Comment by powersnail | Hacker News Reader

Comment by powersnail | Hacker News Reader

I read the paper and have doubts on the assessment of absolute pitch.

> Participants heard one synthesized piano tone per trial, for a total of 18 trials, in a semi-randomized order, the only constraint being that successive tones were separated by an interval greater than one octave. The tones were 500 ms in duration and interleaved by 3750 ms of silence, and were identical to the stimuli used during training

Only the first stimuli is a test of absolute pitch. Afterwards, relative pitch can easily help.

There is no measure (that I could see) that prevents the subjects from simply memorizing the feeling of hearing a tune because they had just gone through heavy training.

For instance, I don't have absolute pitch. But because I tune a violin daily, I can remember what A4 sound like, and deduce what other notes sound like based on that. People with absolute pitch doesn't have to do that.

In addition, testing with 12 notes is way too easy.

So, I would say "Valproate" improves the effect of ear-training, instead of saying the subjects really developed AP.

Wouldn't that only hold true if they knew the answer to the first pitch they heard? If that is held back, perfect relative pitch still would not be of help.

Yes. Relative pitch will not definitely help the subject. However, it will still affect the test result, as it is involuntary to think about the relative pitch. Therefore, the test still was not purely testing absolute pitch.

> Yes. Relative pitch will not definitely help the subject. However, it will still affect the test result, as it is involuntary to think about the relative pitch. Therefore, the test still was not purely testing absolute pitch.

How might a test be designed to take that into account?

In abstract, it's impossible to design a test to take that into account because if your subjects have great pitch memory they can recall a reference pitch heard a long time ago and then use relative pitch to fill in the rest. This is indistinguishable from absolute pitch in almost every way except maybe speed.

In practise, spacing the tests out more than a day and asking subjects not to listen for other things between tests might help, without being foolproof.

To be honest, I have no good idea. The obvious thing is to add a really long resting period before the subsequent note. Maybe engage the subject in a task of distraction while waiting.

But that is going to take a long time, so not ideal for a lab setting.

Absolute pitch is easier to recognize in life, because people who possess it can name a pitch completely out of the blue, and of pretty much anything, like a sneeze, door slam, not just piano tones.

So, there are three traits of AP --- which I gathered anecdotally, not scientifically --- that I think that might help to distinguish AP from a good RP.

1. AP works on anything with a stable, audible frequency. Memorizing the sound cannot provide that. (e.g. I cannot tell the frequency of a door slam, even though I can tune a violin without a tuner.) So, potentially, the test would use the type of sound they didn't train with.

2. AP also works really fast. As soon as they hear the sound, they know the pitch. No time for calculation. So, the subject should be asked to answer immediately.

3. People with AP doesn't seem to make mistakes at all.

All of these have led me to believe that AP is not a skill of adeptness. It's like they perceive sounds in a fundamentally different manner than I do, as if there is a separate mapping of frequency in their brain.

Again, this comment is based on my experience with people who have absolute pitch. I don't have a background in neuroscience to understand how AP works precisely, or to state definitively that the traits I described are universal among people with AP.

How are people identified as having AP?

Assuming they display AP-like abilities, the main signifiers of those who do not actually have AP are that

1. They consciously use relative pitch from a known reference tone when identifying another pitch "out of the blue", and

2. They have worked hard on acquiring relative pitch, which is something one consciously practises.

I am the same, I learned it using a piano.

The question is: If you can remember what C1 or whatever sounds like and then deduce from that, is it not just a way of more learning to be able to remember the other ones too?

Is there really a fundamental, functional difference between absolute pitch and "normal" hearing?

> is it not just a way of more learning to be able to remember the other ones too?

Yes, but that will take a long time and much dedication, and I don't know anybody interested in trying to remember many notes. There is also the risk that the more notes you try to memorize, the less reliable it gets. It's a lot more practical to just remember a few anchor notes and put the effort into training relative pitch.

For the purpose of performing music (or any practical activity), I'd say a good relatively pitch and a good memory of a few anchor notes, is as useful as absolute pitch. Functionally, AP isn't much more valuable.

But I do think, based on observation, that there is something fundamentally different with AP. It's like they hear sounds very differently, and their brain has a special mapping of frequency that I don't have.

They recognize very subtle deviations in frequency. I have had a person tell me "It's too sharp" when I was only off by 3 cents according to the tuner.

They don't make mistakes and they are fast, almost naming the pitch as soon as they hear it. It's effortless, precise, and accurate.

> They recognize very subtle deviations in frequency. I have had a person tell me "It's too sharp" when I was only off by 3 cents according to the tuner.

Ok, but THAT I have. I don't have absolute pitch, but I was always able to tune my cello more precisely than my teacher. (and other instruments too, not limited to one person)

Once a single string is tuned it is all relative from that one, so I can totally see how that would be the case. If you are a good listener then you can tell the beats until they are very close indeed.

Another trick: tune then relative to the first string tuned, not incrementally from one tuned string to the next. That way you end up much closer to the ideal.

> They don't make mistakes and they are fast, almost naming the pitch as soon as they hear it. It's effortless, precise, and accurate.

To give an example. I was playing board games with an acquaintance with AP. Someone tinked a glass or something (brief, very high pitch / edge of musical range). She named the pitch without hesitation, and noted that it was slightly sharp. (I of course couldn't confirm, but had no reason to doubt her.)

I have very good relative pitch and pitch recall. E.g. I can tune a guitar quickly by ear without relying on beat notes, and can do so without a prior absolute reference (I have memorized E, A, and Bb), though I may be ~30 cents off the reference. I can reliably identify the key of a song I heard months or years ago by comparing it to memorized references.

But I have to think about it, imagining and computing pitches and intervals in my head. And hearing other pitches in the interim throws off my recall.

To do these things instantly, with precision of cents, is magic to me. It's the difference between a savant producing the product of two large numbers instantly, and me working through the multiplication algorithm in my head.

I would suggest memorizing C4 instead. It's much easier to go up and down a couple of octaves from there than it is to start from such a low position. By the time you reach C4 from C1 you will likely be off considerably.

Thanks for the tip. I actually memorized C4, not C1, mistake on my side.

Reply on news.ycombinator.com