Loudness of Speech and Speech-Like Signals

Abstract

The influence of different properties of speech-likesignals on loudness wasinvestigated by measuring levels at equal loudness in normal-hearing listeners for 22 speech-likes timuli using categorical loudness scaling and an adaptive matching procedure. Eight of the stimuli, referred to as unprocessed stimuli, had the same speech-like long-term spectrum, butdiffered in other speech-related properties, ranging from astationary noise overspeechmodulated signals to real intelligible and unintelligible speech. The other stimuli were designed to investigate the influence of severe modifications of speech-related properties due to spectral filtering, reverberation, compression, and expansion on loudness. In general, results were similar for both measurement methods and revealed only small effects of temporal modulations on loudness perception, and no influence of intelligibility or reverberation. Even severe modifications of the stimuli by highpass-filtering and expansion did not affect levels at equal loudness. Only lowpass-filtering significantly decreased loudness and compression significantly increased loudness. The experimental results indicate that the long-term spectrum of speech has amajor influence on loudness while other properties of speech only have aminor contribution. Some of the results could not be predicted by loudness models based on short-term loudness. Instead, longer time constants were required to not overestimate the loudness of time-varying speech-likesounds.

Effectiveness of a loudness model for time-varying sounds in equating the loudness of sentences subjected to different forms of signal processing

“…The results of the experiments are consistent with earlier results showing that the LTL predicted by the TVL model could give accurate predictions of the loudness of speech that had been subjected to multi-channel amplitude compression of the type that is often used in broadcasting (Moore et al, 2003). It is also consistent with the results of Rennies et al (2013), obtained using both loudness matching and categorical loudness scaling, which showed that the LTL gave reasonably accurate predictions of the loudness of a variety of speech-like signals (including speech-shaped noise, unprocessed speech, and speech that was subjected to filtering, reverberation, and amplitude compression and expansion), whereas the predictions were not as accurate when based on the short-term loudness derived using the TVL model or other loudness models (Chalupper and Fastl, 2002;Rennies et al, 2009). Table II summarizes the results of the two experiments, showing the mean adjustments in level relative to equal RMS required to equalize the peak LTL or the mean LTL and the mean rating obtained for each equalization method and type of processing.…”

Loudness of complex time-varying sounds – A challenge for current Loudness models

“…The level of the reference signal was fixed at 70 dB SPL. The level of the test signal was set adaptively depending on the subject's response (for details, see Rennies et al, 2013). This procedure resulted in an estimate of the level at which the test signal was perceived as equally loud as the reference signal (presented at the reference level).…”

“…For each measurement, this procedure resulted in a categorical loudness function of cu values vs. sound pressure level. These functions were then used to derive level differences at equal loudness by computing the horizontal distance between the loudness function of the reference signal and all other loudness functions at the cu value corresponding to a level of 70 dB SPL for the reference signal (i.e., the reference level; for details, see Rennies et al, 2013). The same apparatus as for the loudness scaling experiments was used.…”

“…All stimuli were at least 2 s long (cutting was made in the next possible speech pause for signals with speech-like envelopes) and were frozen, i.e., the same stimulus samples were used throughout the experiments. Apart from intelligibility and the type of temporal modulations, the stimuli also differed in their ratio between peak and root-mean-square (rms) value and whether or not they contained components with fundamental frequency or temporal gaps (see Rennies et al, 2013). The third-octave-band spectra of the stimuli were equalized to match the male long-term average speech spectrum (LTASS) of Byrne et al (1994) in order to eliminate the influence of long-term spectrum on loudness perception.…”

“…The present study describes a series of experiments that was conducted to test existing loudness models with two sets of real sounds. The first set comprised different types of speech-like signals which varied in their similarity to real speech (Rennies et al, 2013). The second set consisted of different technical sounds including stationary sounds as well as sounds with strong impulsive components.…”