pitch~ or fiddle~, that is the question...


    Mar 18 2006 | 6:40 am
    MIT has some externals called pitch~ abd fiddle~. I am new to doing realtime pitch detection. I was wondering if someone could give me some tips on using these. Of pitch~ or fiddle~, which is better? If I had a prerecorded snipet of the instrument I was going to interact with, is there a way I can do some spectral analysis to help train pitch~ or fiddle~?
    Anthony

    • Mar 20 2006 | 4:05 pm
      Surely someone has used these...
    • Mar 20 2006 | 4:49 pm
      Of course. Use whichever works best for you. Some people may have had more success with fiddle~, some people will prefer pitch~, some people will say both work equally well (or poorly).
      It's not so much a question of "training" either (in the sense of training neural networks), but of tweaking parameters to best suit your input signal.
      Good luck -- P.
      -------------- http://www.bek.no/~pcastine/Litter/ ------------- Peter Castine | +--> Litter Power & Litter Bundle for Jitter |.................................................... p@castine.de | iCE: Sequencing, Recording, and Interface Building pcastine@gmx.net | for Max/MSP pcastine@bek.no | http://www.dspaudio.com/ Extremely cool 4-15@kagi.com |.................................................... | home|chez nous|wir|i nostri http://www.castine.de/
    • Mar 20 2006 | 11:19 pm
      Quote: Anthony Palomba wrote on Fri, 17 March 2006 22:40 ---------------------------------------------------- > MIT has some externals called pitch~ abd fiddle~. > I am new to doing realtime pitch detection. I was wondering > if someone could give me some tips on using these. Of pitch~ or > fiddle~, which is better?
      This sentance from Tristan's website is a tipoff:
      "Pitch tracker based on fiddle~ from Miller Puckette," that they're pretty similar.
      If I had a prerecorded snipet of the > instrument I was going to interact with, is there a way I can do some spectral analysis to help train pitch~ or fiddle~? >
      These objects aren't trainable. But you can increase the quality of your analysis by filtering out extraneous noise. I put some serious high pass filters before either object, and sometimes lowpass, too. I try to keep my microphone placement super close (to avoid any reverb) and consistent.
      It take some fiddling to get these to behave. Some things (flutes, violins) are easier than others.
      mzed
    • Mar 21 2006 | 12:22 am
      Hi,
      Has anyone tried the "WIDI Audio To MIDI" VST plug-in Max/MSP versus fiddle~ ?
      TIA for any feedback. Philippe (just a user ;-)
    • Mar 21 2006 | 12:29 am
    • Mar 21 2006 | 6:28 am
      On 3/20/06, mzed wrote:
      > It take some fiddling to get these to behave. Some things (flutes, violins) are easier than others.
      Worth keeping in mind too that you will tweak fiddle~ in a specific room/studio/hall or whatever. When you go to a different place with different acoustics it will be necessary to tweak again--probably not as much work as getting fiddle~ to respond reliably to a particular sound in the first place, but things do sound different in different places, both to you and to fiddle~. Not to mention the potential effect of using different microphones, etc.
      -- Paul
      -- ----- |(*,+,#,=)(#,=,*,+)(=,#,+,*)(+,*,=,#)| -----