Fundamental Frequency Characteristics of Modal and Vocal Fry Registers

This study examined the distribution of fundamental frequencies in the connected speech of 14 healthy young adults. Acoustic analysis of fundamental frequency was performed on previously collected speech samples of a phonetically balanced reading passage. For the first three sentences of the reading passage, fundamental frequency contours were extracted using PRAAT, a speech analysis software package. The accuracy of these contours were visually verified and manually corrected when needed. The distribution of the fundamental frequency histories for each sample were then analyzed using Gaussian Mixture Model analyses in MATLAB. For most speakers, four statistical modes were identified in the data based on model optimization. The lowest statistical mode was located in a frequency region that was consistent with the vocal fry register. This lowest statistical mode made up only around 5 percent of all glottal cycles, on average across both male and female participants. The results are discussed in relation to normal voice production, voice disorders, and vocal performance