Technical discussions related to Analog Devices DSPs (including Blackfin, TigerSHARC, SHARC and ADSP-21xx DSPs).
|
> -----Ursprüngliche Nachricht----- > Von: Jon Harris [SMTP:] > Gesendet am: Dienstag, 5. Oktober 2004 21:07 > An: Bernhard Holzmayer; 'ayyam perumal'; > Betreff: Re: AW: [adsp] Reg...guitar tuner design... > > > > Or, just evaluate the FFT of the signal, average it and find the > > position of highest energy. > > > > Bernhard > > The main methods out there seem to be FFT, zero-crossing, and PLL. > Here is a link that describes a zero-crossing method that is a > bit smarter to try to deal with the overtone problem: > http://www.xilinx.com/bvdocs/appnotes/xapp366.pdf > > I think the FFT method could work, though quite a large amount of > processing power could be required for good accuracy on low frequencies. Hi Jon, what about this? Do a first FFT to roughly detect the tone range. Then downsample the signal with an appropriate factor. Another FFT at that new sampling rate should now provide a result, where the accuracy is almost independent of the tone's frequency. The accuracy should be comparable to that of a FFT at original rate with the FFT length multiplied by the downsampling factor. Bernhard |