Where are the blind streaming quality tests?


I've been searching around for awhile now trying to find a good article/report on how many audiophiles on a decent system (at least $1000) can hear the difference between Apple Music and TIDAL and can reliably pick which one is "better sounding" in a blind test environment. It seems most blind tests I have found show that people really can't say one service sounds better than the others. But they often are using a <$100 audio setup. Many people on here claim that they can hear a huge difference between AAC and FLAC or MQA etc. but without any evidence or test results.

So I'll challenge everyone: get someone else to switch between lossless audio and compressed without your knowledge. Also have them sometimes just do a "fake switch" where they say they changed something but really kept playing the same thing. Or better yet use the NPR test online. Can you still reliably pick the lossless audio over multiple trials with a variety of music? Post which songs you used, if you guessed right on that track, and maybe a list of gear you used for your test. Let's see if we can get some real scientific results here!
mattlathrop
@cleeds, such a study for audiophiles does not have to be comparable to drug studies and such. I think a broad range of listeners is an okay idea but I don't think it is necessary. I think any group of audiophile journalists would be where you would start. One type of music. Test no more that three variables. Start slow and simple. Maybe 10 listeners/subjects. The point being, there are no decent studies at all. _Anything_ would be an improvement.

mattlathrop, agree about photography. I think immediately about lens MTF charts. They don't tell the whole story about a lens.....but they tell and awful lot.
I have done a blind test on Tidal vs the actual cd. I picked and preferred Tidal all three times. The CDs were ripped into my Sound Science music server using DB poweramp. I used the same server to stream Tidal.
I took the NPR test and did quite well, missing one. 

Here's one from cnbc done back in 2016:
https://www.cnbc.com/2016/08/05/hifi-music-streaming-services-people-cant-tell-it-when-they-hear-it.html

Here's also some contrary thoughts on the matter of blind listening tests:

https://www.audiostream.com/content/trouble-audio-tests
&
https://www.audiostream.com/content/blind-testing-golden-ears-and-envy-oh-my

The last two are not meant to start a flame war but intend to keep an open mind as to the supposed standards set by "objectivists" who, in the end, are just as subjective as anyone.

All the best,
Nonoise
The types of tests I'm referring to would be quantification of subjective responses. And the test would have to consist of more than simply saying this is Tidal on my system or this is a ripped CD on my system. I've only been at this a few months and can do that sort of thing myself. 

The type of test I would find useful is testing several controls; several sets of cables or several streaming services,  a few DACs etc. Common criteria would be set and presented on paper with gradations on a scale of 1-10 or 1-5 for each presentation of the variable to be ranked by the subjects.  Music (one piece) would be played to test subjects multiple times for each variable in randomized order, in other words you could hear the same variable several times in a row, etc. The test would have to be double blinded meaning the test designer would not know which variables were being tested at the time of the test. That would be the only significantly tricky part.

For grins I would have each variable reviewed by each of the test subjects in open fashion (they would know what they are reviewing and would not be comparing it to anything else at the time they listened) before the double blinded test.

All of this could be completed in an afternoon with one system in one room with maybe one or two "blinded" assistants and a computer to randomize the order of variable presentation.

Assemble the data and report it. Application of statistical analysis would probably not even be needed if the number of subjects were small enough, say 8 or 10.

Such a study would have some inherent weaknesses and its scope would be very narrow. There is a chance that the data would not give a clear statement on the variables........but as stated above THAT in itself would be valuable to those looking for something other than someone's opinion about potentially expensive gear and/or services.


This is a quote from one of the articles cited above:

"Clearly any analysis that chooses to discount individual results in favor of the group result is to ignore the most basic and most important ingredient of listening to music on the hi-fi—our preference."

I agree and disagree. When it comes to knowing what you like, sure, I agree. But magazines and websites are full of reviews about specific components and such. Some of which come with astronomical prices and, quite frankly, incredible claims. If the only thing such a review can give me is one person’s preference then it has extremely limited value even if I trust the reviewer.