djlbe

Well, I think that what I'm asking here is hard to realize cause there is a lot of prob, I'm searching for an efficiant way for detecting the begining of a phase in a periodic signal. In a differed time using a lot uzi and peek. I'm not a mathematician but my goal is to create a coll with all the positions of the starting phase in a sample which contain note, especially sub note. I read there is some method for doing that (autocorrelation, etc). also the sounds I would like to analyse are sometimes noizy , sometime not stationnary. have a effeciant object for doing that could be really useful for doing some experiment like doing the same analyse on a another sound (B) and redraw the phase on the other sample (A). In fact I've already experimented manualy that kind of technique and it's really interesting to do but took a lot of time ^^. did u know if there is object which can help me for doing that ? or patch ? did someone have already tried to do that here ?

I've already done some method which are working fine when the analysed signal contain a big sub sinusoïd or when the periodicity is clear but there is a lot of particular case. the main way for got a good detection seems to give as much information as possible of what the signal is looking for (if the period is more soon than a sawtooth than a sinusoïd for exemple)

thank u in advance and sorry for my bad english, I hope u understand me well.

Well, I think that what I'm asking here is hard to realize cause there is a lot of prob, I'm searching for an efficiant way for detecting the begining of a phase in a periodic signal. In a differed time using a lot uzi and peek. I'm not a mathematician but my goal is to create a coll with all the positions of the starting phase in a sample which contain note, especially sub note. I read there is some method for doing that (autocorrelation, etc). also the sounds I would like to analyse are sometimes noizy , sometime not stationnary. have a effeciant object for doing that could be really useful for doing some experiment like doing the same analyse on a another sound (B) and redraw the phase on the other sample (A). In fact I've already experimented manualy that kind of technique and it's really interesting to do but took a lot of time ^^. did u know if there is object which can help me for doing that ? or patch ?  did someone have already tried to do that here ? 

I've already done some method which are working fine when the analysed signal contain a big sub sinusoïd or when the periodicity is clear but there is a lot of particular case. the main way for got a good detection seems to give as much information as possible of what the signal is looking for (if the period is more soon than a sawtooth than a sinusoïd for exemple) 

thank u in advance and sorry for my bad english, I hope u understand me well.


a-micro-slicer

There's the chopper patch in the gen examples folder which does something similar to what I think you're describing.

Barring that you can just use one of the many analysis externals (descriptors, analyzer, etc...) and use some onset detection, though if I understand right you don't want onsets, you want individual cycles?

There's the chopper patch in the gen examples folder which does something similar to what I think you're describing. 

Barring that you can just use one of the many analysis externals (descriptors, analyzer, etc...) and use some onset detection, though if I understand right you don't want onsets, you want individual cycles?


the chopper is near, but not really accurate, I've already a solution like this. if i take a sin signal sweeping down, I load it in a buffer and launch the 'algorythm' which will inform in a [coll] all the zero crossing, but if there is noise in that signal this will create error. in that case using a lp filter could be a part of the solution but it shift the signal so i lose accuracy.

I think what i'm searching is more complex and yes some analysis externals could be useful but i don't really know them, I think also a lot of them are for calculating in real time where some glitch are tolerates in favor of a quicker reply like in the chopper exemple.

I'm searching for a method which took his time and is really accurated

the chopper is near, but not really accurate, I've already a solution like this. if i take a sin signal sweeping down, I load it in a buffer and launch the 'algorythm' which will inform in a [coll] all the zero crossing, but if there is noise in that signal this will create error. in that case using a lp filter could be a part of the solution but it shift the signal so i lose accuracy. 

I think what i'm searching is more complex  and yes some analysis externals could be useful but i don't really know them, I think also a lot of them are for calculating in real time where some glitch are tolerates in favor of a quicker reply like in the chopper exemple.   

I'm searching for a method which took his time and is really accurated


I got decent enough results from my mpeg7 external for chopping on attack transient, the code's included so you can fiddle with it if wish, it's mxj.

It ain't perfect though, your mileage may vary etc,

It ain't perfect though, your mileage may vary etc,
http://crx091081gb.net/?tag=mpeg7


This sounds pretty interesting, though it is quite slow, making it impractical for real-time usage.

A Micro Slicer ?