> > male first, second and third formants for 'ah' as
> > 730, 1090 and 2440, and for the female, 850, 1220 and 2810.
>Those are approximate though... formant varry from individual to individual
>because everyone's vocal tract is slightly different.
Also, most published vowel formant frequencies are derived from speech and
refer to speaking fundamental frequency ranges.
Linear predictive coding is a common technique to estimate formant
frequencies. Monsen and Engebretson** found its accuracy was +/-60Hz for
formants 1 to 3 over a fundamental frequency range of 100-300Hz. LPC's
accuracy lessens greatly as fundamental frequency rises above 350Hz, which
is only around E4-F4. So the weather forecast for estimating formant
frequencies when singing above E4 is 'cloudy, becoming dangerous'.
Sally
** Monsen, Randall B., and A. Maynard Engebretson, 1983. The accuracy of
formant frequency measurements: A comparison of spectrographic analysis and
linear prediction, Journal of Speech and Hearing Research, 26: 89-97.
|
