Old 8th September 2007, 04:59   #1
muzicman82
Junior Member
 
Join Date: Aug 2002
Location: Woodstown, NJ
Posts: 19
Send a message via ICQ to muzicman82 Send a message via AIM to muzicman82 Send a message via Yahoo to muzicman82
FLAC vs. MP3 Comparison

I did a short one. Read it here - http://www.muzicman82.net/2007/09/07...p3-comparison/
muzicman82 is offline   Reply With Quote
Old 9th September 2007, 06:25   #2
Rocker
Hiding in plain sight (mod)
 
Join Date: Jun 2000
Location: Melbourne, Australia
Posts: 9,910
its like comparing apples to oranges.
Rocker is offline   Reply With Quote
Old 9th September 2007, 10:30   #3
gaekwad2
Foorum King
 
gaekwad2's Avatar
 
Join Date: Jul 2003
Location: bar2000
Posts: 11,457
Listening to the difference won't tell you anything about whether that difference is actually audible when comparing the lossless and lossy version.

Welcome to the wonderful world of perceptual coding. ABX (not sighted A/B) is your friend, without it you'll most likely get lost and end up wasting time and effort on pointless tests, such as...
gaekwad2 is offline   Reply With Quote
Old 9th September 2007, 16:08   #4
muzicman82
Junior Member
 
Join Date: Aug 2002
Location: Woodstown, NJ
Posts: 19
Send a message via ICQ to muzicman82 Send a message via AIM to muzicman82 Send a message via Yahoo to muzicman82
Most people probably can't tell the difference. I can A/B test all day and tell the difference. I'm a full time studio and live audio engineer. I can recognize 2.5 kHz vs. 3.1 kHz. Now that disk space is cheaper and iPods hold double the capacity that they did last week, it worth the change.
muzicman82 is offline   Reply With Quote
Old 9th September 2007, 16:22   #5
gaekwad2
Foorum King
 
gaekwad2's Avatar
 
Join Date: Jul 2003
Location: bar2000
Posts: 11,457
You can believe you hear a difference as much as you want, if your tests aren't blind it may always be placebo.

But the main problem is that you use your mix-paste experiment that tells nothing about mp3's sound quality as an argument against lossy compression.
gaekwad2 is offline   Reply With Quote
Old 9th September 2007, 19:54   #6
muzicman82
Junior Member
 
Join Date: Aug 2002
Location: Woodstown, NJ
Posts: 19
Send a message via ICQ to muzicman82 Send a message via AIM to muzicman82 Send a message via Yahoo to muzicman82
Quote:
Originally posted by gaekwad2
But the main problem is that you use your mix-paste experiment that tells nothing about mp3's sound quality as an argument against lossy compression.
How doesn't it tell you anything? The resulting audio shows exactly how the two formats compare.. the resulting file is everything that is pulled out of 16-bit/44.1 kHz audio to make these low quality MP3s. It tells you a lot actually. It tells you what you've been missing.

Just because you don't notice the difference listening to 192 kbps MP3s, doesn't mean they aren't different. They're very different to a trained ear.

Besides, what point are you trying to make? That I can't tell the difference? That FLAC isn't better than MP3s? You can give me any blind test you want and I'll pick out the lossless every time. It's not that hard. MP3s roll off the high end starting at about 12kHz. You might not be able to tell, but I can tell when my highs are there... especially 14kHz-18kHz.

My post wasn't to show that one is better than the other. We already know that FLAC is LOSSLESS. The post shows HOW they are different in terms of frequency/gain differences.
muzicman82 is offline   Reply With Quote
Old 9th September 2007, 20:59   #7
Benski
Ben Allison
Former Winamp Developer
 
Benski's Avatar
 
Join Date: Jan 2005
Location: Brooklyn, NY
Posts: 1,057
Post a log using http://www.kikeg.arrakis.es/winabx/ and then we'll talk
Benski is offline   Reply With Quote
Old 9th September 2007, 21:17   #8
gaekwad2
Foorum King
 
gaekwad2's Avatar
 
Join Date: Jul 2003
Location: bar2000
Posts: 11,457
Quote:
Originally posted by muzicman82
How doesn't it tell you anything? The resulting audio shows exactly how the two formats compare.. the resulting file is everything that is pulled out of 16-bit/44.1 kHz audio to make these low quality MP3s. It tells you a lot actually. It tells you what you've been missing.
No, it shows the difference between two signals, nothing more
Quote:
Originally posted by muzicman82
Just because you don't notice the difference listening to 192 kbps MP3s, doesn't mean they aren't different. They're very different to a trained ear.
To someone trained to hear mp3 artifacts. (Hint: they don't include bass or treble roll-off.)
Quote:
Originally posted by muzicman82
Besides, what point are you trying to make? That I can't tell the difference? That FLAC isn't better than MP3s?
No, that your test methodology is flawed.
Quote:
Originally posted by muzicman82
You can give me any blind test you want and I'll pick out the lossless every time. It's not that hard. MP3s roll off the high end starting at about 12kHz.
Funny, your own graph shows a sharp lowpass at 16khz*, underneath that the volume looks fairly similar, doesn't it?

*and if you'd inform yourself about the mp3 format and its design limitations you might even find out why it's there
Quote:
Originally posted by muzicman82
You might not be able to tell, but I can tell when my highs are there... especially 14kHz-18kHz.

My post wasn't to show that one is better than the other. We already know that FLAC is LOSSLESS. The post shows HOW they are different in terms of frequency/gain differences.
It shows the 16khz lowpass.
(btw, if you don't want that use Lame --alt-preset standard)

Oh, and if you really are an audio engineer, shouldn't you be more concerned with other kinds of compression that are far more destructive than even low bitrate mp3?
gaekwad2 is offline   Reply With Quote
Old 9th September 2007, 23:05   #9
muzicman82
Junior Member
 
Join Date: Aug 2002
Location: Woodstown, NJ
Posts: 19
Send a message via ICQ to muzicman82 Send a message via AIM to muzicman82 Send a message via Yahoo to muzicman82
gaekwad2 -

So far, all you've managed to do is tell me my test is flawed, but have failed to say WHY or how it is flawed. It's a simple test to see what is ripped out of the MP3 versus the WAV file. Nothing more. I'm not comparing human interpretation.

If all you like to do is go around to forums and bash people's posts without logical explanation, that's fine. I'll remember to ignore you.

Mixing one channel of a stereo file with the inverse of another yields the difference of the two channels. Doing these test while one channel is a 192 kbps MP3 and the other lossess will provide an accurate representation of the difference between the two filetypes. It might not provide a representation of what you hear as the difference, but it's a calculated difference. It's the same technique that's use to cut vocals from a stereo track. Why does it work? Things that are the same in both channels get ditched. Things that are different stay. Lead vocals are almost always panned center in the stereo field, except for BGV's.

I'm not comparing the two formats to tell me which is better. I already know that. I went into this asking myself, "What is the MP3 taking away?" And this gives me the answer. But please, you obviously know more than I about this stuff, so tell me how this isn't showing me what is subtracted from a WAV to result in an MP3.

I know and completely understand there are psychoacoustics involved with MP3 theory. I'm not out to test that. I can listen to MP3's all day and tell you they sound great, for being 1/10th the size of a WAV.

I know there are high and low pass MP3 settings, but I am testing against what I've been using.
muzicman82 is offline   Reply With Quote
Old 10th September 2007, 00:32   #10
gaekwad2
Foorum King
 
gaekwad2's Avatar
 
Join Date: Jul 2003
Location: bar2000
Posts: 11,457
First of all, mp3 doesn't simply subtract anything* (also listening to the difference doesn't tell you whether what you're hearing was removed or rather added during processing).

But even if it did, it's not designed to produce pretty graphs or sound good mix-pasted.
The only thing that matters is whether the mp3 as is sounds like the original or if it doesn't how obvious/annoying the artifacts are.

*apart from the low pass (high pass is never used in practice (at least by Lame))
gaekwad2 is offline   Reply With Quote
Old 10th September 2007, 02:41   #11
muzicman82
Junior Member
 
Join Date: Aug 2002
Location: Woodstown, NJ
Posts: 19
Send a message via ICQ to muzicman82 Send a message via AIM to muzicman82 Send a message via Yahoo to muzicman82
Ok, folks. You might read the post again . I did it again suspecting that the samples weren't perfectly aligned the first time, and I was right. What's interesting is that the first time I did this, both files were the same number of samples, just offset by a few. I'm surprised no one else mentioned/questioned/caught that. The result is different, but that doesn't change whether this is a valid method for comparing two formats mathematically.

I wasn't trying to make it sound good mix-paste. The fact is that there are differences. Whether they are added or subtracted doesn't matter. The end result are the differences.. whether added or subtracted.

Again, I'll say that I'm not trying to replace an ABX test with calculations. The calculations are just an interesting listen to the DIFFERENCE between two formats. It's also interesting that by doing these with different bitrates of MP3s and AAC, as the bitrate increases, the result of the inverse mix is less audio. The goal, ultimately is flat line.

LAME didn't exist when I made some of these MP3s originally. I have been using Easy CD-DA Extractor for some time, but at some point, I was listening to MP3s and noticed something was up. I then checked settings and the default for 192 kbps has HP/LP set to automatic, which for all practical purposes means ON.
muzicman82 is offline   Reply With Quote
Reply
Go Back   Winamp & Shoutcast Forums > Winamp > Winamp Discussion

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump