Test Your Sense of Pitch | Hacker News Reader

Test Your Sense of Pitch

nidcd.nih.gov

48 pointsherendin61 comments

It's interesting. I didn't recognize all the tunes (most of them I did, of course, having gone to summer camp and Christmas services in the US).

I got 'em all. Some of them sounded to me like sentences with grammatical errors. They were all musically very adherent to Euro tradition. That is, they were all in some key or other. When the key shifted, it shifted. Nothing subtle.

I don't think this test is really a test of tonal recognition. It's more of a test of culturally determined expectations.

Looking at the data it looks like it's designed to identify a relatively small number of atypical people. I hope they're careful about following up; if this were a test for actual hearing anomalies, it might have both false negatives and false positives that would swamp the real diagnoses if the test were applied to a broad population.

And, General MIDI still sucks after all these years.

I'm confused as to how you could get anything but 26/26...

The tunes which are "wrong" aren't subtly off, they're grossly distorted - to the degree where on some of them it's hard to tell what the tune was meant to be, but you know that you're listening to some kind of godawful mess, and the answer must therefore be "No".

Well, for one, you could not know the tunes and think it's all random chromatic scale piano playing and thus all played « correctly ».

Except it's made clear from the way the test is designed that that's not the case

If you have some musical knowledge or experience it's pretty easy to tell which tunes are wrong even if you've never heard them. Assuming you have knowledge of the western musical tradition, e.g. modes, major/minor keys, etc, it's usually pretty clear where a melody should be going even if you haven't heard it (i.e. it will probably remain in the tonal centre of the piece).

If you don't recognize the songs you don't know if something is intentional or not. You can stay in-tune while not playing a song correctly.

That's interesting, do you have any links you could point me to in order to learn more about this? For instance, let's say we're in C Major.

Tune1 = C D F A G C

Tune2 = C F D A G C

How can you tell which (if any) is wrong or not and how?

I got 21/26, as of my comment the lowest score in the comment section here, but the website said it was a fine sense of pitch. Honestly I thought I'd get less than 50% before I began. I also did not recognise some of the melodies.

Some also sounded right, but cut out the last tone. Not sure if those are supposed to be counted as correct or incorrect so I did a 50/50 on those.

> Some also sounded right, but cut out the last tone. Not sure if those are supposed to be counted as correct or incorrect so I did a 50/50 on those.

Those should be counted as correct.

The webpage states the following:

"Research shows that 2 to 5 percent of the U.S. population has problems with pitch perception."

24/26 -- I guess this test is tailored to Americans quite a bit... As a European I recognise a few of the melodies, but for some of them I just have no idea what the true melody is supposed to be.

I also got 24 of 26 ... two samples were cut short omitting one or more notes that naturally would come at the end. Not being sure whether "completeness" was part of the test (probably not since the focus was "pitch") I wrongly replied "no" for those.

When designing experiments it's good to point out what to do if the tests come across unexpected situations. E.g. I've witnessed QA where the verification suggestion "no RED text should appear in the generated report" would result in PASS if no report got generated in the first place.

Same here. Even though I suspected the odds were against me, I couldn't bring myself to identify phrases with their final notes missing as "correct". There's probably a moral about pedantry here.

Same here. I have very good ear but I didn't recognize 2-3 decently sounding melodies. Of course some are so distorted I don't make an effort to recognize.

Swede, half deaf, 26/26. Didn't recognize all, but the wrong ones just sounds... wrong, to the point of making me cringe.

I'm confused, are we supposed to tell whether it matches any Major/Minor diatonic scale?

This sounds more like a melodic/scale test, not a pitch test, I was excepting it would ask something more like whether the instrument is well-tuned or whether a fretless instrument plays correctly.

For the curious, I've analyzed the first 5 songs with a program I'm developing, using the default C Major scale, here's the result : http://imgur.com/a/yx2H9 (I guess you could do it with something like Melodyne too).

EDIT: nevermind, this is just a cultural/memory test. You are supposed to know the tunes in advance and see if there is any « diff » with what you hear, melody-wise.

EDIT2: 21/26, I only knew more or less half of the songs, and knew well like a third. I Guess I have either poor culture or poor hear (or a little bit of both, we'll never know).

> http://imgur.com/a/yx2H9

That's neat. I know little about music. Would it be close to correct to say that this program finds the notes (hz, fourier) and checks if they the progression fit some predetermined pattern. eg. B4 must be followed by one of X1,X2,..., osv. ?

> EDIT: nevermind, ...

Does this mean there's songs which the program fails to classify?

It would be interesting to see the output of the program for all the songs.

It does indeed fourier analysis to find the notes. But I'm sorry to disappoint you, it does no progression analysis, so it's useless for this particular exercise (it's only handy to see the notes when you don't have an absolute pitch earing).

Well, I got 26 out of 26, but only by second-guessing the tonality that they expected, "what would classical music want me to do?".

If you had said that #3 was written by Bach, I would say it was wrong. If you had said it was by Tchaikovsky I would have said it was correct. I guessed that the test wanted me to say 'wrong' (which was the correct answer). Tchaikovsky was over a century ago and a piece of music by him would still be 'wrong'.

UNLESS I'm expected to have heard all these before.

Some of the tunes were familiar, some were not. There should be a control for 'don't know the tune' because without that the data's pretty meaningless.

Are you an American? As an American I found all of these very familiar, and they're explicitly patriotic, like yankee doodle, the national anthem, my country 'tis of thee, and so forth. Others like La Cucaracha would be very familiar to Americans.

No. Do I need to be? The landing page doesn't mention it. This may be an NIH study, but NIH studies aren't limited to American participants.

Of course I know some of them, but a handful are unknown to me.

As an American with (apparently) okay pitch this was easy to ace, but without a "Never heard this" button I don't see how the results can be meaningful. Assuming an international audience, surely the number of missed due to not knowing the song must be non-negligible - if not indeed higher than the number of misses due to poor pitch!

25/26 and I didn't recognise 1/4 of the tunes. It's harder when you don't know the tune, but it's possible to tell where the tune should be going.

Sure. Nonetheless when you don't know the tune, the test measures a different quantity than it was intended to.

It's a "standardized survey in use for over 50 years" by the National Institute on Deafness and Other Communication Disorders, so I imagine it assumes an American audience familiar with western music and its tropes.

There's a wide variety of tests and information about "tone deafness" out there. This NIH one is actually one of the better tests, though it does suffer from the problems mentioned in comments already - that it assumes familiarity with a particular set of melodies.

Tone deafness is a real condition (it's a subset of the more general condition "amusia") but is often confused with learned musical abilities, particularly being able to sing in tune. True tone deafness is something (which most studies agree) only 1-3% of the general population suffer from.

If you are truly tone deaf you will be unable to really enjoy or appreciate music. For that reason many people take these tests and declare them "too easy" but if the test is well designed it should be easy for 99% of people. This is an example of people misusing the phrase "tone deaf" to mean "not skilled in music".

If anybody would like to try some other tests for tone deafness (some good, some terrible, and disclaimer: one made by me) I wrote up a summary of the most popular ones online here: http://www.easyeartraining.com/learn/tone-deaf-tests-online/

Most common score is 26/26

Histogram: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2673726/figure/F...

Submitter here. The discussion at the site where I found this may also be interesting - http://www.metafilter.com/153003/Test-Your-Sense-of-Pitch

Like here, the most common comment seems to be that this NIH test is too easy. Maybe the title should be 'Find out if you are REALLY tone deaf'?

I seem to have the lowest score yet, at 17/26. I recognized most but not all of the tunes. There were many I wasn't sure about, and if I took it again I'd probably give different answers.

One time I tried to tune a ukulele, using a sound played from a laptop as a reference note, but I couldn't even tell whether the string pitch was higher or lower than the reference.

And yet - I rarely play, but I own an ocarina, and when I get a note wrong, it's usually quite obvious to me.

I wish they'd give me a more precise breakdown of my score. Did I get more false positives or false negatives?

Here are my answers - Y correct, N incorrect. 9/26 Y, 17/26 N, makes me guess I had too many N, which is already what I was expecting. Does anyone feel like marking me?

    (Example) - Y
     1	Y *
     2	Y
     3	Y *
     4	N
     5	Y
     6	N *
     7	N
     8	Y
     9	N
    10	Y *
    11	N
    12	N *
    13	N
    14	N
    15	Y
    16	N
    17	N
    18	Y
    19	N
    20	N
    21	N
    22	N *
    23	Y *
    24	N
    25	N *
    26	Y *

edit - starred the ones I got wrong. My prediction was wrong too. 5 false Ys, 4 false Ns.

Answers: https://news.ycombinator.com/item?id=10220145

24/26, which is amazing as when singing I can't carry a tune in a bucket. For fun, my 9 year old daughter will sing a note and get me to try and match it, and then laugh at my efforts.

Scored 26/26. Clearly the music samples are meant to be familiar to nearly all Americans. As mentioned before, the errors aren't subtle ones. One should be considerate that 2 to 5 percent of the U.S. population has problems with pitch perception (and evidence indicates this has genetic causes), and although it seems incredible that these samples would be challenging for those with the condition, let's be reflective on the things that are easy for some and hard for others.

Hmm, well I stopped at the second song from the example one. I think tests like this should not assume that the listener knows those songs. Also, if the listener knows those songs the test shouldn't assume a perfect memory of them. It tests the sense of pitch, not how well the listener remembers those songs. I mean without even knowing those songs, one only can guess by giving a no answer to those songs that contain a section with an unexpected or unfitting dissonant note. If the notes are fit to the song, dissonant or not, there is no way of noting that the song was played correctly if you haven't heard the original song multiple times in the past or just right before listening to the test ones.

I only recognized 3 of the tunes, but still got 26 perfect score. Nothing to do with familiarity with the material, just whether you can hear notes being in the correct key or not.

That's my point. You can't really know if they are correct. You just guess, if the notes are fit or not to the song. We could argue that what really tests is the listeners ability to recognize dissonant notes if all songs in the test don't have dissonant notes in them. But then its not a pitch sense test. There was one other test that I took that I see more accurate than this one: http://tonometric.com/adaptivepitch/ This one measures how little a difference of pitch can you note. It starts with a difference of 12hz, and ask you to tell if the second tone is lower or higher in frequency than the first. When you have answered a few of them correctly, it reduces the difference of pitch. I remember I was able to recognize differences of 0.75Hz before I started to fail to answer correctly.

The test you linked is much more interesting to take, thanks!

Despite not knowing almost half of the tunes, I got everything right. Though, I have to say, as someone who appreciates alternative music a lot, some of the 'wrong' samples sounded quite intriguing!

Heavens, yes. Me, too. Some of the "wrong" samples were very fun -- especially if you like diminished or augmented scales.

It's important to remember that this test only focuses on the pitch of the notes. Some of the tunes cut off before the phrase finishes, which feel weird and wrong but isn't part of the test.

Answers: http://pastebin.com/yWjhWtsm (they are embedded in the javascript function 'countChecked')

To clarify, 0 here means the answer is no (the tune was played incorrectly) and 1 means the answer is yes.

26/26, although to be fair I have played various musical instruments since a very young age. I was actually expecting the distortions to be a lot more subtle. Fun test!

25/26. I was trained as a drummer and taught myself guitar and some piano. I think the only one I missed was #13, which only sounded like it had one off-note, still sounded in key and might have been a nice harmony off the melody.

I'd love to see something which started off with these blatant distortions, and then progressed so I could train myself to hear better (assuming that's possible).

Even better if it could, based on that, what percentile I am compared with other 40yo males.

26/26. I had to guess one of the answers, as I wasn't sure whether it was a tune I hadn't heard, or whether it was just really badly played.

26/26

I agree. It is much more difficult to tell which note played, in each clip, is out of pitch if you have no frame of reference. Having heard a version of the sound clip goes a long way in this test.

This is a pitch test. IE: If I play middle C for you, then play a flat or sharp version, you should be able to identify the difference.

Asking people who have never heard middle C to identify middle C doesn't make the test relevant.

What strikes me is distorted in _which_ sense? Without a training example (i.e. one distorted and one non-distorted tune). In particular each waveform is a non-distorted version of itself.

I am not in musics (or not an American), so I am missing all cultural "obvious truths". But perceptually, there are _differences_ than can be spotted, and tastes, but not "ground truths".

23/26 here.

Score yourself: paste the following into your browser's JS console.

  a=[0,1,0,0,1,1,0,1,0,0,0,1,0,0,1,0,0,1,0,0,0,1,0,0,1,0];for(i=1;i<a.length;i++){
  with(document.querySelector("label[for='correct"+i+(a[i]?'yes':'no')+"']")){
  style.color='green';innerText+=" (correct)";}}

This test seems to be mostly about a) familiarity with the song and a little bit about b) knowing the notes that are part of a scale. Several of the "incorrect" melodies could be correct in that the notes played are pretty much all within scale, they're just not the right ones for the song.

I got 19/26 but couldn't recognise the majority of the tunes... Are you supposed to be able to tell whether it's played correctly even if you've never heard the song?

Some of them sounded "weird" but hard to say whether that's just how it sounds.

got 26/26

yes you are supposed to be able to tell even if you don't know the tunes

my daughter who is 9 took the test, she said she only recognized about 50% of the tunes, yet she scored 100%

although the test is likely skewed towards people with experience with western musical styles and scales

Got full marks despite knowing nothing about music at all. The wrong ones are just so wrong it's impossible to make a mistake. I heard PP has something to do with tonal languages, is that relevant?

I got 24/26, but there were two that I would have liked to listen to again, so those were probably the ones I got wrong. I'm not sure why you could only listen to each clip once.

Probably depends on your browser. I could listen multiple times (chrome)

* 10 were played right.

* 16 were played wrong but so, so right.

Signed, music nerd who used to end major-key choral pieces with an "anti-picardy" third during rehearsal.

25/26 for me. From the selection I've played at least most of these tunes in middle and high school or have just heard them.

Is this meant to test for disability, or also to test for "giftedness"?

Seeing many people here with a 26/26 score, I suspect the former.

26/26. Didn't recognise a few tunes, but easy enough to spot when something's wrong.

Oh and from Northern Ireland, so UK.

23 out of 26, not too bad.

26/26

Reply on news.ycombinator.com