WAV compared to MP3 - by popular demand

JC.

I got it wrong. I did however prefer the sound I chose. I listened to both samples lots, at different volumes as well. To me I though I heard more tonal timbre in sample 2, I was wrong.

 

Right at the start I did say that people shouldn't worry about getting it wrong.
All sorts of effects and sound shaping is used in recording to get a particular sound or presentation, so it is perfectly valid to prefer the MP3 version IMO.

Different effect, but just look at how some music actually sounds better when heavily compressed. The default reaction to compression is that it has to be wrong, but if used creatively it is fine. Now the sort of effects we are hearing in a test like this are (by comparison) very subtle, but preferring them is fine in the same way that preferring a little tube distortion is fine, or a bit extra bass warmth from your phono cartridge etc.

 

To be honest Rob they both sounded pretty crappy through my laptop and Skype headset.

 

Proper headphones required then.

 

I think that's spot on as we all tailor our systems to suit our own preferences

the real casuality in this test is preconception

 

But were you really wrong .............. and if so in what way ??

I do accept that you have done a lot of very careful listening for a long time to a lot of kit.

So, in this case, you say you heard more tonal timbre in sample 2, with the implication that therefore you thought that was the best track.

Do you now think that your ears deceived you ?? or is it that really you prefer mp3's to wavs ??

JC.

 

Hi Jcbrum, I quite agree with you.

I readily admit that I cannot hear the difference between MP3 and WAV.

The sample (although smaller than I would have liked) suggests that the average audiophile can't either.

I recently met a hifi shop owner, who complained that young people came into his shop with MP3's.

How can they POSSIBLY hope to hear the difference with quality hifi when listening through an MP3? He asked, with incredulity and a raised eyebrow.


We need to educate ourselves so that we can speak with more credibility. I speak for myself here, I don't wish to alienate anyone..


I also would like now to have a bigger test - perhaps ''cross forums'' getting everybody involved.

Perhaps with 4 or 6 snippets of tracks, of only 30 seconds in length.

It would also be fun and advantageous for us to try alternative MP3 codecs (if this is the right terminology).

If it's established that we can hear a difference with MP3 (and we're a long, long, way from that) - is this the case for all methods of coding the MP3?

 

If we are going to have cross-forum participation, and i assume you mean a thread or notification on other forums explaining what is going on here with a link to the test, then we'll need the permission of the admin on those forums IMO.

I'm not comfortable putting links to ZG on other forums in this way - feels wrong.

A more complex test with more samples is fine and I can arrange that.
Perhaps others could chip in and we can agree a plan.

 

Bottleneck, the price of hard drives is so low these days that you might as well keep your library in lossless format, and it doesn't matter which.

When using a portable player you probably need to uses a lossy format like MP3, AAC(+) or OGG. The format you choose will be determined by the player you use.

And of course there is a trade off between bitate and quality - take the bitrate too low and you will hear a degradation in quality. Where the optimum balance between quality and storage space is only you can determine on your portable player using your ears! The quality of the headphones needs to be considered too.

You can listen to the same progamme in different formats and bitrates by trying an internet radio station.

Try this one

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

Top right you will see the options.

Depending on your software installed you may not be able to play all of them.

I think the free player Songbird will play them all

http://www.getsongbird.com/

On a PC if you right click on (say) and OGG stream of D-Dur it will open in a new tab or window. The URL will show at the top of the browser.

Copy that to Songbird.

In Songbird go File | Open location - and enter the URL in the box

For example the 256kb/s OGG stream is

http://www.rozhlas.cz/audio/download/ddur_maxogg.m3u

Check out the differences yourself.

A 48kb/sec stream will be small tranny quality but OK for speech, a 256 stream you will fail to tell apart from a lossless stream, as Rob's experiment showed.

 

I agree with Brian that the main music library should be stored lossless.
It might not offer any audible advantage to the listener but I think it important to preserve all of the data in order to be best paced for future technological developments.

Discard the data and its lost forever.

 

You are very welcome to use pfm.

Tony.

 

Thanks Tony.

Perhaps those posting frequently in other places can do likewise when we get a thread up and running. Computer Audiophile and Hydrogen Audio love this sort of thing so if anyone regularly posts there it would be useful.

 

Totally agree on hard disk prices.

A mate of mine says he can't listen to MP3 because of sound quality reasons - I think this test shows that as something more likely in his perception.

Robs experiment doesnt show mp3 as ''small tranny quality'' unless I am looking at a different experiment?

The one I'm seeing, its split pretty equally between people picking all options, showing no clear diffrentiator.

 

I too agree that the main archive should be lossless.

I don't think there is much mileage in testing mp3 files themselves, as it seems to me to be adequately established that no group of users, especially 'audiophiles' can tell the difference between compressed mp3 and lossless files in everyday recreational listening even with very good equipment.

(this is what I said on PF 3 years ago and got howled down )

What is much more interesting is that many people claim to be able to tell the difference.

We need to test individual listeners, to determine whether their claim is true or false

I think that would be much more fun. Im glad to see Tony is prepared to put himself forward, and we should all have a go, - the more the merrier.

JC.

 

I store lossless whenever I can.

hard disks are so cheap.

However, I stored most of my music when they were NOT cheap.. so its the huge task of re-ripping everything I can't be arsed to do - its a mamoth task, with literally thousands of cd's.

For somethign I cant hear the difference between (yet) I haven't done it.

wish I knew some bloke in china who would do it for £20..

 

Low bit rates are not good however - see for yourselves

http://www.mediafire.com/?kcde1ccclxgj5f5

Sort of 'Skynyrd Underwater' as opposed to Skynyrd Unplugged!

 

Several issues here.

Part of what confirmed Tony's identification was his knowledge of artefacts associated with mp3 - if I understand what he said correctly. So it would be fair to say that, to an extent, you can learn how to tell them apart.

Also, 256kbs is relatively high, eg higher than the iPlayer stream I listen to most of the time and why I tried them on a system that I know I can listen to mp3s on, ie it's 'good enough' for the way I listen to that system.

I listened to both tracks twice switching between the two as I did so and came to a pretty quick decision. Once you start trying to analyse and identify differences I think you are into a whole different area.

To come back to (I think) what Rob said: it's not so much about proving you can tell the difference but more - do those differences make one or other more enjoyable?

On the basis of this sample of one I am happy to continue the process I have only just started which is to rip to .wav and leave it be.

 

It's certainly a safe option Coda, but I happen to prefer alac or aiff, because of the tagging facilities which are missing from wavs.

JC.

 

Any thoughts on how to do a more comprehensive test?

To use the forum poll system we cannot make it too complicated.

We could just have two groups A & B with one group containing lossless files and the other lossy.
Each group would contain say five different tracks in order to give more variety.
Listeners can choose any of the five tracks to compare. So if you are happy that you can detect the difference using one track, you can vote, but if you need all five you can download them.

 

I don't think two groups would be much better than just two files, it's still a 50/50 chance.

What is required, particularly if you want to test the users as well as the files, is a new section containing, say, 10 threads marked A, B C, through to J.

Each thread would contain two files one of which is the mp3.

Then an eleventh thread could be available for members to post their answers such as A1 B2 C1 etc ..... It is important that the edit feature be disabled on this thread.

Following the submitted answer it could then be scored out of ten by a mod, and the percentage of correct marks could be added by the mod.

There would be no need to for the mods to publish the correct answer list immediately, just post the percentage score, and anyone who wished to could have another go to see whether they could improve their score.

Dunno whether this helps JC.