Tonedeaf Test

[filarion]

Here's a fun way to waste 6 minutes: Jake Mandell (used to work for Native Instruments and releases electronic music on a bunch of labels) is working on a test to screen people for tonedeafness/short-term memory capabilities as part of his work at Harvard Medical School. The test is a series of 36 phrases repeated twice, with user input to discern whether the phrases were identical or altered the second time around. Sounds easy enough huh? Well, I scored 80,6% which is less than I thought I would, but maybe you shouldn't smoke Fatties before taking the test.

I'd be interested to know how you guys fare: http://jakemandell.com/tonedeaf/

Performance seems to be fairly unrelated to musical skill, it's more of a memory game.

[Spock]

Yea, it's more of a memory game. The first few I wasn't sure what a was going to hear, when the second pharse would start etc. From a design standpoint, they should have a few samples to get used to the format, an easy one or two, and some feedback so you know you are doing it right, then start the real test. They also shouldn't show how many are left to be done, it provides a bias, people will pay more attention at the start of the task and when it gets near the end.

I also scored an 80.6.

[J.G.]

88.9 pour moi. (Hey--that's one of the Jazz stations I listen to in Paris! ; )

Not a bad score, but not 'Grande', heheh...

[Mixerman]

NOTE: IF YOU HAVE NOT TAKEN THE TEST DO NOT READ THIS POST!!!!













































This test is completely flawed. I scored just over 80% on my first attempt, which got me curious, because there were some very suspect examples in there.

So, I went through and answered "Different" on each example, which gave me a 50% score. This means 18 of the 36 answers are "Same."

WRONG.

I went through it twice afterwards, and there are NOT 18 that are the same. Period. The problem is, this guy is playing these by hand. There are discrepancies in tuning and performance that make what the tester considers to be the same, different in the ears of the test-taker. For example. I suspect the tester is calling #16 the same, but clearly there is a tuning difference on the first note.

If good musician's are scoring 80%, then there is something wrong with the test. It's too subjective. Any discrepency might be enough to throw off an answer, and there are clearly discrepencies. I'm interested in throwing these in a DAW and comparing them one at a time.

Mixerman

[J.G.]

I'm interested in throwing thes in a DAW and comparing them one at a time.

Mixerman

And I'm interested in reading those results.
(And wahaaaaay too lazy to do it myself.)

; J

[volthause]

Wow. I scored a 75%, in the 43rd percentile. I love being average. :Roll eyes:

[Jackpine]

86.3 here but i agree with Mixerman. In the results you can relisten to any of the examples you were incorrect on. Quite a few of the ones he said were the same just flat out were not.

Mostly down to timing issues.

[J.G.]

In the results you can relisten to any of the examples you were incorrect on.

Merde, I didn't even realize that. (Too busy making tones of my own...)

; J

[E. Shaun]

I got a 75%, which left me rather baffled, because I'm SURE that there were more "different" than the same. There was one point where I was wondering if it was one of those "trick question" kind of tests, where all the right answers are the same for every question.

I would bet a large sum of money that there are many more that are "different" than the same. This isn't to say I believe I would have scored 100% (maybe on a good day), but I KNOW I have a much keener pattern recognition and musical sense than this test indicates.

[sqkychair]

87%

But, yeah, it's kind of an iffy test. More memory than anything after a while.

[Trazan]

97.2%. I often thought I heard tiny differences too, but chose "same" when I felt whatever difference there might've been was too small to be a part of the test. Might be the mind playing tricks though...would be fun to analyze that audio!

[filarion]

What I'm trying to understand is, the guy is working at a neuroimaging laboratory and is using obviously nonscientific methods for the test (not to mention the fact that a lot of people that are taking the test multiple times are going to render his statistics useless). At first I thought the idea was awesome, but yeah, maybe halfway into it I first blamed the general stonedness for thinking there might be miniscule differences between the intentionally "identical" pieces, but there are obviously timing differences and small timbre changes.. but why the hell? he could have saved bandwidth and kept the test more to the point by just using the very same sample for the identical phrases. Very wierd, maybe I'll read the forum over there to see if there's similar sentiments.

[Spock]

I was talking to the wife, PhD in experimental Psych, on a long car trip about this test. Her first questions was, so what exactly is Tone Deaf, unless you define this clearly, any test you make will be flawed.

When I think of someone that is tone deaf I think of someone that can't sing along with a simple song, they can't hear that what they are doing doesn't match. Thus, it is not memory.

[J.G.]

And then there's those folks who DO realize their sing-a-longs are off--they just can't do anything about it, (besides laugh)...

*Talk about a NIGHTMARE SCENARIO*

[crunch]

80.6

No fatties.

I was counting things like timing and phrasing, not just notes... It was a memory test as well, yes? I think Trazan had a good idea when he gave the author wiggle room for error. Because there was plenty of them. So a margin for error is definitely called for.

Interesting nevertheless, but I hope the author isn't planning on using the results as basis for a thesis or anything, because this is the very definition of "junk science"... Totally flawed, skewed results. I don't claim to be Walter Becker or anything, more like an Angus Young from Texas (hehheh... and a lot taller), but I heard differences where there were definitely differences...

[Mixerman]

97.2%. I often thought I heard tiny differences too, but chose "same" when I felt whatever difference there might've been was too small to be a part of the test. Might be the mind playing tricks though...would be fun to analyze that audio!

I held the bar considerably higher than this. "Same," to me, means the same. Not close. There was one that was a guitar playing a phrase that ended with a timbre splat. The second time the splat was slightly different. In my first time taking the test, I thought the splats were different and marked the phrase in that manner. Clearly, he intends this one to be the same.

It's a retarded test and he's just wrong in some cases. And clearly, if you can debate after the fact, whether one example is the same or different, it is a subjective test. If you want to do the test, for the "same" examples, use the exact same recording. For the different examples, use the same exact recording, and edit in a slight alteration. Do that, and musical people will consistently get 100%.

Mixerman

[otek]

86.1.

I was totally thrown off by this test, as I felt practically all the examples were different, and marked them accordingly.

[seagate]

83.3

Stupid test...

[J.G.]

I too was taking in all parameters, not only the whats but the whens--and at times, second-guessed myself wondering if infact the test was 'whens-relevant' or not...

I think that the duuuuuuuuuuuuuuude should have made the project in the DAW; not by playing the bits by hand. Me thinks that the test clearly made a premature exodus from ye olllllllllllle workshop.

But it does beat surfin' fer pern... :D

[CurtZHP]

I'm interested in throwing these in a DAW and comparing them one at a time.

Mixerman


But what summing box will you use?





Here we go again!!!
:lol:

[pounce]

at the end of the day i got 80 percent on that, i expected to do better.

[aframe9999]

75%.... right smack in the middle of average....yay

It took me two or three quesitons just to figure out what the hell was going on. and as soon as i hit my answer, the next one started immediately, which took me by surprise for a few questions...

so all in all, i could spend 6 minutes doing that, or six minutes looking at porn... hmmmm seems about like a wash to me....

[clicktrack]

I held the bar considerably higher than this. "Same," to me, means the same. Not close. There was one that was a guitar playing a phrase that ended with a timbre splat. The second time the splat was slightly different. In my first time taking the test, I thought the splats were different and marked the phrase in that manner. Clearly, he intends this one to be the same.
Mixerman

I have to agree.

I got 83.3%, but there were more than a few I would have preferred to relisten to again simply for those factors in themselves...I heard a minor variation, but I wasn't sure if it was actually something I HEARD or a variance of the performance.

I do see room for improvement. After all, musicians will be hearing and looking for things significantly more specific than the average lay person, so the test should test to that group and be stringent enough to remove all idea of subjectivity.

My two cents, though...