What techniques exist for periodicity detection of arbitrary signals? My idea is to be able to split the audio into periodic chunks. --------------------------------------- Posted through http://www.DSPRelated.com
Periodicity detection of arbitrary signals
Started by ●January 22, 2016
Reply by ●January 22, 20162016-01-22
On 22.01.2016 10:42, mavavilj wrote:> What techniques exist for periodicity detection of arbitrary signals? > > My idea is to be able to split the audio into periodic chunks. > --------------------------------------- > Posted through http://www.DSPRelated.com >I would suggest to start with searching the web for "cepstrum"...
Reply by ●January 22, 20162016-01-22
>On 22.01.2016 10:42, mavavilj wrote: >> What techniques exist for periodicity detection of arbitrary signals? >> >> My idea is to be able to split the audio into periodic chunks. >> --------------------------------------- >> Posted through http://www.DSPRelated.com >> >I would suggest to start with searching the web for "cepstrum"...You may also try calculation of the spectral correlation function. -Doug --------------------------------------- Posted through http://www.DSPRelated.com
Reply by ●January 22, 20162016-01-22
On Friday, January 22, 2016 at 4:42:10 AM UTC-5, mavavilj wrote:> What techniques exist for periodicity detection of arbitrary signals? > > My idea is to be able to split the audio into periodic chunks. > --------------------------------------- > Posted through http://www.DSPRelated.comgoogle for "pitch" and "periodicity histogram" don't forget to write a check :-)
Reply by ●January 22, 20162016-01-22
On Friday, January 22, 2016 at 7:43:46 AM UTC-5, Andre Lodwig wrote:> On 22.01.2016 10:42, mavavilj wrote: > > What techniques exist for periodicity detection of arbitrary signals? > > > > My idea is to be able to split the audio into periodic chunks. > > --------------------------------------- > > Posted through http://www.DSPRelated.com > > > I would suggest to start with searching the web for "cepstrum"...cepstrum does not detect period of the simplest periodic signal -a sine wave
Reply by ●January 22, 20162016-01-22
On Fri, 22 Jan 2016 03:42:04 -0600, mavavilj wrote:> What techniques exist for periodicity detection of arbitrary signals? > > My idea is to be able to split the audio into periodic chunks. > --------------------------------------- > Posted through http://www.DSPRelated.comYou might look up 'autocorrelation' too.
Reply by ●January 22, 20162016-01-22
On Fri, 22 Jan 2016 03:42:04 -0600, mavavilj wrote:> What techniques exist for periodicity detection of arbitrary signals?Second-order linear prediction will result in two useful parameters: The center frequency of the two-pole response is the spectral peak, and therefore the inverse of the predominant period. The derivative of the frequency response equals zero at this peak. The Q tells you how periodic the signal is. Also, this calculation requires little in the way of resources. Steve
Reply by ●January 23, 20162016-01-23
>On Fri, 22 Jan 2016 03:42:04 -0600, mavavilj wrote: > >> What techniques exist for periodicity detection of arbitrary signals? > >Second-order linear prediction will result in two useful >parameters: > >The center frequency of the two-pole response is the spectral peak, >and therefore the inverse of the predominant period. The derivative >of the frequency response equals zero at this peak. > >The Q tells you how periodic the signal is. > >Also, this calculation requires little in the way of resources. > >SteveCould you clarify what other terms are used for "second-order linear prediction". I'm trying to find Python or C++ libraries and very little turns out with "second-order linear prediction". --------------------------------------- Posted through http://www.DSPRelated.com
Reply by ●January 23, 20162016-01-23
mavavilj <106909@DSPRelated> responds to my post,>>Second-order linear prediction will result in two useful >>parameters: >> >>The center frequency of the two-pole response is the spectral peak, >>and therefore the inverse of the predominant period. The derivative >>of the frequency response equals zero at this peak. >> >>The Q tells you how periodic the signal is. >> >>Also, this calculation requires little in the way of resources.>Could you clarify what other terms are used for "second-order linear >prediction". I'm trying to find Python or C++ libraries and very little >turns out with "second-order linear prediction".Googling on: linear prediction coding C++ libary give you a ton of hits, and the top hits are relevant to what you want. Hope this helps. Steve
Reply by ●January 23, 20162016-01-23
>mavavilj <106909@DSPRelated> responds to my post, > >>>Second-order linear prediction will result in two useful >>>parameters: >>> >>>The center frequency of the two-pole response is the spectral peak, >>>and therefore the inverse of the predominant period. The derivative >>>of the frequency response equals zero at this peak. >>> >>>The Q tells you how periodic the signal is. >>> >>>Also, this calculation requires little in the way of resources. > >>Could you clarify what other terms are used for "second-order linear >>prediction". I'm trying to find Python or C++ libraries and very little >>turns out with "second-order linear prediction". > >Googling on: > > linear prediction coding C++ libary > >give you a ton of hits, and the top hits are relevant to what you want. > >Hope this helps. > >SteveI don't find the parameters you mention though. Care to elaborate? --------------------------------------- Posted through http://www.DSPRelated.com
