Spectral loudness summation for short and long signals as a function of level

The equal-loudness-ratio hypothesis (ELRH) assumes that the loudness ratio between equal-level long and short signals with the same spectrum is independent of level and spectrum. Thus, it predicts the same amount of spectral loudness summation for equal-level short and long signals. To investigate how duration and level affect spectral loudness summation, loudness functions for 10- and 1000-ms bandpass-filtered noise centered at 2 kHz were obtained for normal-hearing listeners using categorical loudness scaling. The bandwidth was varied from 200 to 6400 Hz. From the loudness functions the level difference between the 3200-Hz-wide reference signal and the equally loud test signals was determined for reference levels in the range from 25 to 105 dB SPL. In addition, for two reference levels (45 and 65 dB SPL), this level difference was measured using a loudness-matching procedure. Both procedures showed similar results for the two reference levels. Over a wide level range spectral loudness summation is larger for the short than for the equal-level long signals. This is in contrast to the predictions of the ELRH. However, the data at moderate levels are consistent with the expectations of a modified ELRH which assumes a smaller loudness ratio for broadband than for narrow-band signals.