MP3 bit rates debate - can YOU tell?

DRZ · 20 Apr 2006 at 19:24

I have three tests and I am going to release them one by one here. It is paramount you follow the instructions properly for it to be a fair test!

A few requests:

Please don't post what you think in this thread unless you "hide" it - your results mustnt hinder anyone else from making their own mind up.

If you have done this test for me before, you are welcome to have another go but again, please dont disclose the right answers

I cant tell you much about the test because the more you know the less accurate your brain is (a characteristic of the human brain always trying to guess "correctly"

FIRST TEST

http://www.statichiss.co.uk/CDvsMP3.zip

Download that zip file and BURN THE ISO TO CD!. Decompile the ISO all you want but at the end of the day "getting it right" isnt the aim of the game, there is no prize, just a clearer understanding of the MP3 world. If you do bother to take it to bits, congratulations, you spoiled it for yourself

If you can tell me which bit rates are used for each track, I will be very impressed indeed, but for the benefits of the test, simply tell me if the track playing is better or worse than the track before and guess at the lowest bit rate used.

Once you have got the hang of it I will introduce you to part 2, which makes things a bit clearer again

_______________________________________________________________

Well, part 1 seems to have attracted some criticism for the type of music used. This is for a reason, I will do a spoiler post at the end telling you why etc. For now, here is part 2.

Part 2

This should be a little easier to tell...

http://www.statichiss.co.uk/CDvsMP3v2.zip

Same deal as before, burn the ISO to CD and have a listen. Which bit rate is which?

DRZ · 20 Apr 2006 at 21:15

lemonkettaz said:
I can.

treble on >160kbps mp3's is hideous.... loses all quality compared to orignal CD.

Why havnt you told me which track is which then? If it is so obvious, you should be able to tell me!

DRZ · 20 Apr 2006 at 21:31

lemonkettaz said:
opps...

p.s. im not good enough for the test

What do you mean by "not good enough"? You cant tell?

Keep an eye out for the results before I put up part 2 if you truly believe what you posted earlier

DRZ · 20 Apr 2006 at 21:43

Real music? Brahms isnt real?

There are two more to come and there are reasons for the music choices used

Considering everyone is so damn quick to slate low bit rate MP3s, everyone seems a bit reluctant to jump in here, dont you think?

DRZ · 20 Apr 2006 at 22:42

Dunky said:
For sound testing, Classical is probably better since most pieces use quite a large range of frequencies.

Depends entirely on the instruments used. For example, a Clarinet produces no energy above ~22kHz, whereas Cymbals can still have audible harmonics at 103kHz (where the best measurement microphones in the world run out of linearity).

For the purposes of this test, we are assuming a frequency bandwidth of ~22kHz (the Nyquist limit of CD) so we can assume the frequency content of every instrument to be roughly similar in terms of bandwidth (although obviously harmonic content will differ significantly). Tonality therefore will only differ from that of CD and so any tonality changes may be used to pick out differing bit rates, if you feel you can notice any tonality changes

FWIW, compressing distortion is far harder than compressing a nice clean flute note, for obvious reasons. Distortion also masks quality to a certain extent and for that reason it was explicitly decided not to be represented here

DRZ · 20 Apr 2006 at 23:24

If you quote / code your own results and use the [color] tags to make them dark blue, it ought to make it so people need to highlight them to see. That way anyone reading the thread has to make an active choice to cheat.

As for me mentioning that if you decompiled the ISO you might see things etc, well, yes but the fact I am telling you this doesnt change the fact a difference is expected etc because it is fairly obvious from the rest of the post a comparison between tracks needs to be drawn. I dont feel this flaws the test, anyway.

DRZ · 21 Apr 2006 at 00:25

Whoops, I appear to have lost the text file I put the order for number 2 in. I will have a listen myself and compare the originals to the burnt tracks and figure out which one is which. Bit daft, ah well... Still, I have them for number 1:

Number 1 results said:
1 96k
2 320k
3 128k
4 160k
5 CD (wav)
6 Lossless

Fish99, impressively close but does 160k really sound better than 320k?

Dunky, I think you need to listen again

DRZ · 21 Apr 2006 at 01:20

willd58 said:
tracks downloaded from variouse share programs are clearly different, i thought this was because of the bit rate? are there other factors involved?

Finally!!!!!!!!!!!

Somebody has clocked the REAL reason my low bit rate MP3s are so comparatively good!

Yes, there are factors FAR FAR FAR more important than bit rate!

Lossy MP3 compression works by removing information that is deemed to be inaudible. Certain frequencies are implied by others and can be removed, high frequencies often go first etc. How this is done depends on the algorithm used to create the compression.

A lot of people create MP3s using terrible software and then blame it on the bit rate. In actual fact, all that is happening is the poor bit rates highlight the terrible conversion done!

I used Adobe Audition's MP3 converter (which uses the Fraunhoffer algorithm) to create these MP3s. It is the best I have come across so far, although I am fairly sure there is better out there after I came across some staggeringly good 128k MP3s.

You see, these sorts of tests show that those people who STATE that anything that isnt lossless is garbage / anything below 160k is dire etc etc shouldnt make such bold claims. Well ripped MP3s might not stand up to intense scrutiny against CD when you play them back to a hardcore audiophile but you need pretty special hearing to pick them apart IMO