lin323

I will talk about the harmonic and formant structure of [a]-falsetto voice. For this

purpose, we have to think about spectra domain. Since we are easy to see

harmonic and formant from spectra domain comparing to time domain. There

are four methods to convert time domain to spectra domain. First way is using

Helmholtz resonators. Second way is using controlled bank of filter. Third way is

using Fourier analysis, for short FFT. The last way is using Linear Predictive

coding, for short LPC. Helmholtz is using series brass sphere and listeners

depend on the loudness they hear to judge the frequency of sound source. So

Helmholtz resonators is consuming time and the result is more subjective, we

don’t consider it here. Bank of filter converts sound to electricity using a

microphone. Even bank of filter is more objective and less time consuming than

Helmholtz resonator, we don’t consider it neither. We use FFT and LPC here.

Firstly we talk about FFT. FFT is based on Fourier Analysis which is a

mathematics method through infinite series calculation to get result. FFT is a

computer program to do the calculation in few seconds. Like ipod, there is FFT

in it. FFT converts time domain to frequency domain based on fundament

frequency .