ray).

It's important to only use lossy compression for the final step, and to never re-compress it. This is why I never buy compressed audio, but I use it every day for listening (unless I want to listen to an entire album). I want to have an uncompressed copy to start with so any compression I do is from a clean original.
 
i rip all my records to WAV and 320k mp3. with a good soundcard, its very very difficult to tell the mp3 from the original. the key is the soundcard. a $100 soundblaster isnt the answer.
 
You must log in or register to reply here.
Back
Top Bottom