The viability of speech-in-noise audiometric screening using domestic audio equipment

Abstract

Speech-in-noise audiometry has potential application as a low-cost, self-screening test for sensorineural hearing loss. To realize this potential, the influence of variations in audio equipment and listening environment need assessment. The present study assessed: 1) the frequency response and distortion produced by a wide range of commercially available audio equipment; 2) the effects of such variations upon test results with normally hearing subjects using a simple, open-set, word-identification test; 3) the effect of distortion on the speech reception threshold using digitally applied distortion; and 4) the reliability of the test in listening environments with different levels of reverberation. In addition, preliminary tests were conducted with elderly listeners. The results indicate that variations in equipment have negligible effects on speech-in-noise audiometry. The only factor that substantially elevated normally hearing listeners’ thresholds was high levels of room reverberation when using loudspeaker presentation. Variations in equipment and environment thus present no significant obstacle to the development of a self-administered audiometric screening test based on speech in noise.

Sumario

La logoaudiometría en ruido tiene una aplicación potencial como prueba de auto-evaluación a bajo costo para pérdidas auditivas sensorineurales. Para desarrollar este potencial se necesita evaluar la influencia de las variaciones en el equipo de audio y en el ambiente de escucha. El presente estudio evaluó: 1) la respuesta frecuencial y la distorsión producida por una amplia gama de equipos de audio disponibles comercialmente; 2) los efectos de tales variaciones sobre los resultados de las evaluaciones con sujetos normo-oyentes usando una prueba de identificación de palabras de contexto abierto; 3) el efecto de distorsión sobre el umbral de recepción del lenguaje utilizando distorsión digitalmente aplicada; 4) la confiabilidad de la prueba en ambientes para escuchar con diferentes niveles de reverberación. Los resultados indican que las variaciones en el equipo generan efectos despreciables en la logoaudiometría en ruido. El único factor que elevó sustancialmente los umbrales en sujetos normo-oyentes fueron los altos niveles de reverberación de la cabina cuando la presentación se hizo a través de altoparlantes. Las variaciones en el equipo o en el ambiente, por lo tanto, no constituyeron un obstáculo significativo en el desarrollo de una prueba auto-administrada de tamizaje auditivo basada en lenguaje en ruido.

Notes

¹Kucera-and-Francis frequency is a count of the total number of occurrences of a word in the million-word corpus of text they analysed.

²It should be noted that subsequent analysis showed the second stage of calibration was probably not performed with sufficient precision; more data were required to reliably optimize the SNR of each word. Nonetheless, when normally hearing listeners were tested with the pilot version of the test on CD (following this second stage of calibration and rescaling), their scores increased from 29.1% to 82.1% over a 3 dB range, an average slope of 17.7 %/dB. A logistic psychometric function fitted to these average data had a slope parameter equivalent to 16.5 %/dB at the 50% point. Both of these statistics are superior to those observed in either calibration stage, despite the averaging of data across seven different pairs of headphones. It is not clear how large a role selection of materials may have played in this improvement, but it does indicate that overall the second stage of calibration was effective.

³The observed SRTs are somewhat lower than those normally reported for a noise interferer. This is because the Brown noise used in this experiment had approximately 2/3 of its energy below 0.1 kHz, where it had little masking effect upon the speech.

⁴The simulation was based upon 63 trials in order to simulate the data in Figure 8 (i.e. the average of 3 runs of the pilot test on each ear using different headphones). Logistic psychometric functions were fitted to the data from normally hearing listeners using the same stimuli, in order to produce probabilities of correct answer for any SNR. In order to model a hearing-impaired individual, this psychometric function was shifted by an amount proportional to his or her individual PTA (0.08 dB/dB(HL)). The resulting probabilities of correct answer at each test SNR were compared with random numbers between 0 and 1 to determine the outcome of each simulated trial. The resulting scores were correlated back with the individual PTAs.

⁵The optimum SNR depends upon the psychometric functions of both groups. If one places on one side the problem that hearing-impaired listeners are a heterogeneous group, and assumes, instead, that they have similar psychometric functions to normally hearing listeners, but shifted upward by approximately 4 dB (Plomp, 1994), then the most discriminating SNR will be at 2 dB above the 50% point for normally hearing listeners. Using this optimistic assumption, a high level of sensitivity (99.6%) and specificity (99.9%) is predicted using only 21 trials and a pass threshold of 10 items correct. The effect of using a less optimal SNR is not dramatic, however. Using the 50% point itself reduces the sensitivity and specificity to 99.1% and 99.4%, respectively, although the pass threshold has to be reduced to just 5 correct.