Optimizing FastTrack's performance to obtain formant values in fricative noise

Although the FastTrack plugin (as well as formant measurement in general) was originally intended for vowel analysis, we apply it to our sample of laryngeal [h] and epilaryngeal [ħ ʜ] fricatives in Mehweb Dargwa (East Caucasian). This is possible because in these places of articulation (almost) the entire supraglottal tract is involved, and the spectrum is similar in structure to that of vowels. We attempt to determine the optimal FastTrack settings to obtain reliable formant measurements with possibly little manual correction by varying formant ceiling ranges and curvature. Apart from the fricative segments alone, we also tested FastTrack’s performance on longer signal stretches: fricative with entire surrounding vowels, fricative with 35 ms adjacent vowel intervals to the left and to the right.