OBJECTIVE:To achieve a multi-dimensional characterisation of normal-hearing and hearing-impaired listeners' hearing abilities, including standard audiological conditions and virtual acoustic scenes for the evaluation of hearing aids. DESIGN:Within-subjects design, including speech intelligibility and listening effort measurements in two standard conditions and four virtual acoustic scenes, and assessments of loudness scaling, tone-in-noise detection thresholds, and HEAR-COMMAND tool questionnaire. STUDY SAMPLE:76 age-matched listeners, including 20 normal-hearing, 25 hearing-impaired without hearing aids, and 31 hearing aid users. RESULTS:Speech intelligibility and hearing aid benefit in virtual acoustic scenes fell between the results of the standard audiological conditions S0N0 and S0N90. The S0N0 and underground station environment were the most challenging conditions regarding speech intelligibility and listening effort. The pure-tone average explained most of the differences between the listener groups in loudness perception and tone-in-noise detection thresholds. Moderate to strong correlations were found between the HEAR-COMMAND tool speech scores and speech intelligibility. CONCLUSIONS:The study established a unique measurement database including complex virtual acoustic scenes and demonstrated a connection between speech intelligibility, hearing aid benefit, and other perceptual auditory measures. The database and findings provide a valuable foundation for advancing hearing aid evaluation and can serve as a benchmark in computational audiology.

OBJECTIVE:To propose a comprehensive audiological-vestibular test battery for characterising age-related hearing loss, enabling precise phenotyping of suprathreshold functional, physiological, and vestibular factors beyond audibility. To establish age- and audibility-dependent reference data in age-appropriate normal-hearing controls. DESIGN:Multidisciplinary consensus about test battery composition; Statistical analysis of centre effects to assess comparability of the test battery measured at two centres (Germany and France); Statistical analysis of age and pure-tone average (PTA) effects per test to identify potential covariates. SAMPLE:n = 55 (39 German and 16 French) participants with hearing thresholds better than the age-dependent median of the PTA, aged 40 years or older. RESULTS:Due to negligible centre effects, all data were pooled across centres. Age- and PTA-dependent reference data were derived. Age and PTA effects were identified for some tests, especially for audiological-functional tests. No age effects were found for vestibular tests. CONCLUSIONS:Normative values for a clinically feasible, multidimensional audiological-vestibular test battery were provided, including several measures whose age and PTA dependencies were previously unclear. Age and PTA should be considered as covariates for interpretation of these tests in future applications such as, e.g. phenotype-genotype relations in specified cohorts. Extensive data documentation and verification are essential for cross-centre comparability.

Objective: The aim is to introduce the concept of the Virtual Hearing Clinic (VHC), give an overview of the current status and exemplify its feasibility with data from a diagnostics module obtaining hearing thresholds. Design: The architecture of the VHC is described and an overview of functional modules that have been developed and tested in respective studies is given. As a functional example data from an experiment is presented. Hearing thresholds were obtained from 20 subjects with hearing loss with the VHC using the Graded Response Bracketing (GraBr) procedure and compared to reference thresholds obtained with a clinical audiometer. Results: Median VHC-based hearing thresholds over all frequencies did not differ significantly from the reference with values of 57.4 dB SPL (VHC) and 55.5 dB SPL (reference). The shape of the frequency-dependent thresholds was also found to be very similar for 250 Hz to 4 kHz. Results for 6 kHz showed larger differences. Conclusion: The VHC is suitable as a mobile hearing health application and to collect data for audiological research. By offering flexibility in time and location it can lower barriers for hearing health care and enable collecting large datasets that are needed for the advancement of data-driven audiology.

Abstract This study investigates the impact of uncalibrated consumer hardware on the outcomes of the German Matrix Sentence Test (GMST) in a remote setting. Speech-in-noise tests like the MST are relevant in audiology, as they better reflect everyday listening conditions compared to pure-tone audiometry. Due to their suprathreshold signal presentation and robustness against calibration offsets, they are particularly suitable for remote testing. It is unclear, however, if a sensitive and efficient test like the MST exhibits the same robustness against uncontrolled conditions in remote testing with uncalibrated consumer devices. This motivates the systematic assessment of the online MST’s reliability and its agreement with laboratory administration, conducted in normal-hearing participants using the browser-based platform “Virtual Hearing Clinic” (VHC). Twenty normal hearing participants completed the test in a randomized sequence: once in a lab with calibrated equipment and twice remotely via the VHC using uncalibrated consumer devices (smartphones, laptops, headphones). Results revealed no significant influence of headphone type, test environment, or pure-tone average. The remote condition demonstrated high test–retest reliability (r = 0.82) and strong correlation with lab-based measurements (r = 0.78). A mean bias of 0.29 dB in speech recognition thresholds was observed with slightly better performance in the lab, though not statistically significant (p = 0.15). Overall, results demonstrate that the MST, when implemented remotely and using uncalibrated consumer hardware, yields results of accuracy and reliability comparable to those obtained in the lab for normal hearing, supporting further applications of remote hearing assessments with optimized test materials.

OBJECTIVE: To enable reliable smartphone-based hearing assessments by developing methods to estimate device calibration offsets using categorical loudness scaling (CLS). DESIGN: Calibration offsets were simulated from a Gaussian distribution. Two prediction models-a Bayesian regression model and a nearest neighbour model-were trained on CLS-derived parameters and data from the Oldenburg Hearing Health Record (OHHR). CLS was chosen because it provides level-independent measures (e.g., dynamic range) that remain robust despite calibration errors. STUDY SAMPLE: The dataset comprised CLS results from N = 847 participants with a mean age of 70.0 years (SD = 8.7), including 556 male and 291 female listeners with diverse hearing profiles. RESULTS: The Bayesian regression model achieved median absolute errors (MAEs) of about 5 dB between the estimated and "true" calibration offsets. Calibration uncertainty was reduced by factors between 0.41 and 0.79, demonstrating greater robustness in uncontrolled environments. CONCLUSIONS: CLS-based models show potential to compensate for missing calibration in our simulation study, but validation using uncalibrated mobile-device listening tests with real listeners is still needed. This approach provides a practical alternative to threshold-based methods, supporting the use of smartphone-based tests outside laboratory settings and expanding access to reliable hearing healthcare in everyday and resource-limited contexts.

Processing delays can degrade listening experience, especially where direct and delayed sound paths interact. Although noticeability of delays has been modeled successfully, it is unclear how well models generalize across different types of delays. In this study, the noticeability of frequency-dependent and -independent delays was measured and the GPSMq model [Biberger, Fleßner, Huber, and Ewert (2018). J. Audio Eng. Soc. 66(7/8), 578–593] was tested for its ability to predict both existing and previously unreported data. A good prediction performance (linear correlation, ρ=0.94) was measured. The functional relationship between model outputs and measured results differs significantly between participant groups with different levels of experience but not across conditions.

Background In background noise, speakers adapt their speech production, giving rise to Lombard speech, which often improves speech intelligibility (SI). While intelligibility benefits of Lombard speech have been extensively studied in non-tonal languages, it remains unclear whether spectro-temporal cues, which are critical for tonal contrasts, are necessary to predict both the Lombard gain (LG) (i.e., intelligibility improvement relative to plain speech) and absolute SI in Mandarin Chinese. Methods Predictions of two SI-models were compared, namely, an automatic speech recognition (ASR)-based approach using spectral or spectro-temporal features and the speech intelligibility index (SII)-based model using spectral features. Predicted LG and absolute speech recognition threshold (SRT) values, for five female and six male speakers in stationary speech-shaped noise, were compared with empirical data. Results For both models, spectral features alone are sufficient for accurate prediction of the LG for both models. In contrast, predictions of absolute SRT were most accurate when spectro-temporal features were included, capturing substantial inter-speaker variability. Conclusions Despite the tonal nature of Mandarin, spectro-temporal features are not required to predict the LG. However, they are essential to predict the absolute SRTs, which vary across speakers.

Reliable hearing assessment at home can improve accessibility and reduce dependence on in-clinic testing. To be viable, home-based procedures must provide accurate results within short measurement times and remain robust to factors such as ambient noise and variable user attention. In this proof-of-concept study with a limited sample size, we evaluated two such procedures—a Graded Response Bracketing method (GRaBr) for pure-tone threshold estimation and a reinforced adaptive categorical loudness scaling method (rACALOS) for loudness-growth assessment - using remote, smartphone-based testing, and compared their outcomes with those obtained inside the laboratory using established reference procedures. Fifteen young adults with normal hearing completed the tasks at home and in the laboratory. Test–retest reliability was assessed by repeating the home measurements within one week of the initial session. Ambient noise levels in home environments were also recorded. The intraclass correlation coefficients for GRaBr measured at home exceeded 0.75, indicating good test–retest reliability. Similarly, home-based rACALOS showed generally high reliability, with across-run biases below 5 dB at all frequencies; however, mean interquartile ranges reached up to 18.4 dB at some loudness categories. Remote GRaBr measurements showed small, non-significant deviations from laboratory audiometry (0.4 ± 7.1 dB); however, frequency-specific biases of approximately ±5 dB were observed at 250 Hz and 4 kHz, with root-mean-square errors ranging from 4.6 to 8.0 dB across frequencies. Compared with the in-lab ACALOS, remote rACALOS showed an overall bias of 3.38 ± 11.8 dB across loudness categories, with no significant overall effect of test environment. These findings suggest that smartphone-based pure-tone audiometry and loudness-scaling assessments provide accurate and reliable results when using these procedures at home, at least in normal-hearing young adults, who are experienced loudness raters, under suitable acoustic conditions with low ambient noise.

OBJECTIVE:To address the calibration and procedural challenges inherent in remote audiogram assessment for rehabilitative audiology, this study investigated whether calibration-independent adaptive categorical loudness scaling (ACALOS) data can be used to approximate individual audiograms by classifying listeners into standard Bisgaard audiogram types using machine learning (ML). DESIGN:Three classes of ML approaches-unsupervised, supervised, and explainable-were evaluated. Principal component analysis (PCA) was performed to extract the first two principal components, which jointly explained more than 50% of the variance. Seven supervised multi-class ML classifiers were trained and compared, alongside unsupervised and explainable methods. STUDY SAMPLE:A large auditory reference database (n = 847 ears) containing ACALOS data was used for model development and evaluation. RESULTS:The factor map showed substantial overlap between listeners, indicating that cleanly separating participants into six Bisgaard classes based solely on their loudness patterns is challenging. Nevertheless, the ML models demonstrated reasonable classification performance. Among the supervised classifiers, logistic regression achieved the highest accuracy. In addition, the SHAP and feature permutation analyses showed that the highest predictive power of the ML models was attributable to the minimum loudness levels at 1.5 and 4 kHz. CONCLUSIONS:The findings demonstrate that ML models can predict standard Bisgaard audiogram types-within certain limits-from calibration-independent loudness perception data. This approach may support future hearing aid fitting in remote or resource-limited settings without requiring a traditional audiogram.

Assigning individuals with hearing impairment to auditory profiles can support a better understanding of the causes and consequences of hearing loss and facilitate profile-based hearing-aid fitting. However, the factors influencing auditory profile generation remain insufficiently understood, and existing profiling frameworks have rarely been compared systematically. This study therefore investigated the impact of two key factors-the clustering method and the number of profiles-on auditory profile generation. In addition, eight established auditory profiling frameworks were systematically reviewed and compared using intrinsic statistical measures and manifold learning techniques. Frameworks were evaluated with respect to internal consistency (i.e., grouping similar individuals) and cluster separation (i.e., clear differentiation between groups). To ensure comparability, all analyses were conducted on a common open-access dataset, the extended Oldenburg Hearing Health Record (OHHR), comprising 1,127 participants (mean age = 67.2 years, SD = 12.0). Results showed that both the clustering method and the chosen number of profiles substantially influenced the resulting auditory profiles. Among purely audiogram-based approaches, the Bisgaard auditory profiles demonstrated the strongest clustering performance, whereas audiometric phenotypes performed worst. Among frameworks incorporating supra-threshold information in addition to the audiogram, the Hearing4All auditory profiles achieved the lowest normalized Davies-Bouldin (DB) score, while the BEAR auditory profiles performed better on the other intrinsic measures. In conclusion, separability should be considered a primary criterion in auditory profile generation, as it directly determines how meaningfully different profiles can be distinguished in practice. Manifold learning and intrinsic measures enable systematic comparisons of auditory profiling frameworks and identify the Hearing4All auditory profile as a promising approach for future research.

Human walking can be modeled using a springmass-damping (SMD) system. While most studies have focused on standard activities in healthy young to middle-aged populations, less attention has been given to participants' responses to unexpected gait perturbations. These responses may be valuable predictors of falls, particularly in older adults. Our previous study modeled walking in the vertical direction for a diverse group of participants. Since most gait perturbations occur in the medio-lateral and anteriorposterior directions, the next step involves extending the analysis to these directions and treadmill walking between perturbations.The study included 60 adults (aged 18-87 years), who walked on a perturbation treadmill while wearing an inertial measurement unit (IMU) at the lumbar region to capture body acceleration. Participants first walked at their preferred speed on the treadmill ("normal gait data"), followed by "perturbation trials" with gait perturbations. The gait data between the perturbations were analyzed as "inter-perturbation gait data". Force data was recorded using the treadmill's built-in force plates. The SMD model was applied to calculate damping and stiffness coefficients.The lowest median spring stiffness was found in medio-lateral direction and the highest in anterior-posterior direction. The lowest damping coefficient was found in medio-lateral direction and the highest in vertical direction. Compared to "normal gait data", "inter-perturbation gait data" showed higher stiffness and, for some participants, higher damping coefficients, while others exhibited decreased damping. Damping and stiffness coefficients were successfully extracted from treadmill walking data across all directions for a diverse group of participants and linked to gait dynamics. The analysis highlighted human gait adaptability under various conditions. This study provides groundwork for future research on individual responses to unexpected gait perturbations.Clinical relevance- Describing human walking with damping and stiffness coefficients in different directions could contribute to understand reactive dynamic balance, and thus give a sound estimation of a relevant risk factor for falls in older people.

Virtual acoustics enables hearing research and audiology in ecologically relevant and realistic acoustic environments, while offering experimental control and reproducibility of classical psychoacoustics and speech intelligibility tests. Hereby, indoor environments are highly relevant, where listening and speech communication frequently involve multiple targets and interferers, as well as connected adjacent spaces that may create challenging acoustics. Hence, a controllable laboratory environment is evaluated here (by room acoustical parameters and speech intelligibility) which closely resembles a typical German living room with an adjacent kitchen. Target and interferer positions were permuted over four different locations, including an acoustically challenging position of a target in the kitchen with interrupted line of sight. Speech intelligibility was compared in the real room, in virtual acoustic representations, and in standard anechoic audiological configurations. Three presentation modes were tested: headphones, loudspeaker rendering on a small-scale, four-channel loudspeaker array in a sound-attenuated listening booth, and a three-dimensional 86-channel loudspeaker array in an anechoic chamber. The results showed that the target talker in the coupled room requires higher signal to noise ratios (SNRs) at threshold than typical indoor conditions. Moreover, for the stationary speech shaped interferer, effects of room acoustics were negligible. For a majority of target positions, no difference between the four-channel and the large-scale loudspeaker array were found, with an overall good agreement to the real room. This indicates that ecologically valid testing is feasible using a clinically applicable small-scale loudspeaker array.

The gap between hearing abilities measured with standard audiological tests, such as speech audiometry, and perceived hearing performance in daily life has led to interest in testing methods that better reflect real-world communication. Traditional speech audiometry evaluates the proportion of correctly understood prerecorded words, which differs significantly from spontaneous, interactive speech in everyday conversations where listeners actively engage with interlocutors. This study examines how background noise affects communication abilities and subjective effort in hearing-impaired (HI) individuals compared to age-matched normal-hearing (NH) individuals. Thirty HI participants and ten NH participants completed the “classic” OLSA matrix sentence test to measure speech intelligibility and the Diapix communication task with varying levels of background noise, where pairs of participants solved tasks through conversation. The study explores how noise influences objective communication performance, subjective effort, and acoustic speech parameters, comparing differences between HI and NH groups. A degradation of communication efficiency for HI individuals is observed, while NH participants are less affected by increasing noise levels. Both groups adjust their speech to cope with challenging conditions. Results are discussed in the context of ecologically valid testing paradigms to bridge the gap between clinical test outcomes and real-world challenges faced by people with hearing impairments.

Radars are promising tools for contactless vital sign monitoring. As a screening device, radars could supplement polysomnography, the gold standard in sleep medicine. When the radar is placed lateral to the person, vital signs can be extracted simultaneously from multiple body parts. Here, we present a method to select every available breathing and heartbeat signal, instead of selecting only one optimal signal. Using multiple concurrent signals can enhance vital rate robustness and accuracy. We built an algorithm based on persistence diagrams, a modern tool for time series analysis from the field of topological data analysis. Multiple criteria were evaluated on the persistence diagrams to detect breathing and heartbeat signals. We tested the feasibility of the method on simultaneous overnight radar and polysomnography recordings from six healthy participants. Compared against single bin selection, multiple selection lead to improved accuracy for both breathing (mean absolute error: 0.29 vs. 0.20 breaths per minute) and heart rate (mean absolute error: 1.97 vs. 0.66 beats per minute). Additionally, fewer artifactual segments were selected. Furthermore, the distribution of chosen vital signs along the body aligned with basic physiological assumptions. In conclusion, contactless vital sign monitoring could benefit from the improved accuracy achieved by multiple selection. The distribution of vital signs along the body could provide additional information for sleep monitoring.

Individuals have different preferences for setting hearing aid (HA) algorithms that reduce ambient noise but introduce signal distortions. "Noise haters" prefer greater noise reduction, even at the expense of signal quality. "Distortion haters" accept higher noise levels to avoid signal distortion. These preferences have so far been assumed to be stable over time, and individuals were classified on the basis of these stable, trait scores. However, the question remains as to how stable individual listening preferences are and whether day-to-day state-related variability needs to be considered as further criterion for classification. We designed a mobile task to measure noise-distortion preferences over 2 weeks in an ecological momentary assessment study with N = 185 (106 f, Mage = 63.1, SDage = 6.5) individuals. Latent State-Trait Autoregressive (LST-AR) modeling was used to assess stability and dynamics of individual listening preferences for signals simulating the effects of noise reduction algorithms, presented in a web browser app. The analysis revealed a significant amount of state-related variance. The model has been extended to mixture LST-AR model for data-driven classification, taking into account state and trait components of listening preferences. In addition to successful identification of noise haters, distortion haters and a third intermediate class based on longitudinal, outside-of-the-lab data, we further differentiated individuals with different degrees of variability in listening preferences. Individualization of HA fitting could be improved by assessing individual preferences along the noise-distortion trade-off, and the day-to-day variability of these preferences needs to be taken into account for some individuals more than others.

The discrepancy between the hearing aid benefit estimated in standard audiological tests, like speech audiometry, and the perceived benefit in daily life has led to interest in methods better reflecting real-world performance. In contrast to audiological tests, everyday communication commonly takes place in enclosed spaces with acoustic reflections and multiple sound sources, including sounds from adjoining rooms through open doors. This study investigates speech recognition thresholds (SRTs) with a sentence test in a laboratory environment resembling an average German living room with an adjacent kitchen. Additionally, acoustic simulations of the environment were presented in a large-scale (86) and small-scale (4) loudspeaker array, with the latter feasible for a clinical context. Measurements with normal-hearing and hearing-impaired listeners were conducted using different spatial target positions and a fixed masker position. One of the target positions was within the adjacent kitchen without line-of-sight to the sound source, representing a challenging acoustic configuration. Hearing-impaired listeners performed the measurements with and without their hearing aids. SRTs were compared between different presentation settings and to those measured in standard free-field audiological spatial configurations (S0N0, S0N90). An auditory model was employed for further analysis. Results show that SRTs in the simulated living room environment with 86 and 4 loudspeakers matched the real environment, even for aided listeners, indicating that virtual acoustics representations can reflect real-world listening performance. When signal-to-noise ratios were normalized, the measured hearing aid benefit did not differ significantly between the standard audiological spatial configuration S0N90 and any spatial configuration in the living room environment.

Hearing health is shaped by both measurable auditory function and the perceived ability to navigate daily life. To fully understand its complexities, it is essential to integrate objective assessments behavioural tests that quantify hearing acuity and functional performance with subjective reports on how individuals navigate and manage their hearing in everyday life. The Oldenburg Hearing Health Repository (OHHR) has been developed to unite these perspectives, providing a comprehensive dataset on hearing health. Collected between 2013 and 2015 at the Hoerzentrum Oldenburg in collaboration with Hearing4all, OHHR includes data from 581 individuals (aged 18 – 86 years; 255 females; mean age = 67.31 years; SD = 11.93 ) with varying degrees of hearing loss. This publicly accessible dataset combines audiometric tests (Pure Tone Audiometry, Loudness Scaling, Speech in Noise tests) with self-reports on hearing difficulties, lifestyle, technology use, and cognitive assessments (DemTect, Vocabulary size test). By integrating subjective experience with objective measures, OHHR will enable researchers to explore the links between hearing ability, cognition, and quality of life, providing valuable insights to advance precision medicine. ### Competing Interest Statement The authors have declared no competing interest. ### Funding Statement This research has been funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Germany's Excellence Strategy - EXC 2177/1 - Project ID 390895286. ### Author Declarations I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained. Yes The details of the IRB/oversight body that provided approval or exemption for the research described are given below: Data protection approval for this study was obtained from the Data Protection and Information Security Management Office at Carl von Ossietzky Universitaet Oldenburg (Application Number: DSM-H4A Open Dataset/20241113-0009). All procedures comply with General Data Protection Regulation. Informed consent was collected from participants whose contact information was available (40%). For the remaining 60%, whose pseudonymized data precluded re-contact, a consent waiver was granted by the board. Additionally, a k-anonymity process (k=4) was applied to further minimize re-identification risk prior to data publication. This process is only described here; the manuscript includes information about the k-anonymity procedure and the approval from the data protection officer in the Methods and Technical Validation sections. The local ethics committee at the Carl von Ossietzky Universität Oldenburg reviewed and authorized the data collection I confirm that all necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived, and that any patient/participant/sample identifiers included were not known to anyone (e.g., hospital staff, patients or participants themselves) outside the research group so cannot be used to identify individuals. Yes I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance). Yes I have followed all appropriate research reporting guidelines, such as any relevant EQUATOR Network research reporting checklist(s) and other pertinent material, if applicable. Yes All data collected in this study and described in the preprint is available for open access at Zenodo under the "Hearing4all" Community.

Adaptive tracking procedures in psychophysics may produce erroneous, "untypical" results and non-converging tracks due to, e.g., inattention of the test subject or external disturbances. This paper presents a multi-state psychometric model, which is used to rate the outcome of psychometric measurement procedures with a consistency measure. The consistency measure may be used for a post hoc, automated consistency estimation for any psychometric measurement procedure that can be modeled with a sigmoid psychometric function. The model calculates the log likelihood difference between single and two interleaved psychometric functions, potentially underlying a recorded adaptive track. A binary classifier was tested with a range of candidates for consistency measures with simulated, inconsistent tracks, and expert ratings of empirical tracks. The proposed consistency measure was identified as the best candidate to classify inconsistent tracks, while expert ratings were best predicted with the spectrum of the stimulus level, which is shown to be a suboptimal predictor of consistency. A threshold of the proposed measure for the German matrix sentence test is 10 to test for inconsistency, with a sensitivity of 60% and a specificity of 80%.

Processing delays can negatively affect listening experience, especially in cases where the (processed) delayed sound interferes with an un-delayed (or direct) sound component, as it is the case for (open-fit) hearing devices. In this study, psychometric functions for delay perception in individual frequency bands were measured. Also, it was assessed how noticeability adds up across frequency bands for frequency-dependent processing delays. Noticeability of delays depends largely on the phase shift they introduce in each frequency band. Psychometric functions are non-monotonous, with maximum noticeability at phase shifts (2n+1)π. When using the sensitivity index d' to describe the noticeability of a delay, the overall noticeability dtotal' of a frequency-dependent delay was found to be the RMS of the noticeabilities dn' in each frequency band n. Additionally, different auditory models were tested regarding their capability to predict the experimental results. The audio quality model GPSMq [Biberger et al. (2018). J. Audio Eng. Soc. 66(7), 578-593] showed the best performance for the majority of conditions, yielding predictions that are highly correlated (ρ>0.85) with the participants' results. Model performance could confirm that delays are mainly perceived based on spectral effects.