Can We Really Tell Lossless From MP3?

EddieSpinola writes "Everyone knows that lossless codecs like FLAC produce better sounding music than lossy codecs like MP3. Well that's the theory anyway. The reality is that most of us can't tell the difference between MP3 and FLAC. In this quick and dirty test, a worrying preponderance of subjects rated the MP3 encodes higher than the FLAC files. Very interesting, if slightly disturbing reading!" Visiting with adblock and flashblock is highly recommended, lest you be blinded. The article is spread over 6 pages and there is no print version.

Comments

Re:It does depend on the recording

[BobNET]

They save me thousands of dollars in high end equipment

There's your problem. If you had spent more on your audio system you'd hear the differences.

Even if there weren't any...

Re:The hiss is where it hides

[Pieroxy]

Ah ah ah!!!
From the story:

The reality is that most of us can't tell the difference between MP3 and FLAC. In this quick and dirty test, a worrying preponderance of subjects rated the MP3 encodes higher than the FLAC files.

Rarely if ever you can find such a contradiction right in the summary. If most of us can't tell the difference, how come subjects rated the two encodes differently?

You Don't Know Nothin

[DynaSoar]

The present study suffers from that methodological malady known in scientific circles as being "fucked". Please bear with me as I explain this technical term.

The question posed in the text is 'can we tell the difference'. One assumes from this that the answer is yes or no. Testing this question would require playing two versions and asking whether they're the same (can't tell the difference) or different (can etc.).

But that's not what gets asked. The subjects get asked to tell which version sounds better. The question assumes they can tell the difference. Even if they can't tell the difference they are forced by the design to choose one over the other as if they can.

Since they are forced to say which sounds better even if they can't tell the difference (something impossible to determine from this design) then they are simply guessing or picking one arbitrarily, and there is no way to determine if or when this occurred. Thus, the results are not only unable to answer the original question, they are unable to answer anything because the data do not even necessarily represent answers.

The design is so fatally flawed that there is nothing that can be pulled out of it. It's complete garbage.

As an aside, I'm not familiar with the musical pieces used, but I'm betting they're fairly new. For years now recordings have been increasingly compressed by the engineers. Most popular works produced in this decade are already so compressed that you can't tell much difference between the original and a recording of it having been compressed yet again, no matter by what method.

To tell the difference between compressed versions one should start with an uncompressed source. And for a person to be able to hear a difference in two versions, they should already be familiar with the original in uncompressed form so they can try to say whether one sounds more like the original than the other (the alternative being both sound worse or both sound like it). If they have no clue what it's supposed to sound like, any attempt to say which sounds better is badly broken due to having no reference with which to compare them.

No attempt was made to determine whether the subjects even had normal hearing. And I don't mean just asked (though that should be done) but tested. People can have frequency drop outs that they're unaware of and that would affect the results.

There are so many problems with the study that it is completely useless. The problems were of the authors' making. Thus, they did not know what they were doing. This is what we mean by "fucked".

I want to know who determined that 'trusted' was a good name for the magazine/blog/honey wagon in which the article appears. I wouldn't trust them to test light bulbs to see if they're burnt out.