formant [-nn -on -pn -fn -in -wn -Wn -tn -ln -hn -Ps] <speech_file>If the sampling frequency of the speech file is greater than 10kHz (ornHz if the -f option is specified), "formant" downsamples the file to10kHz (or nHz) and saves the results in a .ds file.  The signal isthen highpassed to remove DC and rumble and the result is saved in a.hp file.  LPC poles are computed and saved in a .pole file.  Thefundamental frequency, VUV and rms energy are estimated and saved in a.f0 file.  Finally, a dynamic programming algorithm estimates the bestglobal mapping of LPC poles to four formants and places the results ina .fb file.  All of the files produced by "formant" have a SIG speechheader and all, except the .pole files, may be viewed as waveforms using the"waves" program.  The .fb and .pole files may be overlaid on spectrogramsusing waves.Other options are:  -on	n specifies the LPC order to use in the analysis [12]  -nn	n specifies the number of formants to attempt tracking [4]  -pn   n specifies the preemphasis constant (0.0 to 1.0) [0.7]  -in	n specifies the frame interval in seconds [0.01]  -wn	n specifies the window duration in sec. [0.049]  -Wn   n specifies weighting type: 0=>rectangular; 1=>Hamming; 2->cos**4 [2]  -tn   n = 0 implies autocorrelation LPC	n = 1 means use Atal's stabilized covariance LPC (-w and -p ignored)		(w forced to 0.025; p to exp(-1800 * pi * T))	n = 2 means use standard covariance LPC [0]	 (no weighting is applied if n > 0)  -F	only compute F0; no .pole or .fb files are produced.  -l	n specifies the low frequency limit for F0 tracking [50Hz]  -h	n specifies the high frequency limit for F0 tracking [500Hz]  -P	s is a F0 analysis param. file (see /usr/local/lib/formant/misc/param)A SIG header may be added to the input speech file with	addhead [-fn] <input_file>  where n is the actual sampling frequency in Hz.  If not specified, n        is assumed to be 10kHz.  <input_file>.d is the output file	with the header prepended.