On 07/01/2015 14:40, .2 dB S/N Ratio wrote:
I don't believe it's been mentioned yet (unless this is what Andy is referring to by transient), but I seem to remember part of the "trickery" that was employed by the Axon system was that the fundamental frequency was first guessed based upon only *half* the waveform -- either baseline to peak, or baseline to trough.
Something like that. If you know the waveshape then you can get an idea of frequency before one period. What happens when the pick hits the string? Before vibration is established. The transient travels away from where the pick hit the string in 2 direstions. Going towards the p/u it creates the first part of the waveform, lets' call it a +VE pulse. Going away from the p/u it eventually hits the fret and bounces back down to the p/u creating a -VE pulse. If you examine the waveform from a guitar you'll see those +VE and -VE pulse repeated, with the position of the -VE pulse relative to the +VE being varied depending on how far along the string the pick hits. ( hit at the 12th fret, and they are symmetric) Mind you, it's perhaps only that clear if the p/u is right on the bridge. ..but.. by the time that -VE pulse hits the p/u it's not much under a period. I think half the waveform is over-optimistic. If you're looking at tha waveform to judge how quick you guessed the pitch then notice that you're not taking into account the time it takes the first impulse to travel from the pick to the p/u. (which may explain any claims of being able to do it in just over half a waveform)
Of course, depending upon the waveform shape, that could possibly place you pretty far off pitch, which is one reason they would then have to correct.
Exactly. ..and if the pick position changes then the waveform changes and you get a wrong answer.
It's probably another place where there's always room for improvement in the algorithm.
Indeed, but to get faster than that isn't possible. It's like hoping that a radar can show the position of a plane before the bounced signal gets back to the radar. Energy simply has to make the journey all the way to the fret and back before there's a period to measure. andy