Spectral loudness summation as a function of duration for hearing-impaired listeners

Abstract

Spectral loudness summation was measured for 10- and 1000-ms long bandpass-noise signals with bandwidths in the range of 0.2 to 6.4 kHz centered at 2 kHz for nine listeners with hearing impairment of primarily cochlear origin. Loudness matches between signals with the same duration and different bandwidth were obtained using an adaptive two-interval, two-alternative forced-choice procedure with interleaved tracks. The 3200-Hz wide reference signal had a level of 45 dB or 65 dB SPL. Every signal was individually frequency equalized prior to presentation in order to ensure audibility of all spectral components for the two reference input levels. Generally the same amount of spectral loudness summation for 10- and 1000-ms long signals was obtained for both reference levels. However, some hearing-impaired listeners show, similarly to normal-hearing listeners, a larger spectral loudness summation for short than for long duration signals, indicating that duration effects in spectral loudness summation are caused by retrocochlear processes.

• Loudness
• Hearing-impaired
• Spectrotemporal effects
• Compression

Acronyms
ΔL	=	Level difference between test and reference signal needed to produce the sensation of equal loudness
AGC	=	Automatic gain control

Acronyms
ΔL	=	Level difference between test and reference signal needed to produce the sensation of equal loudness
AGC	=	Automatic gain control

Sumario

La sumación espectral de la intensidad subjetiva se midió con señales largas de ruido pasa banda de 10 y 1000 ms, con rangos de ancho de banda de 0.2 a 6.4 kHz, centrados en 2 kHz, en nueve sujetos con hipoacusia de origen fundamentalmente coclear. Se obtuvieron intensidades subjetivas equiparadas entre señales de la misma duración y diferente ancho de banda, usando un procedimiento adaptativo de dos intervalos y dos selecciones forzadas de dos alternativas, con pistas intercaladas. La señal de referencia de 3200 Hz de ancho tuvo un nivel de 45 o 65dB SPL. Cada señal fue ecualizada en frecuencias de manera individual antes de la presentación, con objeto de asegurar la audibilidad de todos los componentes espectrales para cada uno de los dos niveles de ingreso. Generalmente, se obtuvo la misma cantidad en la sumación espectral de la intensidad subjetiva para las señales de 10 y 1000 ms en ambos niveles de referencia. No obstante, algunos hipoacúsicos mostraron, al igual que los normooyentes, una sumación espectral de la intensidad subjetiva mas amplia para las señales cortas en vez de las señales largas, indicando que los efectos de duración de la sumación espectral de la intensidad subjetiva son causados por procesos retro-cocleares.

Notes

¹For one reference level, Verhey and Kollmeier (2002) also matched the loudness for signal and reference with the same bandwidth as a control. As expected, the data of almost all subjects show no level difference when test signal and reference signal have the same bandwidth. Hence, for the present study a level difference of 0 dB is assumed for this bandwidth.