[ntp:questions] Using NTP to calibrate sound app

Jeroen Mostert jmostert at xs4all.nl
Mon Jan 28 00:35:34 UTC 2013

On 2013-01-27 23:43, unruh wrote:
> On 2013-01-27, David Taylor<david-taylor at blueyonder.co.uk.invalid>  wrote:
>> On 27/01/2013 19:33, unruh wrote:
>>> On 2013-01-27, no-one at no-place.org<no-one at no-place.org>  wrote:
>> []
>>>> In case you are wondering, my app is a professional piano tuning app.
>>>> The standard in this industry is that tuning devices should be
>>>> accurate to 12 parts per million.  I know that is probably overkill
>>>> for tuning pianos, but that is what the professionals expect from
>>>> their equipment.
>>> Ah. I would expect 1 cent, which is more like 500PPM.
>> 1% (10,000 ppm) is a 4.4 cycles per second beat at 440 Hz!  Completely
>> unacceptable.  You want an imperceptible beat, ideally, well under 1 Hz.
>>    Agreed that 12 ppm is overkill.
> I agree that 1% is pretty bad-- that is 1/6 of a semitone, which is
> clealy preceptible. However 1 cent, 1/100 of a semitone, is the limit of
> audibility

Not really. A cent is simply 1/100th of a semitone, no more, no less. It's true 
that few if any should be able to distinguish a note from a note that's one cent 
off when heard in isolation, but the cent is not some sort of biological limit, 
as far as I know. When played together, a difference of one cent between notes 
is certainly audible in the beating (at least on artificial waves, I have no 
idea if the same is true for physical pianos). Whether you can even physically 
tune a piano that accurately is another matter altogether. Even if you can't, 
you should still like to be able to tell that you didn't.

Even if you consider accurate detection of 1 cent to be good enough for tuning 
purposes, your measuring equipment still needs to be an order of magnitude 
better. A 0.1 cent difference is 57.8 ppm. 12 ppm is 0.02 cent, which isn't 
excessive if you're going for 0.2 cent accuracy.


More information about the questions mailing list