Re: [MLUG - DISCUSSION] Audio file quality tests

[Mike Miller <EMAIL:PROTECTED>]

On Tue, 8 Jan 2008, Joseph Ondrus wrote:

> If I rip a disc to 128, 192, and FLAC, I just have a pair of 
> Sennheiser's attached to a Carver pre-amp and if I switch from 128, 192, 
> to FLAC, 128 to 192 is very easily distinguished

That is a credible claim, but how are you doing the test to verify that 
you aren't just confirming what you already knew?  In other words, don't 
you need to be given the two files in some random order with no knowledge 
of which is which, then pick which is the 128 to prove that you can tell 
the difference?


> the jump from 192 to FLAC is a bit more difficult, but you can pick out 
> parts of songs that sound different, you may not be able to describe how 
> they are different very well but you can tell them apart.

But can you pick out which is the better file based only on how they sound 
with no other information about which is which?


> I don't want to imply that the someone could choose what bit-rate the 
> song is by listening to the song, but when comparing the two together, 
> you can tell one is higher than the other.  Like you said cymbals might 
> really matter at higher bitrates, but I'm sure if we sat down we could 
> pick out many other instruments (or noises) that the human ear (or does 
> this begin to fall under brain processing ;) )can differentiate in a 
> sense of quality at different bit-rates.

I know that people can detect bitrate differences if the differences are 
large enough and one of the bitrates is low enough.  The important 
question for me is how low of a bitrate is nearly always indistinguishable 
from full CD audio (~1400 kbps) -- that is for nearly all listeners on 
nearly all tracks on high-quality equipment.  I'd also want to know what a 
really good listener can do.  I think above 320 kbps, no one can tell the 
difference anymore.

So the question is, how can I found out the truth about this?  The only 
way to get there is with a fairly carefully-controlled study, but has 
anyone done such a study yet?  I've seen a lot of claims on the web, but 
no good studies.  I've seen pages where people post files, label them, and 
then ask people to vote on which sounded best, but that is a flawed 
design.

We could try to do a study if we can't find one that has already been 
done.

Mike

_______________________________________________
discussion mailing list
EMAIL:PROTECTED
http://mlug.missouri.edu/mailman/listinfo/discussion