It's proven: 16/44.1 as good as 24/96

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Disabled Account
Joined 2006
Originally posted by I_Forgot
You can't apply scientific techniques to music.
Music is about beauty and truth and love and ennui. How can a scientific test ever hope to capture any of that? The article is worthless.
The effect of music on the mind can be studied using psychological and neuroscience techniques. Indeed, psychology of music is an active field.

But this is beside the point of this thread, as the article is not at all about music; it's about whether people can or cannot detect a difference between audio setups where a specific thing was changed--the bit depth and sampling rate of the audio. All subjective effects are taken into account by the fact that the measurement is done by humans with their ears and brains, not a machine. That invalidates any and all objections you could come up with.

So really, the only thing worthless here was your post.
 
Disabled Account
Joined 2006
Let's get the thread back on topic. There are various criticisms of this study's methodology elsewhere, but I wanted to see what people here would have to say. Criticizing methodology is one thing; criticizing the idea of blind testing, however, is not acceptable--it is a rejection of rationality, and thus in essence one is arguing against logic (and thus against logical arguments--the argument becomes self-defeating).
 
frugal-phile™
Joined 2001
Paid Member
I've actually read the paper.

The test showed that in their test, except for a few anomolies in the data (a few people who fairly reliably chose hi-rez, and a test group (women) that had a significant anti-result), the results were no different than random chance. They say in their conclusion that it does not prove that there is no difference, but that it casts doubt.

The paper was weak on detail of the actual setups.

The method which they aquired 16/44material was quite clever. They also said that it became clear fairly quickly that SACDs & DVD-A were clearly superior to CDs.

The test used ABX. There is a valid arguement that test conditions change the way one listens (ie the forced choice changes the mental state under which the listening takes place) such that test results are valid only for listening conditions unlike those conditions extant listening for musical enjoyment.

ie, it is interesting paper but proves nothing.

dave
 
Disabled Account
Joined 2006
The same testing methodology as the ITU specifies for codecs should be used for hardware testing: double-blind multi-stimulus with hidden reference. The specifics such as natural listening environment, long term listening, etc., do not contradict this.
 
I find it interesting that when ABX comes up, there are always people that claim that the 'stress' of testing, the fact that the testers are 'forced' to take a decision, somehow invalidates the results.

In any and all human activities, it has been proven time and again that putting stress on people ALWAYS IMPROVES the results, whether it is answers at a quiz or laptimes at speedskating.

Jan Didden
 
Disabled Account
Joined 2006
janneman said:
In any and all human activities, it has been proven time and again that putting stress on people ALWAYS IMPROVES the results, whether it is answers at a quiz or laptimes at speedskating.
In my case, it is the converse. I work poorly under pressure, and my work is mostly software and algorithm design. I don't know about speed, but the quality of my work is best when I can work at a moderate pace, i.e. in the flow--not too relaxed but not too pressured either. The problem is that external pressure occupies a portion of attention, and thus is a hindrance.
 
frugal-phile™
Joined 2001
Paid Member
janneman said:
I find it interesting that when ABX comes up, there are always people that claim that the 'stress' of testing, the fact that the testers are 'forced' to take a decision, somehow invalidates the results.

I understand that this line of questioning came about from some scientific experiments on detection of subliminal visual stimuli.

It is also consistent with what Levitin has to say in "This is your brain on Music" (highly recommended reading)

I know i don't do as well when stressed. My best 100 butterfly times were when i very consiously put myself into a non-stressed state before the race and my worst were when i was stressing myself to do well. I do not do my best work under pressure. I am certainly able to listen deeper into the music when relaxed.

dave
 
According to this study, SACD is indistinguishable from SACD resampled to CD quality -- the difference apparently lies in the mastering, not in the bit/sampling rate. Furthermore, they showed that 16/44.1k is perceived to have a higher noise floor (when the volume is turned up), as expected.

I read this paper some time ago, and IMO it's pretty solid. Hopefully more studies like this will be conducted in the future. More statistics = good.

Requisite blurb: http://www.bostonaudiosociety.org/explanation.htm
 
Administrator
Joined 2004
Paid Member
To take a tangent.

What about the stress of reward and failure? It can be hard to "do well" on a test if there is no consequence to the test.

If you want to test whether people can distinguish A from B, or low from high, why not offer a reward? Immediate feedback rewards like animal training. Get it right, you earn a dollar, get it wrong, you lose one. That's motivation! And most of use learn well in situations like that.

Of course, that wouldn't prove or disprove that ordinary folks can tell A from B in a casual setting, but it might show that the difference is readily detected by a trained observer.

I can't be the first person to think of this.....
 
I used to record musicians, and I could never distinguish between CD quality and higher resolutions. I really enjoyed seeing this paper, its what I would qualify as hard evidence. I wish people would do more rigorous studies like this - even if it isn't perfect, it is more reliable then hear-say.

Science results in very few absolute 'laws' or proof - its purpose is to accumulate data that people agree is reasonably accurate (and is always open to re-interpretation).
 
I work with both 14/44.1 and 24/96 and I can't hear a difference -- so I hope there isn't much of one LOL

In any case the "test stress" objection seems stretching things a bit -- you'd first have to show that stress consistently made the subjects either more inclined or less inclined to detect differences and in the context at hand. Otherwise one is countering systematic empirical observation with unfounded hypothesis -- not a flattering position in which to place one's self. Eh?
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.