How many people can hear the difference between...

I was mid pack, although I admit that I screwed up the Killers because I though the track hadn't changed. Didn't get the hang of the test until the third flip. Plus, I found the Killers track to just be a massive wall of sound that gives little to distinguish. Like Rob, it seemed like the Dixie Chicks had more "clues" to use for distinguishing the two. For me it was in ambience and soundstage, neither of which I found in the Killers. This was also the case when I did a similar test years ago. Some tracks were more affected by the compression.

I wonder if its significant that I improved continually?

You got 52% correct
There is a 100% likelihood of getting this or a more extreme score by chance
Track Correct p-value
The Killers 20% (p >= 0.020)
James Blake 40% (p >= 0.020)
Daft Punk 60% (p >= 0.020)
The Eagles 60% (p >= 0.020)
Dixie Chicks 80% (p >= 0.020)
 
Along with the quality of hearing, the gear being listened on and the environment, there is what they are used to hearing, I worked at an FM station and I could switch the studio monitor to the source (going to the xmtter) and what was broadcast. What people got was compressed to the point that one would need some high-end gear to try to restore it to the real sound. If people spend all their time listening to radio, they probably would not recognize half the music played from the source.

The same may happen to those hearing it streamed, as MP3s, on poor earbuds, etc.

Opinions are like (name the body part); everyone's got one.
 
Along with the quality of hearing, the gear being listened on and the environment, there is what they are used to hearing, I worked at an FM station and I could switch the studio monitor to the source (going to the xmtter) and what was broadcast. What people got was compressed to the point that one would need some high-end gear to try to restore it to the real sound. If people spend all their time listening to radio, they probably would not recognize half the music played from the source.

The same may happen to those hearing it streamed, as MP3s, on poor earbuds, etc.

That is an interesting point. In the past it was often not possible (equipment and/or media prohibitely expensive) to get anything even close to "the source". These days it's very possible, but still might not be practical enough for most people.
It's also interesting to think that "the same" could happen multiple times, for the lack of a better wording. Some people are young enough to never have heard anything from an analog tape so they certainly won't compare an MP3 to that experience, and some people have listened to tons of tapes but have no concept of what vinyl sounds like.
 

Interesting! Not a study, but something closer to the original point (yes indeed). It's good to be reminded of how the math for this sort of thing works.

I realized now more directly that something further to consider is of course the relevance of only being able to tell the difference. Like others have already touched on. More specifically, in this test I knew there was a difference to start with. I feel if I hadn't known (if there was an added possibility of A and B being the same), it would have approached being impossible. That would almost certainly have added true errors... aka something not just a wrong guess, I guess.

Also something to consider is of course whether or not the difference is large enough to "matter", which is almost a completely different topic. I found that in some of those tracks the only difference was exactly that they sounded different from each other. One didn't sound "bad" and the other "good", generally. In at least two of them I could only tell them apart because the high treble sounded different, which could even be a result of an optional low-pass filter on the compressed track. Different encoder settings, that difference might disappear.

My results? Well ok, for science!

You
probably can't
hear the difference between the lossy and lossless samples (p >= 0.10)
You got 68% correct
There is a 11% likelihood of getting this or a more extreme score by chance
Track
Correct
p-value
The Killers 60% (p >= 0.020)
James Blake 60% (p >= 0.020)
Daft Punk 40% (p >= 0.020)
The Eagles 100% (p >= 0.020)
Dixie Chicks 80% (p >= 0.020)
Is sharing scores your thing?
 
Something to note here - obviously if there is a quality difference when it comes to the audio being compressed, bouncing out in higher khz/bitrate is sort of like comparing the resolution of a picture - You will notice a quality difference of a 720 and 1080 video.. but at 4k or 5k resolution the difference starts to become really small, at some point the difference will be so small that you as a human can't really point it out.

i have noticed in the digital domain of recording, that some plugins and effects work better when i tried bouncing out in 192khz, sometimes not at all.

and i mean, what you are referencing the content through also is really important.


I don't think the average joe and jill actually cares about the quality as long as it hits industry standard or whatever radio station/spotify they are used to listening to.

And i also think audio engineers such as myself are biased when listening to music as we tend to spend so much time listening to how the music was mixed etc that we have a harder time just sitting back and enjoying the ride!

If it sounds good, it's most likely good ;)
 
Back
Top