Mythbusters I-Need volunteers- blind test b/w lossless/320/256/192 kbps MP3

Status
Not open for further replies.
superczar

superczar

Keymaster
Me & Saiyan were setting up my HTPC last evening when I stumbled upon the idea of making Saiyan the first scapegoat.

Saiyan failed, I couldn't spot differences either.

(Test Track was Alive - Pearl Jam (Apple lossless vs 256 kbps MP3)

Need volunteers who can :P

And before someone comes cribbing about the bad placement of speakers in my room, the speakers have been shifted to the living room in a way which can be termed as near ideal listening conditions

(good weather looms ahead for the next 3-4 months so the living room will be , well, livable :ohyeah: )

Pics of the setup will be posted tonight....

Who's willing to volunteer?

Volunteers can get their own fave CD which will subsequently be ripped to Apple lossless/wav and also transcoded to MP3s

The volunteer shall then be blindfolded and made to sit on a chair where he/she will be sitting in a near equilateral triangle with the speakers with ears at the height of the speaker :cool2:

I hear Rave laughing like a maniac

Spot the difference (between lossless and upto 256kbps), and I shall treat the volunteer to an Irish coffee and Chocolate Mousse topped with whipped cream

Fail to spot, be ready to eat humble pie....

Make a wrong selection, you eat a sickeningly humble pie.... :ohyeah:

So who's game?
 
i assume its for chennai people . so its out of question for me :)

and usually , it is with headphones that these blindfold tests are carried out , if i am not mistaken :)
 
can a blind test be carried out on an Ipod , creative earphones , trying to find difference between a 256 / apple lossless ?
 
I hope the pie is tasty :D Most people won't be able to distinguish high bitrate mp3 from lossless most of the times.

To make this test more scientific, get audio CDs and rip the tracks using eac. Then compress to flac and mp3. You can use foobar for this. Why not apple lossless? So that you get to use foobar's built-in abx comparator, which is so sexy you can't believe it :P
 
^ MORE FORMATS REQUEST.

FLAC vs. OggVorbis 192 kbps vs. 320kbps VBR MP3

btw.

I can easily spot the differences when comparing upto 128 kbps CBR MP3. Not sure after 192 kbps.

It is a well known fact that after 192 kbps, mp3s are virtually indistingishable to the untrained ear on consumer equipment.

Since you are talking of an HTPC rig and presumably have excellent knowledge of the task at hand, I wud request you to use atleast a best of 10 score i.e. 10 songs of different genres because of varying degree of compression artefacts in different types of sound. Classic music and those with more vocals suffer less from compression as mp3s supposedly cover the frequency range required for the same. Try using very deep bass (especially if you have got an exceptional subwoofer calibrated for very low crossover frequencies) or metallic tones, and you'll know where they are missing eg. Sting's Fragile, Diffuser's Karma (MI2) and Afterglow by INXS.

Too bad I don't live in Chennai. :P

Edit:

of course use foobar with asio/ks and not that pathetic wmp / itunes which are optimized for their own proprietary codecs. everything sounds definitely worse on them.
 
of course use foobar with asio/ks and not that pathetic wmp / itunes which are optimized for their own proprietary codecs. everything sounds definitely worse on them
Click to expand...

connected to a mac.. yesterday evening was done via direct passthrough to a DA converter USB card

So that you get to use foobar's built-in abx comparator, which is so sexy you can't believe it
Click to expand...

No foobar on Mac :(
 
Now that you said it, I gotta do it :P. Will post results of foo_abx with my HD650 tonight. Any volunteers in bangy? :P
 
sure man...we'll do it with foobar on the lappy if you want...

but it doesn't make a difference as Apple Core Audio (The audio kernel) is universally acknowledged to be a superior to the the kernel mixer on Windows thereby remving the need for an ASIO passthrough..

Thats the reason why most professional audio folks use OS X

To make this test more scientific, get audio CDs and rip the tracks using eac.
Click to expand...

The lossless Pearl Jam copy was a straight wav copy from Chaos' disc later converted to apple lossless....

I presume the results should be the same vs a EAC rip?
 
Heh, it will be easy to spot the difference with the good headphones.

First listen to original CD. Then to 2 test file formats. I am sure someone will be able to spot the difference almost immediately.
 
^^ Thats the whole point of the thread funky...

Is diffrentiating between Lossless/320/256kbps MP3s:

- As easy as you make it sound

- Difficult but possible

- Near impossible

- Or is the whole audiophilia debate a pseudoscience which fails under controlled tests (ok..ok, this wouldn't really be a strictly controlled test :P...but whatever :ohyeah: )
 
superczar said:
but it doesn't make a difference as Apple Core Audio (The audio kernel) is universally acknowledged to be a superior to the the kernel mixer on Windows thereby remving the need for an ASIO passthrough..

Thats the reason why most professional audio folks use OS X
Click to expand...

Wonder what said professional would say about Vista then(seems they reworked how audio works there), prolly that there's nothing to gain since it was already there in Apple they use anyway. But the result is that now the gap is closed.

Use v0 (alt preset extreme) with lame (3.97 final) for vbr, i prefer v2(alt preset standard) as v0 is just a waste of space (only in maybe 5% of cases which i'm happy to ignore).

Forget CBR since VBR tries to get you to the same sounds in less space.

I hope you know what a double blind test is, meaning the tester *also* does not know what is being served but the app you use for the purpose *will*, in short an ABX test.

The kind of music you use matters too, if its orchestral with many instruments, i think you will lose some of that with any lossy codec. Lossless is the only way here. But for most other stuff where say only a few instruments or voice is the dominating factor, lossy should do quite well as thats one of the things it uses to reduce filesize.

Another factor is how familiar the test subject is with the material, there are 2 arguments here :
- if you know the music well then you can spot the shortcomings easier, therefore its better to use as it, then it won't matter in the cases where you do not know the music, you can be rest assured that its indisitinguishable from the orginal. This feels like cheating to me somehow.

- that you *don'* know the music and have to make up your mind on the spot so to speak.

I favour the 2nd one since it helps (disk space wise) and if you can't tell the difference then to your ears that's all that matters.

Did you say Irish Coffee ?..as in with whisky in it..thats very nice :D
 
^^No using words like whisky as on tech forums it is very risky :P

but a blind test should suffice , a double blind test would not prove much here . :)
 
The foobar abx is double blind :D

foo%20abx%202.jpg


Edit: And btw, do apply replaygain before testing. Differences in level can skew the results.
 
SharekhaN said:
but a blind test should suffice , a double blind test would not prove much here . :)
Click to expand...

Removes any bias that can be felt through non-verbal expressions or intonations. Very subtle but detectable if the listener is alert.

The guy seems to be taking the trouble to set it up & all, might as well try to do it right.

Amongst the very first questions asked in forums about said tests is whether they were done dbl-blind or not and if not then it taints the results.
 
Status
Not open for further replies.
Top Bottom