Chaslav V. Pavlovic

Derivation of primary parameters and procedures for use in speech intelligibility predictions

The literature on various parameters that appear in the articulation index-type calculations of speech intelligibility is reexamined. Based on the reported data, the best estimates of these parameters and the most appropriate procedures for their use are suggested. These included: (1) the analysis and specification of the importance of various frequency bands to speech intelligibility; (2) the procedures used for measuring threshold and the calculation of threshold-based parameters used for predicting intelligibility of low-level speech; and (3) the calculation and measurement of relevant speech parameters. All results are given so that the calculations can be performed either in critical bands, 1/3 octaves, or octaves.

A frequency importance function for continuous discourse

Normal hearing subjects estimated the intelligibility of continuous discourse (CD) passages spoken by three talkers (two male and one female) under 135 conditions of filtering and signal-to-noise ratio. The relationship between the intelligibility of CD and the articulation index (the transfer function) was different from any found in ANSI S3.5-1969. Also, the lower frequencies were found to be relatively more important for the intelligibility of CD than for identification of nonsense syllables and other types of speech for which data are available except for synthetic sentences [Speaks, J. Speech Hear. Res. 10, 289-298 (1967)]. The frequency which divides the auditory spectrum into two equally important halves (the crossover frequency) was found to be about 0.5 oct lower for the CD used in this study than the crossover frequency for male talkers of nonsense syllables found in ANSI S3.5-1969 and about 0.7 oct lower than the one for combined male and female talkers of nonsense syllables reported by French and Steinberg [J. Acoust. Soc. Am. 19, 90-119 (1947)].

Transfer functions and correction factors used in hearing aid evaluation and research.

In hearing aid research and application it is often necessary to transform the sound pressure level or acoustic gain measured at one physical point (e.g., 2 cc coupler) to an equivalent value at another physical point (e.g., tympanic membrane). Results of currently available studies on various transformations are summarized in this report. When no data were available, the desired transfer function was calculated as a linear combination of known contributing transfer functions. Values are given for the center frequencies of 1/3 octaves and critical bands.

Use of the articulation index for assessing residual auditory function in listeners with sensorineural hearing impairment

Two experiments were performed to explore the feasibility of deriving an articulation index (AI) based scheme for quantifying residual auditory function in listeners with sensorineural hearing impairment. In the first experiment the AI procedure was applied to audiograms of normal and sensorineural hearing-impaired individuals in order to predict speech discrimination under various conditions. Good predictions were found for the normal and the less impaired subjects, but not for those with greater impairments. These subjects exhibited a disproportionate loss in speech discrimination compared to that predicted on the basis of the AI procedure. In the second experiment it was investigated whether the AI predictions for presbycusic individuals could be improved by making use of the proficiency factor that is not a function of frequency. The results indicate that the deficit in suprathreshold speech processing is frequency-specific and that, therefore, the proficiency factor is not independent of frequency. It assumes the value of 1 at those frequencies where hearing sensitivity is normal and is substantially reduced for frequencies where hearing threshold is elevated.

An evaluation of some assumptions underlying the articulation index

The purposes of this investigation were to derive measurement and calculation procedures for the articulation index (AI) that were as precise and accurate as possible, and to evaluate the significance of the differences in assumptions made by early investigators. A nonsense syllable test specifically designed for this application was recorded. The implications of various assumptions utilized in different approaches to predicting speech sound identification were tested using 13 different conditions which differed with respect to filtering, level of interfering noise, or level of the speech signal. The listeners were three normal hearing individuals. A set of preferred assumptions was chosen based on a post hoc evaluation of the data. Predictions of speech sound identification were good, both for the average listener and for individual listeners.

Band importance functions for audiological applications

Recently, band importance functions have been developed for a number of speech tests used in audiology. These functions, as well as the importance functions for average (everyday) speech, are also being considered for the inclusion in the revised Articulation Index standard (submitted for vote to the Acoustical Society of America). In this paper, the band importance functions for different speech materials (usually reported in literature for 113 octave bands) have been recalculated to correspond to frequency bands normally used in audiological applications. In addition, criteria for selecting appropriate importance functions and transfer functions are discussed.

Frequency importance functions for a feature recognition test material

The relative importance of different parts of the auditory spectrum to recognition of the Diagnostic Rhyme Test (DRT) and its six speech feature subtests was determined. Three normal hearing subjects were tested twice in each of 70 experimental conditions. The analytical procedures of French and Steinberg [J. Acoust. Soc. Am. 19, 90-119 (1947)] were applied to the data to derive frequency importance functions for each of the DRT subtests and the test as a whole over the frequency range 178-8912 Hz. For the DRT as a whole, the low frequencies were found to be more important than is the case for nonsense syllables. Importance functions for the feature subtests also differed from those for nonsense syllables and from each other as well. These results suggest that test materials loaded with different proportions of particular phonemes have different frequency importance functions. Comparison of the results with those from other studies suggests that importance functions depend to a degree on the available response options as well.

Comparison of discomfort levels obtained with pure tones and multitone complexes

The relationship between threshold of discomfort (TD) estimates and the number of components in a complex signal has been investigated. The thresholds of discomfort were first obtained for 16 pure tones located at the center frequency of critical bands from 250 to 4000 Hz. Subsequently, thresholds of discomfort were obtained for 2, 4, 8, and 16 tone complexes. The pure-tone components of the complexes were systematically selected from the same 16 pure tones. For each subject, the relative intensities of the components in the four complexes were determined in such a way so as to parallel the pure tone TD contour obtained for that subject. Data were obtained from 15 normal and 15 hearing impaired adults. The individuals in the latter group all had mild to moderate sensorineural hearing loss. Summation of discomfort (S) was defined as the difference between the threshold of discomfort for a pure tone presented in isolation and within the complex. The two groups demonstrated different summation values. For both groups, however, the summation was shown to be a linear function of the logarithm of the number of components in the complex: S = a + b log (n) where n is the number of components (2, 4, 8, 16). For the normal hearing group, a and b are 2.05 and 11.51, respectively, while for the hearing impaired group, they are 3.95 and 12.88, respectively. While the future digital hearing aids can easily regulate their limiting levels so as to accurately account for this summation, present day hearing aids may underestimate this effect.(ABSTRACT TRUNCATED AT 250 WORDS)

Use of the magnitude estimation technique for assessing the performance of text‐to‐speech synthesis systems

As text-to-speech systems develop, it becomes necessary to compare various solutions and to evaluate whether a change in the synthesis procedure has an effect on the listeners attitude to the system. The possibility of directly scaling intelligibility, naturalness, and users satisfaction (i.e., acceptability) with the magnitude estimation technique is investigated. A magnitude estimation protocol suitable for this purpose is described. In general, within the limits of the methodological constraints discussed in this paper, the procedure appears to be reliable and valid for quantifying the perceived attributes of synthesized speech.

Addendum to “Transfer Functions and Correction Factors Used in Hearing Aid Evaluation and Research”

In an earlier publication, various transformations used in hearing aid research and its application were summarized. As a result of continued interest and requests, additional transfer functions are provided in this addendum.