.
Routine occupational audiometry combined with machine-learning models can stratify airport catering workers by high-frequency hearing risk and support targeted hearing-conservation strategies.
Despite its high reliability, the Audiometric Weber Test demonstrates insufficient clinical validity, suggesting it should not be recommended as a primary diagnostic tool for verifying hearing thresholds in routine practice.
This study investigates the effect of hearing acuity on daily-life conversation characteristics, detected using a machine learning algorithm trained on acoustic features and Ecological Momentary Assessment data.
A neural network trained on physiologically simulated data can decode cochlear damage profiles from non-invasive recordings, even in ears with identical audiograms.
This study investigated whether 7–9-month-old infants process degraded audio-visual speech similarly to adults; behavioural results showed that infants significantly increased looking time toward the speaker’s mouth when speech was degraded, suggesting an early compensatory strategy, while analyses of the fNIRS and EEG data are ongoing to determine whether adult-like adaptations are also observed at the neural level.
This study established that Virtual Acoustic Space Identification (VASI) and the spatial domain of the SSQ scale are highly accurate tools for diagnosing spatial hearing deficits in individuals with sensorineural hearing loss, with VASI recommended as the primary objective clinical measure.
Older adults with mild to moderate hearing loss showed robust TRF even in presence of noise indicating their ability to listen in noise and comprehend speech.
This study compares perceived and objective time in audiometric testing using a fast Bayesian method (BAL) and the conventional Hughson-Westlake (HW) method in young musicians, examining the impact of stimulus structure and anticipation. Results show that HW leads to an underestimation of perceived duration, whereas BAL does not. We investigate the sequence of tones generated by each protocol as the output of a dynamical system, whose temporal predictability properties are analyzed.
NEMA+ Watch extends ecological momentary assessment to smartwatches, enabling low-burden, user-initiated reporting of real-world listening experiences alongside contextual and physiological data.
A closed-loop framework combines auditory modeling and deep learning to learn cochlear implant stimulation patterns that better match model-based normal-hearing neural responses.
A machine learning model optimized for everyday hearing tasks exhibits human-like psychoacoustic thresholds, suggesting perceptual sensitivity is limited by efficient representations of sound rather than internal noise.
A deep learning model trained on 120,000+ consumer ratings of commercial hearing aids predicts listener-rated ease of speech understanding from audio with high accuracy on held-out devices.
A comparison of the performance of current industry speech recognition systems to humans on many types of distorted speech revealed that industry speech recognition systems had high correlations with humans but were superhuman in some cases.
This study finds that the prevalence of tinnitus among school-aged adolescents in the southern region of India is 8.22%. The severity of tinnitus is notably affected by its frequency of occurrence and perceived loudness, but not by factors such as hearing thresholds, age, or gender.
DeepFS4 introduces a novel end-to-end deep learning architecture that integrates speech enhancement directly into the FS4 sound coding strategy by applying noise reduction on the temporal envelope while maintaining the temporal fine structure for the most apical channels, significantly improving speech intelligibility in noise for cochlear implant users without adding latency.
This project investigates how adults with hearing aids perceive voice cues and vocal emotions, compared to adults without hearing aids.
This study investigated how different auditory maskers and visual conditions affect audio-visual speech recognition.
This study applies traditional statistical analysis and machine learning to evaluate Frequency Following Response (FFR) patterns in children with school difficulties or stroke.
Alignment of measured DPOAE growth functions with simulated level maps enables quantitative inference of cochlear amplifier gain and reveals pronounced gain-dependent changes in DPOAE responses.
People who stutter show reduced auditory detection under cognitive load, suggesting increased listening effort and altered resource allocation during auditory processing.
Hearing Data, the tool that unlocks the potential of the Noah database to enhance patient care and clinical research.
Replacing old PTA-4 averages with R-based algorithms for high-precision audiology.
A computational modeling framework was designed to estimate electrocochleography measurements in response to acoustic and electric stimulation, which has been applied in an acoustic-only and an electric-acoustic stimulation experiment.
We present a UMAP-based Hearing Loss Map that identifies novel sensorineural hearing loss phenotypes and longitudinal trajectories from a large clinical dataset, going beyond conventional audiogram-based classification.
Regression modelling reveals a statistically significant contribution of interaural cue distortion to perceived music quality with simulated hearing aids and hearing loss.
This study uses efficient coding to explain how hidden hearing loss reorganizes adaptive auditory-midbrain representations in ways not captured by threshold audiometry or Fisher information alone.
Contralateral sound attenuation significantly improves speech intelligibility for hearing-aid users and normal-hearing listeners in "cocktail party" listening environments.
Individuals with misophonia demonstrated a significantly delayed FFR offset peak, suggesting abnormal subcortical auditory encoding that may underlie their heightened sensitivity to trigger sounds.
Extended training with synchronized temporal fine-structure cues produces rate-dependent improvements in interaural timing sensitivity in bilateral cochlear implant users.
This perspective highlights emerging research directions to improve the ecological validity of hearing assessment by bridging laboratory testing and real-world listening conditions.
During CI rehabilitation, M/C-levels reach stability around 3 months, whereas T-levels stabilize at approximately 5 weeks post-activation, for both Advanced Bionics™ and Cochlear™ implants users.
This presentation examines the design and integration of a real-time edge-AI neural denoiser for hearing aids, highlighting the challenges of meeting strict latency and computational constraints while improving speech intelligibility, preserving speech quality.
The Panoramic ECAP Method is accessible online (now with 35+ users) to allow researchers and clinicians alike to collect the relevant Electrically Evoked Compound Action Potentials and subsequently extract patient-specific estimates of current spread and neural responsiveness for cochlear-implant users.
Computational closed-loop hearing compensation can personalize signal processing for both standard and hidden hearing loss, improving objective intelligibility outcomes while preserving perceived listening quality.