Loudness Model Research Articles

Implementing Zwicker sharpness using specific loudness calculated from the “Procedure for the Computation of Loudness of Steady Sounds.”

Sharpness as defined by Fastl and Zwicker [2007] is a measure of the weighted spectral balance of a sound, i.e., sounds with relatively greater high-frequency content have a greater sharpness than those with proportionally greater low-frequency content. The sharpness percept also influences the overall “pleasantness” of a sound and is thus important to sound quality. Measures of sharpness have been predominantly constructed to work with Zwicker's loudness model [DIN 45631/ISO 532B], with less attention to the standardized model given in “Procedure for the Computation of Loudness of Steady Sounds” [ANSI S3.4-2007]. In this paper, a mathematical method for adapting Zwicker's sharpness measure for use with the ANSI S3.4-2007 loudness standard (and other loudness metrics producing a specific loudness distribution) is discussed. Benchmarking results for the resultant metric and potential limitations of this approach are also addressed.

The loudness of an amplitude-modulated sinusoid as a function of interaural modulator phase, modulation rate, and level

The aim was to test a model of loudness for binaurally presented time-varying sounds (Moore et al., Trends in Hearing, in press). A 1000-Hz sinusoidal carrier was 100% sinusoidally amplitude modulated. The effect on its loudness of varying the interaural modulation phase difference (the IMPD) was assessed. The IMPD of the test sound was 90° or 180° and that of the comparison sound was 0°. A two-interval, two-alternative forced-choice method with a one-up/one-down rule was used to estimate the level difference between the test and the comparison sounds at the point of equal loudness (the LDEL) for baseline levels of 30 and 70 dB SPL and modulation rates of 1, 2, 4, 8, 16, and 32 Hz. The LDELs were negative (mean = -1.2 and -1.6 dB for IMPDs of 90° and 180°), indicating that non-zero IMPDs led to increased loudness. The model predicted that the LDELs should be most negative for modulation rates up to 4 Hz, and should be close to zero for the rate of 32 Hz. The data showed a pattern similar to the predicted pattern, but the LDELs for the modulation rate of 32 Hz were about 0.6 dB more negative than predicted.

Physiologically motivated individual loudness model for normal hearing and hearing impaired

Physiologically motivated individual loudness model for normal hearing and hearing impaired

Perceptual Evaluation and Analysis of Reverberation in Multitrack Music Production

This work was made possible by the Engineering and Physical Sciences Research Council Grant EP/K009559/1 Platform Grant: Digital Music, 2013-18. The authors also wish to thank Dominic Ward for a fruitful discussion on loudness models and related features.

Evaluation of near-end speech enhancement under equal-loudness constraint for listeners with normal-hearing and mild-to-moderate hearing loss

Four algorithms designed to enhance the intelligibility of speech when noise is added after processing were evaluated under the constraint that the speech should have the same loudness before and after processing, as determined using a loudness model. The algorithms applied spectral modifications and two of them included dynamic-range compression. On average, the methods with dynamic-range compression required the least level adjustment to equate loudness for the unprocessed and processed speech. Subjects with normal-hearing (experiment 1) and mild-to-moderate hearing loss (experiment 2) were tested using unmodified and enhanced speech presented in speech-shaped noise (SSN) and a competing speaker (CS). The results showed (a) the algorithms with dynamic-range compression yielded the largest intelligibility gains in both experiments and for both types of background; (b) the algorithms without dynamic-range compression either yielded benefit only with the SSN or yielded no consistent benefit; (c) speech reception thresholds for unprocessed speech were higher for hearing-impaired than for normal-hearing subjects, by about 2 dB for the SSN and 6 dB for the CS. It is concluded that the enhancement methods incorporating dynamic-range compression can improve intelligibility under the equal-loudness constraint for both normal-hearing and hearing-impaired subjects and for both steady and fluctuating backgrounds.

Pure-Tone-Spondee Threshold Relationships in Functional Hearing Loss: A Test of Loudness Contribution.

The purpose of this article is to examine explanations for pure-tone average-spondee threshold differences in functional hearing loss. Loudness magnitude estimation functions were obtained from 24 participants for pure tones (0.5 and 1.0 kHz), vowels, spondees, and speech-shaped noise as a function of level (20-90 dB SPL). Participants listened monaurally through earphones. Loudness predictions were obtained for the same stimuli by using a computational, dynamic loudness model. When evaluated at the same SPL, speech-shaped noise was judged louder than vowels/spondees, which were judged louder than tones. Equal-loudness levels were inferred from fitted loudness functions for the group. For the clinical application, the 2.1-dB difference between spondees and tones at equal loudness became a 12.1-dB difference when the stimuli were converted from SPL to HL. Nearly all of the pure-tone average-spondee threshold differences in functional hearing loss are attributable to references for calibration for 0 dB HL for tones and speech, which are based on detection and recognition, respectively. The recognition threshold for spondees is roughly 9 dB higher than the speech detection threshold; persons feigning a loss, who base loss magnitude on loudness, do not consider this difference. Furthermore, the dynamic loudness model was more accurate than the static model.

Tonotopic representation of loudness in the human cortex

A prominent feature of the auditory system is that neurons show tuning to audio frequency; each neuron has a characteristic frequency (CF) to which it is most sensitive. Furthermore, there is an orderly mapping of CF to position, which is called tonotopic organization and which is observed at many levels of the auditory system. In a previous study (Thwaites et al., 2016) we examined cortical entrainment to two auditory transforms predicted by a model of loudness, instantaneous loudness and short-term loudness, using speech as the input signal. The model is based on the assumption that neural activity is combined across CFs (i.e. across frequency channels) before the transform to short-term loudness. However, it is also possible that short-term loudness is determined on a channel-specific basis. Here we tested these possibilities by assessing neural entrainment to the overall and channel-specific instantaneous loudness and the overall and channel-specific short-term loudness. The results showed entrainment to channel-specific instantaneous loudness at latencies of 45 and 100 ms (bilaterally, in and around Heschl's gyrus). There was entrainment to overall instantaneous loudness at 165 ms in dorso-lateral sulcus (DLS). Entrainment to overall short-term loudness occurred primarily at 275 ms, bilaterally in DLS and superior temporal sulcus. There was only weak evidence for entrainment to channel-specific short-term loudness.

Frequency characteristics of neuromagnetic auditory steady-state responses at the supra-threshold level

Auditory steady-state response (ASSR) is a neuronal electrical component recorded as a continuous sinusoidal signal phase-locked to a modulation frequency of sounds. Since ASSR amplitude shows high correlation with perceptual characteristics at hearing threshold level, the ASSR has been examined as a candidate for an objective index of the human hearing ability per frequency. However, the relationship between ASSR and perception at the supra-threshold level has not been clarified. In this study, characteristics of ASSR magnitude relative to loudness at the supra-threshold level were investigated. Neuromagnetic 40-Hz ASSR was recorded in response to sinusoidally amplitude-modulated sweep tones with carrier frequency covering the frequency range of 0.1-12.5 kHz. Sound intensity was equalized at 50-, 60-, and 70-dB SPL with an accuracy of ±0.5-dB SPL at the phasic peak of the modulation frequency. Corresponding loudness characteristics were modeled by substituting the detected individual hearing thresholds into a standard formula (ISO226:2003(E)). The strength of the ASSR component was maximum at 0.5 kHz, and it decreased linearly on logarithmic scale toward lower and higher frequencies, whereas loudness model was plateaued between 0.5 and 4 kHz. The results indicated that frequency characteristics of the ASSR were not equivalent to those of SPL and loudness model.

Influence of simultaneous grouping mechanisms in the loudness judgment of a multi-tone mixture

Our main concern is to understand how the perceptual organization of an acoustical mixture into distinct auditory objects (known as an ASA mechanism) can affect the judgments of loudness. Three main questions have been raised: Is the loudness of one auditory object influenced by concurrent objects? How the partial loudness of different auditory objects are combined into one global loudness percept? Does global loudness depend on the perceptual organization of the mixture? In the present experiment, the mixtures consisted of frequency-modulated harmonic complexes. The same or different modulation frequencies have been applied to the odd and even harmonics to ensure that participants perceived either one or two auditory objects, respectively. As a first result, when two objects were well segregated (without spectral masking), the loudness of one object in the mixture equaled the loudness of the same object presented alone. It suggests that spectral grouping mechanisms prevail over loudness. Then, the global loudness pattern obtained for two auditory objects as a function of their partial loudness will be drawn up. Global loudness modeling for two auditory objects will be discussed with regards to the predictions given by classical loudness models. [Research supported by the chair “MOUVIE”, carried by the UPMC.]

Effects of spectral gaps on the contribution of specific frequency bands to the loudness of broadband sounds

Contributions of individual frequency bands to judgments of total loudness can be assessed by varying the level of each band independently from one presentation to the next and determining the relation between the change in level of each band and the loudness judgment. Recent studies using stimuli consisting of 15 bands of noise, each two critical bands wide, have shown that frequencies near the lower and upper edges of a broadband noise make a greater contribution to overall loudness than frequencies in the center of the spectrum. In the present study, eight listeners with normal hearing compared the loudness of two 15-band stimuli with average levels of 59 dB SPL per band and of stimuli where 1, 3, 5, or 7 bands were missing from the center of the spectrum. Listeners placed greater weight on the bands at the edges of the spectral gap and the effect was larger for greater gap widths. Edge bands received more weight regardless of absolute frequency and the high-frequency edge of the low bands received greater weight than the low-frequency edge of the high bands. These effects are not predicted by loudness models. [Work supported by NIH.]

Physiological motivated transmission-lines as front end for loudness models.

The perception of loudness is strongly influenced by peripheral auditory processing, which calls for a physiologically correct peripheral auditory processing stage when constructing advanced loudness models. Most loudness models, however, rather follow a functional approach: a parallel auditory filter bank combined with a compression stage, followed by spectral and temporal integration. Such classical loudness models do not allow to directly link physiological measurements like otoacoustic emissions to properties of their auditory filterbank. However, this can be achieved with physiologically motivated transmission-line models (TLMs) of the cochlea. Here two active and nonlinear TLMs were tested as the peripheral front end of a loudness model. The TLMs are followed by a simple generic back end which performs integration of basilar-membrane "excitation" across place and time to yield a loudness estimate. The proposed model approach reaches similar performance as other state-of-the-art loudness models regarding the prediction of loudness in sones, equal-loudness contours (including spectral fine structure), and loudness as a function of bandwidth. The suggested model provides a powerful tool to directly connect objective measures of basilar membrane compression, such as distortion product otoacoustic emissions, and loudness in future studies.

Zwickerのラウドネスモデルに基づく音響構造連成メタサーフェスのトポロジー最適化

Zwickerのラウドネスモデルに基づく音響構造連成メタサーフェスのトポロジー最適化

A Loudness Model for Time-Varying Sounds Incorporating Binaural Inhibition.

This article describes a model of loudness for time-varying sounds that incorporates the concept of binaural inhibition, namely, that the signal applied to one ear can reduce the internal response to a signal at the other ear. For each ear, the model includes the following: a filter to allow for the effects of transfer of sound through the outer and middle ear; a short-term spectral analysis with greater frequency resolution at low than at high frequencies; calculation of an excitation pattern, representing the magnitudes of the outputs of the auditory filters as a function of center frequency; application of a compressive nonlinearity to the output of each auditory filter; and smoothing over time of the resulting instantaneous specific loudness pattern using an averaging process resembling an automatic gain control. The resulting short-term specific loudness patterns are used to calculate broadly tuned binaural inhibition functions, the amount of inhibition depending on the relative short-term specific loudness at the two ears. The inhibited specific loudness patterns are summed across frequency to give an estimate of the short-term loudness for each ear. The overall short-term loudness is calculated as the sum of the short-term loudness values for the two ears. The long-term loudness for each ear is calculated by smoothing the short-term loudness for that ear, again by a process resembling automatic gain control, and the overall loudness impression is obtained by summing the long-term loudness across ears. The predictions of the model are more accurate than those of an earlier model that did not incorporate binaural inhibition.

Loudness of Speech Transmitted via Handsfree Telephone Systems - Perceptual Measurements and Loudness Models in Free Field Listening

Loudness of Speech Transmitted via Handsfree Telephone Systems - Perceptual Measurements and Loudness Models in Free Field Listening

How should we move forward with the next version of the ANSI S3.4 loudness standard?

Recent updates to the international loudness standard ISO 532 reflect both increased understanding of loudness and the use of loudness standards by the acoustics and engineering communities. Both are inspiration to consider revising ANSI S3.4-2007. The current ANSI standard is based on Moore and Glasberg’s model of loudness for stationary sounds. The new ISO 532-2 is a revision of this. The new ISO 532-1 is based on Zwicker’s time-varying loudness, and will replace the current ISO 532B, also based on Zwicker’s loudness model but for stationary sounds. For people who are not developing loudness models but are using them to assess sounds or noise, there are a number of other issues, e.g., loudness of very low frequency sounds, and exposure assessment based on loudness. These will be discussed and possible directions in updating the current ANSI standard are presented.

The effect of stimulus bandwidth on binaural loudness summation.

Binaural loudness summation is an important property of the human auditory system. This paper presents an experimental investigation of how binaural loudness summation varies with stimulus bandwidth. Loudness matches were obtained between dichotic stimuli, with interaural level differences (ILDs) of 2-12 dB, and diotic stimuli. The stimuli were noise bands with seven center frequencies and four bandwidths. Results showed that the loudness of dichotic stimuli increased nonlinearly with ILD, the increase being slightly less with broader bandwidths. There was a bandwidth-dependent difference between the listening tests results and the predictions of Moore and Glasberg's [(2007) J. Acoust. Soc. Am. 121, 1604-1612] loudness model. The size of the difference was, however, small. A characteristic function was derived describing how overall loudness depends on stimulus bandwidth and ILD.

Loudness of temporally varying environmental sounds

Loudness of speech, speech-like signals, and other dynamic environmental sounds were measured and compared to predictions of current loudness models and predictions on the basis of the relevant standards. Loudness was assessed experimentally by using a loudness matching procedure and categorical loudness scaling. The scaling method used in our study is in agreement with the requirements of the international standard on categorical loudness scaling. Categorical scaling allows for a fast assessment of loudness over a large level range but has the disadvantage that loudness is not measured in sones, as commonly used in loudness models. The data of the two methods are compared by deriving levels at equal loudness from the categorical loudness data. In addition, the present study discusses to what extend the scaling data can be compared to loudness predictions by using recently proposed equations relating categorical units to sones. The comparison of the measured levels at equal loudness and simulations revealed that for speech and speech-like signals the long-term spectrum largely determines the loudness of the sound. Dynamic models with short time constants tend to overestimate loudness. In general, this is also true for the other environmental sounds although for some technical signals discrepancies remain.

Some factors affecting loudness measurement and prediction

The current ANSI standard (ANSI S3.4, 2007) estimates the loudness of sustained sound, but many naturally occurring sounds have time-varying levels. Glasberg and Moore (2002) published a model for the prediction of time-varying sounds and this aspect should be considered as part of a revision to the ANSI standard. To assess the predictions of the dynamic model, loudness magnitude estimation functions were obtained for 24 listeners using pure tones (0.5 and 1.0 kHz), vowels, spondees, and speech-shaped noise (SSN) presented at levels from 40 to 90 dB SPL. Inferred equal-loudness levels from the fitted loudness functions were compared to the model predictions. The loudness model was qualitatively consistent with the behavioral data. The model predicted SSN to be louder than vowels and spondees which would be louder than tones; however, the model over-predicted the loudness differences. Possible explanations for the over-prediction include biases in loudness judgments, cognitive factors regarding learned expectations for loudness, assumptions regarding the free-field-to-headphone transfer function, and the model’s over-prediction of spectral, loudness summation, a result found recently (Schlittenlacher et al., 2014) for the ANSI standard and DIN 45631.

On time-varying loudness and nonlinearly propagated sound

Time varying loudness models have been found to be highly correlated with subject judgments for many sources of noise. Among these sources are impulsive sounds such as sonic booms, blast noise, and thunder. Aside from the short temporal duration, all of these sound sources are distinctive in that each is of high enough amplitude to have nonlinear propagation effects occur, which change the temporal and spectral character of a signal over distance. Because of this, it is possible that differences in how a model handles spectral content may result in differences in loudness predictions over distance. A series of simulated nonlinear sources were propagated and its time-varying loudness calculated from both the Moore & Glasberg and Zwicker model. These predictions will be compared to examine how these models differ when the spectral content of the signal changes. Other parameters, including historical level metrics and those derived from loudness time histories, such as the loudness time derivative, will also be discussed.

Differences between ANSI S3.4-2007 and the proposed ISO532-2

The ANSI standard for the calculation of loudness (ANSI S3.4-2007) is based on the model developed by Moore, Glasberg and co-workers, and uses the assumption that a diotic sound is twice as loud as that same sound presented monaurally. However, recent data suggest that loudness summation across ears is less than assumed in the ANSI standard. This was taken into account in the revised loudness model of Moore and Glasberg [J. Acoust. Soc. Am., 121, 1604–1612, 2007] using the concept of “binaural inhibition”, whereby the signal at one ear inhibits the internal response to a signal at the other ear. This model predicts that a diotic sound is 1.5 times as loud as that same sound presented monaurally, and it forms the basis for the proposed ISO532-2 standard. The revised model also gives reasonably accurate predictions of loudness in cases where the sounds differ across the two ears, for example, sounds recorded via a dummy head. An extension of the model to deal with time-varying sounds has been shown to give accurate predictions of loudness for a variety of technical and musical sounds. It is proposed that the revised and extended model be used as the basis for a new ANSI standard.