[bigshot]

Please everybody note the low-fi hardware that I used. It seems to be possible to hear differences even through intermediate quality portable earphones and iPods.

 
I've found that when compressed audio artifacts, it doesn't require a ten thousand dollar system to hear it. Artifacting is plainly audible in midrange systems too.

 

[bigshot]

  @bigshot, 10codecs are already a lot, but it would give a better idea to have a few doubles to reduce the probability of guessing. maybe just one or 2 as not to make the test too heavy and tiresome.

 
I thought it was more important to have longer and more varied bits of music, rather than more versions. 20 minutes of samples seemed to me to be about the limit of an average person's endurance. I wanted to start below my own threshold and go to the top to see if my own results were typical. Actually, I think most people can't tell 192 either.

 

[mironathetin]

  @bigshot, 10codecs are already a lot, but it would give a better idea to have a few doubles to reduce the probability of guessing. maybe just one or 2 as not to make the test too heavy and tiresome.

 
Agreed, don't torture.
It takes already a lot of effort to do it with 10.
Since you have to build a sequence and not only pick right or wrong with every guess, this is already a good test. There are more than 3.5 million ways to build a sequence out of 10 items. If the result has a correct structure, than it should be statistically valid.
 
But bigshot is right in his analysis: it seems to be possible to identify the lowest bitrate, but then it gets harder (who is surprised?).

 

[Terja]

Interesting discussion ///
 
For the test I would agree with FFBookman that a standard needs to be established first in order to determine encoding differences. In other words, the original uncompressed source should serve as the initial parameter, or benchmark sample. After that it may be included in the randomized test samples to see if it can be correctly identified among other samples and whether that makes a difference.

 

[bigshot]

Why would you need to know the lossless sample if lossy is obviously inferior and sounds like music being played in a cardboard box as he says?
 
If I provided a lossless sample, the first thing golden ears cheaters are going to do is compare waveforms against the known lossless track. Not knowing keeps them honest because they have to use their ears to determine sound quality, not peep for differences.

 

[dprimary]

  Why would you need to know the lossless sample if lossy is obviously inferior and sounds like music being played in a cardboard box as he says?
 
If I provided a lossless sample, the first thing golden ears cheaters are going to do is compare waveforms against the known lossless track. Not knowing keeps them honest because they have to use their ears to determine sound quality, not peep for differences.

If you convert the high bit rate lossy files back to wav files is it even going to visible?

 

[Ruben123]

The funny thing is everyone is able to tell MP3 and 16 and 24 bit WAV apart. Until it must be done blind. Then you need some hours to listen to 24 bit and then going back is obvious. But IF they could do that in a blind test environment, science tries to fool then. So no go.
 
Same for all audiophile stuff it seems.

 

[bigshot]

  If you convert the high bit rate lossy files back to wav files is it even going to visible?

 
Yes, because the wave form has been altered in the lossy files. It would be a simple matter to take the known lossless and just compare it to the ten samples and find the one it matches. (I had someone try a similar trick on me once, but I caught him at it.)
 
The point of this test is listening... everyone knows that lossy files are *different* than lossless ones, but the test is to determine if the difference is audible.

 

[Terja]

P.S> People who are really interested in finding out the veracity of their own hearing wouldn't bother with such childish antics ....

 

[bigshot]

Absolutely. Unfortunately some people are more interested in proving they are right than actually being right.

 

[castleofargh]

  Interesting discussion ///
 
For the test I would agree with FFBookman that a standard needs to be established first in order to determine encoding differences. In other words, the original uncompressed source should serve as the initial parameter, or benchmark sample. After that it may be included in the randomized test samples to see if it can be correctly identified among other samples and whether that makes a difference.

 
but it's not really the point. he's been posting everywhere anytime the subject comes up, that mp3 sounds really bad. on the pono troll topic, he's been saying that an old walkman playing a k7 tape sounded better than a mp3 in a modern DAP. so if only for the tape hiss, if mp3 is worst, audibility should really never be a problem right?
then he goes on saying that no test is right to test that.

see the irony? so obvious you don't need a test, so tricky a test can't be used. pretty funny. and of course as always with people against blind testing, he has no better alternative to offer but sighted evaluation. and we all know what that's worth and how accurate it is.

 
 
 
I didn't take bigshot's test because I believe that an ABX is a more accurate way to go at it, and have done it countless times. but between nothing and his test, I take his test any day.
and I also don't believe blind tests to be perfect, but I just don't pretend like I can know better in a sighted evaluation, because that's a joke.

 

[FFBookman]

   
but it's not really the point. he's been posting everywhere anytime the subject comes up, that mp3 sounds really bad. on the pono troll topic, he's been saying that an old walkman playing a k7 tape sounded better than a mp3 in a modern DAP. so if only for the tape hiss, if mp3 is worst, audibility should really never be a problem right?
then he goes on saying that no test is right to test that.

see the irony? so obvious you don't need a test, so tricky a test can't be used. pretty funny. and of course as always with people against blind testing, he has no better alternative to offer but sighted evaluation. and we all know what that's worth and how accurate it is.

 
 
 
I didn't take bigshot's test because I believe that an ABX is a more accurate way to go at it, and have done it countless times. but between nothing and his test, I take his test any day.
and I also don't believe blind tests to be perfect, but I just don't pretend like I can know better in a sighted evaluation, because that's a joke.

 
Does Frauhauffer pay you people to defend MP3 sound quality online by spreading FUD about "lossless"?  I thought that program ended in the 90's.
 
Why are you convinced that 320k = 4000k, but only in digital audio?  
 
Why do you denigrate your hearing capabilities so easily to think that your ear-brain only can process what equals 320k of audio data at a time?
 
Why would you support a test that tricks people into thinking that 320k = 4000k?
 
You've already said in previous threads that any artist or producer that purchases samples at higher than 16/44 quality is a fool, which makes you a fool in my book.
 
 
No tests needed, just add up your points --- No one can hear more than 320k of lossy data?  Yet everyone seems to hear it. So a test is devised to show that most of the time people aren't sure what's going on. Confusing people doesn't prove anything and your emotional crutch in believing a compressed, lossy, audio format designed for dial-up modems sounds the same as professional audio is fascinating, it keeps me posting. I can't help the intervention in your backwards thinking.

 

[RRod]

 
No tests needed, just add up your points --- No one can hear more than 320k of lossy data?  Yet everyone seems to hear it. So a test is devised to show that most of the time people aren't sure what's going on. Confusing people doesn't prove anything and your emotional crutch in believing a compressed, lossy, audio format designed for dial-up modems sounds the same as professional audio is fascinating, it keeps me posting. I can't help the intervention in your backwards thinking.

 
People *claim* to hear it, then suddenly don't when the test is blind. That is, if they take time to get off the troll-logic train and actually do the test.

 

[bigshot]

Prove it! You have the file!

 

[FFBookman]

  Absolutely. Unfortunately some people are more interested in proving they are right than actually being right.

Wow that is what you are doing.  Searching and searching for a way to convince us all that 320k = 4000k in digital audio.
 
 
Or that humans only have the hearing capability to process around 320k of stereo audio data.  That is false.
 
Or that humans can't hear digital compression artifacts and the filtering and masking used in MP3.  Also false.
 
Or that everyone in the world listens to horribly produced music from the last 10 years on $5 headphones.  False again.
  
Or that nobody cares anyway, because this is confusing and they just want convenience.  More false every day.