Slashdot Mirror

It's not "elitist fuckery", it's a combination of good equipment, an original source recording of high quality, and perhaps better-than-average hearing.

Every actual study I've seen shows that except on certain "hard" classes of recordings it's wishful thinking or other psychology--with 99% of normal music (be it well-recorded classical, a capella voice, rock, whatever) the 256kbps VBR LAME settings Amazon uses haven't been distinguishable to anyone in any scientific study I'm aware of.

I've had this argument many times, and there are some recordings that reveal obvious flaws in even 320 kbps CBR mp3 to my ears with my headphones and amplifier

If you can do it regularly, then you should pretty easily be able to make thousands of dollars for a few hour's work by claiming any of the numerous prizes offered for people who can succesfully distinguish 256kbps VBR MP3 of a wide array of popular music from uncompressed originals in a double-blind A/B test.

Now, if you're just talking a handful of special-case horrible-for-mp3 recordings (like, say, the well-known Eig "LAME killer" sample) then that's another story; some people can certainly pick that one out at 320 kbps VBR.

You can easily start off using a free ABX program like PC ABX (Windows) or LinABX, or a more expensive hardware solution. Just see if you can actually ABX them at home and if so, you should be good to go claim some cash.

It's pretty fun to see what you can actually distinguish, too. I have a nice setup with a good pair of Grados; out of my library of 4000+ songs there are maybe 3-4 I've found so far that I can pick out a 192kbps Ogg from FLAC. Before doing such testing, I was "sure" I could pick out the difference between the original and the 320kbps encodings I usually make. Most people who come over to my place can't distinguish 96 kbps from uncompressed except on a handful of nasty test samples, but if you actually learn what the common artifacts are it gets a bit easier for some people.

Friend of mine got an iPod a couple of years back, downloaded about 100 CDs onto it. I was listening to it idly, and remarked that the sound quality wasn't up to much (I was probably picking up compression artefacts, as he plays no music that I have ever listened to). He discovered that he'd been using the standard compression, and that switching to recordings made with the higher bit rate was a substantial improvement.

So, last week I had a mess about with Lame and found that 256 was OK for classical music, but 384 was noticeably better. I didn't A/B with the source material, as I knew I wasn't going to keep a CD quality file on my laptop. Incidentally I'd expect classical piano music would be a great way of revealing problems with compression algorithms.

http://www.pcabx.com/ for a very handy comparator tool that allows you to run statistically valid, double blind audio tests.

The only reason why Sonique's decoder sounds 'better' is because they're breaking spec and likely doing filtering behind the user's back. I'm not so sure if the 1.9x series of Sonique players before their eventual failure to realize 2.0 were any different, but I doubt it.

As it says in the article linked above, they weren't do anything they shouldn't in regards to filtering. All things being equal, everyone's MP3 decoder following proper specs, should sound exactly the same -- the way it should be. It's not the decoder's job to filter anything, that's the job of an actual DSP chain you can tweak how you wish.

So the next time someone is making claims about a decoder sounding 'better', provided that they're both using the same output at the same volume levels... One can realize that this is either placebo or there's something going on where nothing should be. Best way to resolve this, is to use whatever available wav logging options to capture samples from each of the disputed players and then ABX them.

Check this out: http://www.pcabx.com/

True double blind testing instead of single blind testing. Of course, the tinfoil-hat audiophiles insist that the DB apparatus "masks" or "conceals" the huge differences between components that are clearly audible during single-blind or sighted testing. Richard Clark is on the right track, and I'm sure no one has collected the $10K.

It does take a bit of preperation, but the results are legit. Not really suited for large organized polls, but fine to see your personal tastes and to understand exactly what a double-blind test is and how it works and why it is the only valid way to scientifically test.

RMS distortion and spectra are not meaningful ways to measure physchoacoustical codecs. Codecs typically take advantage of temporal and frequency masking effects which mean that the in many cases large components of the sound can be discarded because they are masked by other sounds in the recording.

If you really want to know if you can hear the difference after re-ripping, or if the iTunes AACs are really CD quality, the best way is to do a blind ABX test (you can find software to do this here.

- Try MP3 encoded with LAME's "--preset standard". It's far better than whatever you're using. If that's still not good enough, try "--preset insane" (which is a highly-tuned version of 320kbps).

- Download an ABX tool and actually do a proper blind test. Most people who complain about MP3 artifacts can't in fact ABX them.

Learn why you shouldn't use spectral analysis to determine lossy codecs' quality.

The most respected technique is double-blind testing using an ABX tool such as PC ABX, WinABX or ABC/HR.

More info on conducting blind tests can be found at the PC ABX site.

Learn why you shouldn't use spectral analysis to determine lossy codecs' quality.

The most respected technique is double-blind testing using an ABX tool such as PC ABX, WinABX or ABC/HR.

More info on conducting blind tests can be found at the PC ABX site.

As a long time follower of the different audio encoding technologies, I have to agree with you on your point that the quality of the different formats seems to be quite subjective, but personally, I'm one of the many (few?) whose ears are overwhelmingly in favor of Ogg.

I can't STAND WMA at all... it has a high end ringing screech at pretty much any bitrate, while at the same time, some people swear by WMA. WMA also boosts the volume of the encoded material to give the listener the impression that it's better quality, which is bad form, IMO.

MP3 is pretty impressive nowadays, with all of the work that has gone into LAME... Even 128kbit VBR is passable. But as I said, to my ears at least, they all bow before Ogg. That same passable quality you get at 128kbit with LAME you get at 96kbit with Ogg. And the artifacts are also much less offensive to my ears, but again, that is a matter of opinion.

Anyway, I hear too much completely uninformed Ogg bashing, and I wish everyone would do some objective testing of their own. Go read up on blind ABX testing, and do some yourself. If Ogg isn't the one you think sounds best, that's fine, but just don't say it's crap without giving it a shot!

While I don't doubt that you believe you can hear a difference, it is highly unlikely that what you perceive is something that you can attribute to real physical phenomina. There have yet to be any published studies which show a difference between 96kHz/24 bit and a downsampled 44.1kHz/16 bit double-blind, level-matched presentation of the same source material.

If you'd like to conduct some experiments on yourself to see how good your hearing really is, I highly recommend going to PCABX.com and going through some of the materials there.

As an aside, even comparing the "CD-Compatible" layer on many SACDs isn't necessarily a fair comparison. Some record companies modify the equalization settings between the different pressings and even down-sampled versions are perceived as different.

Heck, even a 0.5 dB change in level is small enough that most people don't perceive it as a level difference, but large enough that they perceive it as being "more airy" or "detailed" or whatever.

While high-resolution formats might have advantages on the production/mixing side, and certainly on the DRM side, they have yet to be demonstrated that they show any real advantage to the consumers.

I am Karma Man, hear me Whore.

An honest double-blind listening test is extremely difficult to arrange, and there is no evidence whatsoever on such on the site.

This is how the test was conducted.

The test required access to a Windows machine (probably Win95 and up, didn't try with Win3.1) with a sound card. Users were required to download the ABC/HR "practice" Zip file, which includes the ABC/HR program, the Ogg Vorbis 1.0 command-line encoder and decoder, a LAME command-line encoder/decoder (I forget which version), a FLAC command-line decoder program, and a .flac sample file (the instrumental introduction to The Eagles' "New Kid in Town").

After unzipping this, the user had to run a batch file (encdec_foobar.bat) which un-FLACced the sample file, then encoded it with Ogg Vorbis and LAME, then decoded both of the resulting files back to .wav.

Then the user executed the ABC/HR program, which is a Win32 GUI application. After loading the sample into the application (pull-down menu and file selector dialog), the interface became a row of double-slider pairs. Below each slider was a "Play" button. Below each slider pair was a "Play Ref" button. Below that was a "Stop" button. There was a pair of sliders for each decoded sample -- so for the practice run, there were two pairs of sliders: one for file #1, and one for file #2. The user did not know which file was Ogg Vorbis, and which was LAME MP3.

The user then listened to the Reference file by clicking any of the "Play Ref" buttons. After hearing the Reference, the user could then click any of the normal "Play" buttons. The first task was to determine, for each pair of sliders, which one was the original and which one was the encoded file. Having determined that, the user used the slider (which went from 1.0 to 5.0 in increments of 0.1) to "score" the sample on the subjective quality of the result. There were also text labels on the slider: 4.0 was "perceptible but not annoying", 3.0 was "slightly annoying", 2.0 was "annoying" and 1.0 was "very annoying".

Finally, there was an ABX button, which launched a different window. In the ABX window, the user could select "Original", "Sample 1", or "Sample 2" for the "A" and "B" samples. Normal ABX testing proceeded from that point. (If you don't know what ABX is, go to pcabx.com.) I found that the ABX window sometimes helped me to focus on a specific sample so that I could find its flaws; armed with that knowledge, I was able to make a determination of which of the two sliders, right or left, was the encoded version.

Once a slider was pulled down from the default 5.0 position, another button became active above that slider. Clicking on it opened a new window with a text box, into which comments could be typed. When the user was finished with the test, the slider positions, the comments, and the ABX results (if any) were written to a plain text file (DOS CR/LF format), which was to be mailed to the test administrator. (Though, of course, you weren't supposed to mail the practice results.)

Now, that was just the practice session, which was a prerequisite for participation in the actual test. For the actual test, the process was similar, but differed in a few details.

The actual test samples included copyrighted, patented codecs for which there are no freely distributable decoders. Therefore, the WMA, AAC and MP3Pro samples were distributed as FLAC files, and decoded by the batch file. Since MP3 did not participate in the listening test, the LAME encoder was not used during the actual test. The Vorbis encoder, of course, was used twice: first with -q 0, and then with -b 64 --managed.

With 5 encodings per audio sample in the actual test, there were 5 pairs of active sliders instead of only 2 pairs. But otherwise, the actual test was exactly like the practice session.

(Personal note: I did 10 of the 12 samples, skipping the two classical ones. Out of 50 encoded versions of the 10 samples, there was only one case where I couldn't tell right from left -- "The Source", encoded with MP3Pro.)

When I did that, I also tried some of his tests to see what kind of listener I was. I can barely hear background noise at all, but was able to pick out the original from 256K mp3 encoding 11 out of 14 times, which is proof to ABXers. This was Arny's 'articulation' test at 'probably impossible' level... I am a mastering engineer who writes DSP software and I was using my studio reference system. He tells me there was one other person once who was able to ace that test, and he's considering toning down the language and not calling it 'impossible' anymore if people can get it 11 out of 14 and so on- but to most people it is impossible. Again, that's 256K mp3, and not Xing either.

I'm sure I could tell Ogg Vorbis too, but you have to know what to listen for when the bit rate gets luxurious. There was no real tonal change to listen for, it's just that the 256K file was recognizably characterless, sort of like 'pod people' of audio. Hardly surprising as this is just what I work to avoid in full-resolution CD audio- that too can be rather bland and characterless if you're not careful!

(My own audio work is at www.airwindows.com/dithering/, GPLed, recently added some mid/side stereo features and a GUI Knob class that worked out quite well)

Try a vbr file created with LAME's --r3mix option, or one of the --dm options. There really is no reason to use CBR.

And, if you make claims you can hear the difference, have you tried telling the difference with ABX? This is the only good way to try it. Then post the results again. And use mp3's you encoded yourself, not something you downloaded.

If you really want to do reliable tests on wav files, then visit PCABX to get the PCABX program and to read more about the testing methodology. The program takes in two wav files, and then chooses one of the two randomly and lets the user decide which of the two is the one chosen randomly. Basically, once this done a good number of times (say, 20) the program can then tell whether the user can actually tell the difference between the two files.

Also, a wonderful website dedicated to the task of creating archival quality encoded audio (which is indistinguishable from the original) is r3mix. Lame even has an optimized parameter that comes from the work at the site, --r3mix! This VBR parameter gives incredible quality at a fairly low bitrate. Check out too a listening test carried out at r3mix that showed the blind preferences of 42 users over a month of time.

If you're running Windows, you can get ABX from http://www.pcabx.com/. On UNIX systems, the LAME source code comes with an ABX program (in the misc/ directory, I think).

Here is an example of a test that took place using a slightly different testing methodology, more akin to MUSHRA (which is used to evaluate lots of encoders at the same time): http://www.ff123.net/128tests.html.

Or you can use an ABX program like this that will randomize the order of play for you and will display statistics of how successful you've been in identifying the encoded from the original.
The program above is only for windows though a quick search in freshmeat gave me LinABX which is an ABX program for linux, didn't have the time to check it out though.

The title says it all, really. While Ogg definitely still has artifacts, they're much softer and less noticable than mp3 artifacts. The first time I sent an Ogg to a friend, I didn't really go into how much better it was, but his first words were "WOW, this sounds _alot_ better than mp3s". Add to that the fact that Monty's still working hard on tuning Ogg... remember, it's still only on RC2. And if you're into low bitrates, 64kbit is pretty damn good considering it's only 64kbit.

But in the end, you should probably just go by your own judgement. Try a double-blind test and see which one you prefer. Info on how to do that can be found over on the PCABX page. I think the software is for Windows only, but check it out anyways... there's lots of good info about the double-blind test in general.

I've noticed most of the posts here are saying how awful Vorbis sounds... I've been using it for quite a while now, and have done pretty extensive testing myself as well as reading what alot of other people have had to say. I don't have "Golden Ears" or $10k worth of stereo equipment, just a decent pair of headphones, but it's ALWAYS been my opinion that ogg sounds better than mp3. I sent one to a friend once, and his first reaction was, "WOW! This is ALOT better than mp3!". And that was with the beta 4 encoder. Even those crazy guys over on the r3mix.net forums have lots of praise for Ogg Vorbis.

Like the topic says, I haven't been able to get to RC2 yet, thanks to it being slashdotted, but I seriously doubt RC2 sounds worse than beta 4, and while encode times _are_ slower than mp3, they're nowhere near as slow as some people are saying. (I get about 3x speed on my Duron 850 with b4). Clicks and pops are likely a cause of a bad rip from the CD, not the encoder.

I've been using nothing but Ogg for my CDs for a while now, and have encouraged many friends to do the same. People really need to give Ogg a fair, unbiased try before they go saying it sucks, because it's most definitely at the very least, better than mp3 at the same bitrate. Check out PCABX for info on how to do a good double-blind listening test.

Congrats to Monty and the rest of the Ogg Vorbis team. Keep up the good work.

While it's only tangentially related, www.pcabx.com provides tools and audio training material