WAV compared to MP3 - by popular demand

Headphones on a PC.
I voted one is the MP3

Not much in it!

 

That's 20, come on - the suspense is killing me!

 

Sorry, poor choice of words on my part.

It's MP3 so it's lossy.

Whether that loss involves dynamic compression I don't know. ie is dynamic detail part of what is lost?

 

Ok people want to know, I'm even getting PMs saying 'come on tell me'


Track 1 is WAV
Track 2 is MP3


Discuss!

 

No need to get my ears syringed today

Tony.

 

I'll get my coat!

 

You can't tell the difference, not reliably!

 

That's a good result.

It shows that the difference between the sound quality of the two files is so small that 20 different discerning listeners could not conclusively detect it as a representative group.

Two of the assessors were using the test (maybe inappropriately) as a test of their own judgement as well. This test is very inadequate for that purpose, because it's a one-off choice where you are either going to be 100% right or equally likely to be 100% wrong as we have seen.

In the spirit of fun only, I would like to see Tony L and flatpopely go head to head, on a series of at least ten choices, in the manner I described earlier, to see if either of them can tell the difference between mp3 and wav.

I suspect the reality is that neither of them can, and as many times as they are right, they will be wrong by statistically the same number.

That would show that their individual assessment is no better than the collective average, - which in this test indicates that no group of 20 people can tell the difference.

Well done Rob, it was good fun.

JC.

 

There's nothing wrong with the test at all.

What it, and the result, shows is that collectively a group of 20 discerning listeners can't tell the difference between mp3 and wav.

This result might lead to a conclusion that there is no audible difference, which is a view I would support at the moment.

However Tony L and flatpopely have both suggested that they are confident that they can hear the difference.

But this test is not designed to test the difference between listeners, it was designed to test the difference between the tracks.

Next test should be designed to test the capabilities of those who claim they can hear the difference.

JC

 

I've seen Tony take this type of test before and get it spot on.

 

To my ears this one was pretty easy, there was a lot more ambient information on the .wav, i.e. the reverb around the guitar and drum kit metalwork was much better preserved - the gaps between the instruments rather than the instruments themselves. It is this type of information that I find file compression impacts, and therefore what I listen for in these tests! Also remember that I have a studio background, so I'm used to working with digital recorders, samplers digital FX etc, and I did so back when they were pretty crap! Anyone who has used old 8 bit samplers etc knows *exactly* what digital loss sounds like, so spotting MP3s is simply a matter of listening for far smaller amounts of the same thing. FWIW I'm perfectly happy using similarly compressed files on my iPod, so I'm not trying to score any points here, I just happen to know what to listen for if put to the test!

Tony.

 

So have I, but,

a) we don't know how many times he gets it wrong because he doesn't say.

b) he's admitted that if he's not sure, he doesn't say that either

flatpopely is equally as experienced at listening and probably won't publicly admit that he is unable to tell the difference either.

In fact the forums are full of members who claim they can tell the difference

If you want to attract attention to this forum, then run the 'Have you really got Golden Ears' test to prove whether it is possible to tell the difference.

Who could resist that challenge ?

JC.

 

I never get it wrong, I can either hear it or not! It's not something I'd bother guessing.

Tony.

 

Great, I believe you ..................... and you can say I told you so ........... after the next test.

Go on admit it, it would be good fun

JC.

 

I didn't vote because obviously I knew the identity of the files.

On good headphones (ATHA700) I could detect small differences along the lines Tony describes, but as he says, you have to be looking for it to be sure. But as for me this was a sighted test, nobody should place any weight on my findings.

On speakers - I'd have voted too close to call.

On the advice of Jason (aka Figlet) at Richard's bake-off I've signed up to B&Ws Society of Sound and downloaded the 30 or so albums available, in both 16 bit Apple lossless and 24 bit FLAC. We can try something similar using 24 bit v 16 bit but I appreciate not everyone can play 24 bit files.

I also wonder what happens when we max out the MP3 to 320kbs.


A question for those hearing a difference in this test.
If you hadn't heard the wav as a reference, would you have thought the MP3 in any way poor or missing something?

 

No, it's a very good recording / good master regardless of the loss, especially in light of the dynamically squashed crap we get fed all too often these days (a whole different type of compression!), but in direct comparison I certainly prefer the .wav. I'm sure I'd prefer 24/96 or whatever over that too, though whether a laptop and 80 quid headphones would be enough to sort that one out I've no idea. Might have to fetch the HD600s through from the other room...

Tony.

 

D-Dur is an interesting station streaming the same programme in many CODECs and different bitrates.

See

http://www.rozhlas.cz/d-dur/english

There is (or used to be) a lossless OGG-FLAC stream

http://radio.cesnet.cz:8000/cro-d-dur.flac

but I could nor make it work yesterday or today. Maybe it has been discontinued. If it is still on air a comparison could bee made between the FLAC and the lossy CODECs.

OGG-FLACs are not easy to play - Foobar on the PC will handle it and VLC player on most platforms.

18 months ago I had my system set up so I could flick on my Beresford Caiman DAC from the 256 OGG on the Squeezebox and the OGG-FLAC on an XP PC running Foobar.

I could not reliably tell the difference. If there was a difference it was in ambience - the space round the music.

But there is another test that could be done, if the OGG-FLAC stream is still there and for those with the kit to try it.

 

On 24/96 (or 192) I not so convinced, certainly if a good up-sampling dac is used on the lower res file.

I've downloaded a bucket-load of 24 bit files so we can do a similar test as a future project.

 

In all honesty, I thought this one was easy? It was more "obvious" than some of the other ones I've done (and got wrong). Not quite sure why that should be...

 

as far as MP3.. no not poor. I enjoy the free limited version of Spotify as well.
In some ways similar when you compare directly against the original.

I wasn't moved with the society of sound music offerings I heard, at least not enough to pay for them.