Frequency-Importance and Transfer Functions for Recorded CID W-22 Word Lists

Abstract

Frequency-importance and transfer functions for the Technisonic Studios’ recordings of the CID W-22 word test are reported. These functions may be used to calculate Articulation Index (Al) values or to predict scores on the W-22 test. The functions were derived from the word recognition scores of 8 normal-hearing listeners who were tested under 308 conditions of filtering and masking. The importance function for the W-22 test has a broader frequency range and a different shape than the importance function used in the current ANSI standard on the Articulation Index (ANSI, 1969). The transfer function is similar in slope to to the ANSI transfer function for 256 PB-words, but is shifted to the right of that function by 0.05 Al.

The intelligibility of words, sentences, and continuous discourse using the articulation index

“…However, recent research has documented that there are different importance functions for different 20 speech material (Black, 1959, Studebaker et al. 1987, Studebaker and Sherbecoe. 1991).…”

“…Since the Al is additive, the LP and HP frequencies that had a WRS of 32% intersect their Fitting the curve Overal[, the halving and complementing procedures were repeated until 26 points were derived so that a relative transfer function for the words, sentences and CD passages could be defined. The curve that best fit the relative transfer function data was introduced by Fletcher and Galt (1950), recently advocated by Studebaker and Sherbecoe (1991), and is shown below as:…”

“…Transfer function derived for words in this study with comparable curves for monosyllabic words (Black, 1959), NU-6 word lists (Schum et al, 1991) and the CID W-22 word test (Studebaker and Sherbecoe,199 1 17. The Frequency Importance function for words for the present Sttjdy, for NU-6 word lists (Shcum et al, 1991), for the CID W-22 word test (Studebaker and Sherbecoe, 1991) and for nonsense syllables (ANSI S 3. concerned with speech at various levels of its complexity; words, sentences and continuous discoarse (CD). Some researchers have concentrated solely upon techniques to measure aspects of intelligibility for these three types of speech.…”

“…Instead, they used a zero bands pair estimate (each SNR yielded a pair of band estimates, one HP and one LP) as the lone estimate of the frequeny importance of that band at that particular S/N ratio. Although no explanation was given, possibly Studebaker and Sherbecoe (1991) were of the opinion that band estimates of zero did not demonstrate that listeners were unable to use the available acoustic information, but rather they may have thought that scme condition within the experimental paradigm (such as excessive masking) prevented the listener from using the available acoustic information. This issue will be explored in the next chapter.…”

“…study. Previous studies have concentrated upon just one type of speech (Studebaker et al, 1987, Black, 1959, Studebaker and Sherbecoe, 1991and Schum et al, 1991. This following will compare the results of the present study with other research.…”