Re: Robust method of fundamental frequency estimation.

["James W. Beauchamp" <jwbeauch@xxxxxxxxxxxxxxxxxxxxxx>]

Roisin,

Automatic F0 detection of some instruments can be nontrivial. Low
pitched tones tend to be more difficult because they can have 
their low harmonics suppressed. Short notes or tones with time-
varying pitch can give problems, as do tones with inharmonic 
partials, like the piano.

You might try a harmonic pattern matching method such as the
two-way mismatch method. It is written up in

R. C. Maher and J. W. Beauchamp (1994). "Fundamental frequency estimation 
of musical signals using a Two-Way Mismatch procedure", J. Acoust. Soc. 
Am., Vol. 95, No.4, pp.2254-2263.

and it is implemented in the SMS and SNDAN musical sound analysis/
synthesis packages.

For piano tones, I know that Anssi Klapuri, who has done much work on
music pitch detection, has specially treated that case. Google will
find him quickly.

Jim Beauchamp
Univ. of Illinois at Urbana-Champaign