Variables to discriminate affricate [ts] and fricative [s] at word initial in spoken Japanese words

Abstract

To determine the best acoustic variables to discriminate the affricate [ts] and fricative [s], word materials produced by single-and multi-Japanese speakers were analyzed. An intensity envelope of a rise, steady, and decay part of [ts] and [s] was respectively approximated by a linear line with positive, zero, and negative slopes, and duration of the each part was measured manually and automatically. Discriminant analyses gave the smallest discriminant error with a combination of a rise part duration and sum of a steady and decay part duration in all combinations of methods for measuring duration (manual and automatic) and speakers of word materials (single and multi). This result strongly suggests that the best variables to discriminate the affricate [ts] and fricative [s] are a combination of the rise duration and the sum of steady and decay durations.

Discrimination of Japanese fricatives and affricates by production boundaries in time and spectral domains: A case study of a female native speaker

“…With respect to time domain analysis, we followed the method that proposed by Yamakawa et al [8]. By the automatic fitting method [8], polygonal lines were fitted for the intensity envelope of the rise, steady, and decay parts of /s/, /C/, / ¶/, and /tC/ in the speech materials (Fig.…”

“…By the automatic fitting method [8], polygonal lines were fitted for the intensity envelope of the rise, steady, and decay parts of /s/, /C/, / ¶/, and /tC/ in the speech materials (Fig. 1).…”

“…Despite the similarities contained within the consonants, they are distinguished through certain acoustic features (e.g., [4][5][6][7][8][9]). For example, Strevens [4] showed that in English, the spectral peak is higher in /s/ (about 3500 Hz) than in /S/ (1,600-2,500 Hz).…”

“…Yamakawa et al [8] studied discrimination of Japanese /s/ and / ¶/ in the time domain. In their study, /s/ and / ¶/ were divided into three parts (rise, steady, and decay), and the intensity envelope of the rise, steady, and decay parts was approximated by three lines with positive, zero, and negative slopes, respectively.…”

“…From these approximated lines, the durations of the rise, steady, and decay parts were calculated. From the discriminant analysis using these durations, Yamakawa et al [8] demonstrated that /s/ and / ¶/ are well separated by a combination of the rise duration and the sum of steady and decay durations.…”