Compression test.

Re: Compression test.

John Chiara said:
Saw that today...haven't gotten to it yet.
Click to expand...

My music theory teacher and I took a stab or two at it today, together, and on our own. We could reliably avoid the 128k, but the 320k was sometimes close enough we couldn't tell well enough. They were all mixed, mastered and encoded professionally, so I can understand the good results after compression.

Funnily enough, the one I consistently got was the concerto, but couldn't get Speed of Sound. The theory teacher is a guitarist, so he listened to that to identify Speed of Sound properly. I'm a pianist, so that explains the concerto.

Overall, a fun listening test. Was easier to tell which was which on my phone speakers than with my in ears, actually. Besides, it's not like any of them sounded bad or anything. We enjoyed listening no matter which file was which.
 
Re: Compression test.

On computer speakers, I only got three out of six. But the thing is, the three I got right, I KNEW which one was the good one. The three that I missed, I repeated the tracks multiple times going back and forth, and still had to guess.

I like to think I'd have done better with a system more resolving than my computer desktop. The statisticians would say that, with a score of 3/6, I did no better than chance, but that doesn't tell the whole story.
 
Re: Compression test.

I have been unable to do better than random chance between the 320k and uncompressed samples, with headphones or monitors. Headphones are a little easier since it's easier to hear panning errors in the codec. Some of the tracks I find easier than others, for instance I can pick the uncompressed version of Coldplay 100% of the time. Very difficult!
 
Re: Compression test.

I started doing this the other day, on a pair of PreSonus s6's in the near field. I got the first 4 correct which was pretty exciting, however I stopped there. The amount of time I spent on each example was pretty disproportionate to the amount of time I was willing to waste.
 
Re: Compression test.

Dave Barnett said:
I like to think I'd have done better with a system more resolving than my computer desktop. The statisticians would say that, with a score of 3/6, I did no better than chance, but that doesn't tell the whole story.
Click to expand...

Since each compression ratio is a different test we should say that you got perfect (with a sample size of 1) on 3 tests.
Although it is interesting that on the harder ones you picked the wrong one all 3 times? maybe there's something about a touch of compression that appeals to you?

Jason
 
Re: Compression test.

If your compression tests low you can squirt some oil in the spark plug hole. If the pressure improves it means you have bad piston rings. If the compression stays low you have leaky valve seats.

JR
 
Re: Compression test.

Jason Lavoie said:
Since each compression ratio is a different test we should say that you got perfect (with a sample size of 1) on 3 tests.
Although it is interesting that on the harder ones you picked the wrong one all 3 times? maybe there's something about a touch of compression that appeals to you?

Jason
Click to expand...

It was more that on those, I thought all the samples were grim.
 
Re: Compression test.

Those 128k samples sounded really good! I've never had my 128k's sound that good... My 'crossover' is around 224k and upwards where I start to have trouble telling them apart from lossless.

Anyone knows how this was setup and encoded?
 
Re: Compression test.

Dave Barnett said:
If, by "cleaning it up" you mean "throwing away information".
Click to expand...

While this is perhaps too esoteric, speech and music are very redundant with lots of repetition as notes persist for small fractions of a second. Even speech is very redundant with sounds repeating. Large amounts of this redundant information can be discarded without hurting speech intelligibility, while for music, sound quality perception does not like major alternations to the sound envelope.

Compression schemes intelligently discard less important (redundant) information.

That said even the best data compression is never better than the full bandwidth signal, just less bad than cutting the same bandwidth arbitrarily.

JR
 
Top Bottom