Clinical measurements of speech reception threshold in noise.

New lists of spoken sentences, edited in a computer, word by word, were tested clinically together with a noise, spectrally shaped as the speech. The purpose was to investigate the reliability and the learning effect of the speech reception threshold in noise, and also to assess the relations of this threshold to conventional audiometric measures. The threshold values ranged from -7 to +7 dB signal-to-noise ratio for the 97 ears investigated. Reliability, expressed as standard deviation for repeated measurements, deteriorated from 0.7 to 1.1 dB as the threshold deteriorated. Learning effect between the first and the second threshold increased from 0 to 1 dB as the threshold deteriorated. No conventional audiometric test showed a high correlation to this threshold measure. The discrimination of the sentences without noise as well as the self-rated speech recognition in noise were also investigated.

Some aspects of methodology in speech audiometry.

The aim of this study was to provide more knowledge about the influence of various factors on the reliability of speech audiometric measurements. On the basis of these data, new material on speech has also been developed for measurements of the speech reception threshold in noise. In a theoretical and a clinical investigation it has been shown that the reliability of speech discrimination scores may be predicted by using the binomial distribution. The reliability mainly depends on the score obtained and on the length of the word list. Curves with confidence intervals as well as diagrams which indicate when two different discrimination results deviate significantly from each other are presented. On the whole, the reliability is surprisingly poor. The reliability of speech reception threshold (SRT) measurements was studied by computer simulations and clinical studies. The SRT results of "subjects" who had poor speech discrimination were unreliable and were greatly affected by the level at which the test started. Some new methods were also simulated, one of which (the 2x4-word method) was 8% faster and also somewhat more reliable in the clinical studies than the ordinary 10-word method. It was also less affected by the patient's speech discrimination. Several speech lists with exactly the same content of sound, but with new sentences, were obtained by editing a list of ten spoken Swedish sentences in a computer. A noise was synthesized from the speech material by the computer to obtain the exact same spectrum of the speech and the noise. This material was tested monaurally on normal-hearing subjects and hearing-impaired subjects. A very steep intelligibility curve was obtained in noise (25%/dB), which gave a standard deviation of only 0.44 dB at repeated threshold measurements in noise for the normal-hearing subjects when the learning effect was outbalanced. This figure increased to 1.1 dB for those with severely impaired hearing. The learning effect between the first and the second thresholds did not exceed 1 dB. The threshold values ranged from -8 to -5 dB signal-to-noise ratio for those with normal hearing and from -7 to +7 dB for those with impaired hearing. The new sentence and noise material developed is well suited for clinical measurements of the speech reception threshold in noise.

Attempts to develop an efficient speech test in fully modulated noise.

An earlier developed speech test with sentences in noise (Hagerman. 1982) was modified in an attempt to further increase its efficiency. The noise was thus changed to be fully modulated and the speech-to-noise ratio of each word was controlled in order to make all the words equally difficult both within and between lists. The new version was tested on 10 normal-hearing subjects. The 10 lists were equally difficult, but compared to the old material the reliability was worse, the learning effect was greater, and the slope of the intelligibility function was less than half despite efforts to maintain the quality of the test. Identical masking signals for identical words in the various lists seem to increase the learning effect. The strongly modulated noise seems to give shallow intelligibility functions for separate words and thus also for complete lists.

To find the PB max--simulations of Levitt's adaptive method.

Levitt's (Scand Audiol 1978;6(Suppl):241-291) adaptive method for finding the speech level giving the maximum speech recognition score (PB max) was simulated in a computer. The results show that at least 50 words should be used to estimate the speech level at PB max before measuring the speech recognition at that level. The choice of stepping parameters is critical. Within a word pair, a 20 dB level difference is suggested. Scatter plots show the variability in the measured speech levels. The variability of conventional speech recognition measurements is also clearly illustrated.

Sentences for testing speech intelligibility in noise.

A list of ten spoken Swedish sentences was computer edited to obtain new lists with exactly the same content of sound, but with new sentences. A noise was synthesized from the speech material by the computer to produce exactly the same spectrum of speech and noise. The noise was also amplitude modulated by a low frequency noise to make it sound more natural. This material was tested monaurally on 20 normal-hearing subjects. The equality in intelligibility of some of the lists was investigated. Repeated threshold measurements in noise showed a standard deviation of 0.44 dB when the learning effect was outbalanced. Only a small part of the learning effect was due to learning of the word material. Intelligibility curves fitted to the data points in noise and without noise showed maximum steepnesses of 25 and 10%/dB respectively. At constant signal to noise ratio (S/N) the best performance was achieved at a speech level of 53 dB.

Measurement of speech reception threshold. A comparison between two methods.

In an earlier work (Hagerman, 1979) several speech reception threshold (SRT) methods were simulated in a computer. New methods were found to be more effective than the common 10-word method. In a clinical investigation the new 2 X 4-word method was compared with the common method. The result showed that the new method required fewer words than the common method. New simulations also showed that the 2 X 4-word method was somewhat more reliable than the 10-word method.

Reliability in the determination of speech reception threshold (SRT).

Clinical measurements of the speech reception threshold (SRT) have been made in order to determine the parameter values of a psychometric function. SRT measurements were then simulated in a computer to obtain, among other things, the standard deviation of a number of measurements from one and the same "patient". "Patients" with poor speech discrimination resulted in bad reliability of the SRT and great sensitivity to the hearing level at which the test started. New methods are proposed here, which are both faster and less susceptible to influence from the patient's speech discrimination.

Efficiency of speech audiometry and other tests.

A formula for expressing the efficiency of audiological tests is proposed: E = (vb/vw - 1)/T where vb is the variance between subjects, vw is the variance within subjects and T is the testing time. This makes it possible to compare various tests even if they require different testing times. The formula has been used on various speech tests and on pure-tone thresholds for a specific group of hearing-impaired subjects, showing that pure-tone threshold tests are much more efficient than speech tests. Some suggestions are given on how to make speech tests in noise more efficient. It is concluded that the proposed formula of test efficiency should be considered when new tests are designed in order to get quick tests which add as much information as possible.

A clinical comparison between a laboratory and a commercial impedance audiometer.

Impedance audiometry is a useful tool in the diagnosis of retrocochlear lesions, provided that adequate normal values regarding the stapedius reflex thresholds and decay are used. The normal values in Sweden (and elsewhere) have been obtained with a laboratory impedance audiometer that is not in common use. We have compared a commercial impedance audiometer, the Grason-Stadler 1723, with this laboratory model to determine whether the normal values are valid. We found that the Grason-Stadler 1723 audiometer gives reliable stapedius reflex thresholds, but at 2-4 dB higher levels. The reflex decay curves, however, are unsatisfactory. Possible explanations for the results are discussed. We have also proposed a modification of the recorder of the commercial audiometer. The study demonstrates that normal values cannot be transferred freely between different equipments.

Questionnaires on desirable properties of hearing aids.

Three questionnaires were presented to 49 hearing-impaired patients. They were asked to rank different properties of behind-the-ear hearing aids with regard to their importance. The first questionnaire dealt with physical properties such as size, weight, and battery-cost, but also with the importance of good sound quality. The second questionnaire focused on the relative importance of various sound quality dimensions, such as softness, clarity, fullness, and others. The third questionnaire asked for suitable sound programs to be used in the procedure of hearing aid fitting in the clinic. Sound quality scored the highest rank by 43 of the patients, and clarity was considered to be the most important sound quality dimension. The most preferred sound programs included male and female voices with or without a background of other voices.

Efficient adaptive methods for measuring speech reception threshold in quiet and in noise.

An adaptive method for measuring speech reception threshold in noise (S/N-threshold) was evaluated with regard to reliability and learning effect. A corresponding method was also tested for speech reception threshold in quiet (SRT). The methods are designed for use with a speech and noise material developed by Hagerman (1982a). Ten normal-hearing subjects and 40 hearing-impaired subjects were tested, the results showing that the test-retest reliability of the S/N-threshold was much better with the new, adaptive, method than with the old one, which was not adaptive. With the new method, a S/N-threshold can be measured within two minutes for most patients (after the initial short training) with a reliability, i.e. standard deviation, of 0.78 dB and a learning effect that is negligible for most subjects if one training list is presented before the test session. It was also concluded that the adaptive method designed for the sentences in quiet gives at least as reliable an SRT result as the old spondée method (including learning effects), but in a shorter time.

Perceived sound quality of reproductions with different frequency responses and sound levels.

Three programs (female voice, jazz music, and pink noise) were reproduced using four different frequency responses and two different sound levels. Fourteen normal hearing subjects listened to the reproduction via earphones and judged the sound quality on seven perceptual scales (loudness, clarity, fullness, spaciousness, brightness, softness/gentleness, and nearness) and a fidelity scale. Significant differences among the reproductions appeared in all scales and could be attributed to the differences in frequency response or sound level or both. Interactions between the reproductions and the programs could be explained by the relations between the spectrum of the programs and the frequency responses used. The results for the noise program were similar to those for the jazz music program.

The effects of different frequency responses on sound quality judgments and speech intelligibility.

Four speech programs and two music programs were reproduced using five different frequency responses: one flat and the others combinations of reductions at lower frequencies and/or increases at higher frequencies. Twelve hearing impaired (HI) and 8 normal hearing (NH) subjects listened monaurally to the reproductions at comfortable listening level and judged the sound quality on seven perceptual scales and a scale for total impression. Speech intelligibility was measured for phonetically balanced (PB) word lists and for sentences in noise. Significant differences among the reproductions appeared in practically all scales. The most preferred system was characterized by a flat response at lower frequencies and a 6 dB/octave increase thereafter. There were certain differences between the NH and HI listeners in the judgments of the other systems. The intelligibility of PB word lists did not differ among the systems, and the S/N threshold for the sentences in noise only distinguished the flat response as worse than all others for the HI listeners. There was little correspondence between intelligibility measures and sound quality measures. The latter provided more information and distinctions among systems.

Relations between speech intelligibility and psychoacoustical modulation transfer function (PMTF).

The modulation transfer function, MTF, has proved to be a powerful measure for predicting speech intelligibility in speech transmission channels. We extended it to include the ear, by measuring the psycho-acoustical MTF, i.e. the PMTF. Tone thresholds of 11 normal-hearing and 20 hearing-impaired subjects were measured in presence of unmodulated and intensity modulated noise. Octave frequencies from 500 to 4,000 Hz were used. The noise was octave filtered around the frequency of the probe tone. Six modulation frequencies from 1 to 50 Hz were used. From these results the PMTFs were calculated, as well as the corresponding psycho-acoustical speech transmission indices, i.e. the PSTIs. The subjects' speech discrimination scores in quiet and speech reception thresholds in noise were also measured. A correlation coefficient of 0.85 between the speech discrimination score in quiet and the PSTI was obtained. For the speech reception threshold in noise and the PSTI, the correlation was 0.71. The first of these two figures is promising, but our method needs some improvement, as it gave some problems due to fatigue effects.

Sound quality judgment during acclimatization of hearing aid.

This study included 10 subjects with sensorineural hearing loss. They had not previously used hearing aids. The hearing aid fitting with "behind-the-ear' aids was based on the NAL method. Judging the sounds in their ordinary home environment, three sound situations (programmes) were selected, permitting the subjects to listen to one programme with speech, another with music and a third with "background noise'. For the group as a whole, significant differences were found between the sound quality judgements with and without the hearing aid. When the judgements of the programmes with background noise were excluded in the analysis, there were significant differences over time in the group as a whole on the scales for clarity and total impression.

Perceived sound quality in a hearing aid with vented and closed earmould equalized in frequency response.

Nine normal-hearing subjects listened to speech and music in a hearing aid, either through a vented earmould or a closed earmould. The complex frequency responses of the two systems were made equal by compensating the aid and closed mould combination with a digital filter. The subjects rated the perceived sound quality of the systems on seven perceptual scales and a scale for overall impression. The results of the ratings support the hypothesis that there is no difference in perceived sound quality between vented and closed earmoulds that are equalized in frequency response, provided that the perceived loudness is the same in both cases.