Use of a deep recurrent neural network to reduce transient noise: Effects on subjective speech intelligibility and comfort
Transient noise reduction using a deep recurrent neural network improves the subjective speech Intelligibility and comfort.
(A) We process signals with a binaural processing stage. The signal is converted to phoneme probabilities using a DNN. The degradation of these is used to predict binaural SI. (B) Results for subjective data, our model (BAPSI) and baseline models in terms of speech recognition threshold (SRT).
Binaural prediction of speech intelligibility based on a blind model using automatic phoneme recognition
In this study, we show that phoneme probabilities from a DNN can produce good estimates of speech intelligibility when combined with a blind binaural processing stage.
Estimated distortion component of hearing impairment from the model's prediction errors as a function of the average hearing loss.
Estimating the distortion component of hearing impairment from attenuation-based model predictions using machine learning
Attenuation-component-based model predictions of speech recognition thresholds like FADE seem to facilitate an estimation of the supra-threshold distortion component of hearing impairment.
Deep neural networks were trained to recognize spoken words from simulated auditory nerve representations. We measured effects of simulated OHC loss (broader frequency tuning and elevated thresholds) and ANF loss (reduced fidelity of temporal coding) on network speech recognition in noise.
Hearing-impaired artificial neural networks replicate speech recognition deficits of hearing-impaired humans
We developed a deep learning model of hearing loss by training artificial neural networks to recognize words in noise from simulated auditory nerve input.
A finite element model of the implanted cochlea, showing the voltage spread caused by stimulation of the implanted electrode.
Comparing phonemic information transmission with cochlear implants between human listeners and an end-to-end computational model of speech perception
A finite element model of a cochlea, a computational model of the auditory nerve, and an automatic speech recognition neural network were combined to replicate CI speech perception patterns.
A cochlear implants teenage user trying one of the BEARS games at one of the participatory design sessions we organised during the past two years.
Designing the BEARS (Both Ears) virtual reality training suite for improving spatial hearing abilities in teenage bilateral cochlear implantees
Teenagers with bilateral cochlear implants (CI) often suffer from poor spatial hearing abilities; a set of multi-modal (audio-visual) and multi-domain training tasks (localisation, spatial speech-in-noise and spatial music) was designed by involving teenage CI users as co-creators during the development process.
Graphic representation of how differences in PCA components are reflected in MAPs.
How variation in cochlear implant performance relates to differences in MAP parameters
Statistical analysis of how fitting parameters relate to speech recognition scores finds meaningful differences between the highest- and lowest-scoring tertiles of recipients.
dbTrack (Westone) technology
Sound-Level monitoring earphones with smartphone feedback as an intervention to promote healthy listening behaviors in young adults
In a first of its kind study, we aimed to determine the accuracy and reliability of sound-level monitoring earphones and the effect of smartphone feedback as an intervention to encourage safe listening use among young people.
Figure 1 A: Data streamed from the hearables to a PC running a recording script. B. Data from two different activities, recorded over 5 minutes each. C. Results from a classification using Naïve Bayes classification and a 5-fold cross-validation procedure.
Automatic detection of human activities from accelerometer sensors integrated in hearables
Using a Naive Bayes classifier, we could show that twelve different activities were classified above chance.
Example trial of the auditory Sternberg paradigm. The upper row shows the screen presented to the participant during each phase. The lower row illustrates the sound presented to the participant during each phase.
Assessing listening effort, using EEG and pupillometry, in response to adverse listening conditions and memory load.
This ongoing study aims to investigate how effortful listening becomes, when neurocognitive mechanisms need to be activated while listening to a distorted speech signal through a CI vocoder, using EEG and pupillometry during an auditory digit working memory task.
Schematic showing experimental set-up including the placement of the participant, loudspeakers and two observers.
A classification approach to listening effort: combining features from the pupil and cardiovascular system
Training k-nearest neighbor classifiers to predict intelligibility, social state and participant perception of a listening task
Figure 1. Stimulus spectra of standard (blue) and target (orange) vowel-like stimuli with ΔF2=2% and the corresponding model responses of AN fibers and IC band-enhanced cells (BE), which are excited by fluctuations within a band of modulation frequencies.
Modeling formant-frequency discrimination based on auditory-nerve and midbrain responses: normal hearing and sensorineural hearing loss
Formant-frequency difference limens in human with normal hearing or mild sensorineural hearing loss were estimated based on models for neural fluctuation profiles of neurons in the inferior colliculus.
VCN microcircuit structure of a pair of bushy cells. Green lines show the excitatory inputs originating from ANFs. Blue lines represent inhibitory inputs from DS cells while red lines are inhibitory inputs from TV cells. Yellow lines represent the gap junction connections between the bushy cells.
Modeling the effects of inhibition and gap junctions on synchrony enhancement in bushy cells of the ventral cochlear nucleus
This study explores the effects of inhibition and gap junctions on the synchrony enhancement seen in ventral cochlear nucleus bushy cells by using biophysically detailed neural network models of bushy cells microcircuits.
Top, SFIE model receiving input from a single frequency, bottom, a cross-frequency model receiving an array of excitatory and inhibitory inputs from multiple frequencies. Right, model responses to two opposite directions of frequency chirp.
A computational model of fast spectrotemporal chirp sensitivity in the inferior colliculus
We present a computational model of auditory nerve fiber responses elicited by combined electric and acoustic stimulation which can be used to investigate peripheral electric-acoustic interaction.
The model consists of a phenomenological model of acoustic stimulation (Bruce et al., 2018, green) and an integrate-and-fire model of electric stimulation (Joshi et al., 2017, orange). The model output are simulated spike times. Different coupling methods between both models were investigated (red).
A computational single-fiber model of electric-acoustic stimulation
We present a computational model of auditory nerve fiber responses elicited by combined electric and acoustic stimulation which can be used to investigate peripheral electric-acoustic interaction.
Predicting fusion of dichotic vowels in normal hearing listeners with a physiologically-based model
A psysiologically-based model can simulate some dichotic vowel fusion percepts in human listeners.
“Ear in the Clouds”– A web app supporting computational models for auditory-nerve and midbrain responses
A cloud-based web app provides an accessible tool for simulation and visualization of population responses of model auditory-nerve and midbrain neurons.
Results for V1 with both cochlear implant (CI) and hearing aid (HA) (top), CI only (middle) and HA only (bottom) in response to a set of 30 sentences extracted from Speech Banana auditory training app
Visualization of speech perception errors through phoneme alignment
We present an automated program for aligning stimulus-response phonemes collected from speech testing in order to visualize speech perception errors in individuals with hearing loss
Average scores for age-matched controls (C:AM), high-performing, blast-exposed SMs (B:Hi) and low-performing, blast-exposed SMs (B:Lo) groups. Figure panels show results for tests used to classify subjects, clinical audiometry metrics, and selective behavioral and electrophysiological metrics.
Functional hearing and communication deficits (FHCD) in blast-exposed service members with normal to near-normal hearing thresholds
Difficulties integrating binaural cues and understanding speech in noise amongst blast-exposed Service members with audiometric thresholds within clinical norms are most likely due to sensory, and not cognitive deficits, as indicated by a poorer signal-to-noise ratio in the neural encoding of sound in the peripheral auditory system.
Average thresholds for amplitude modulation detection task (N = 11) at different modulation rates (left: 16 Hz, middle: 32 Hz, right: 64 Hz). Carrier frequency: 500 Hz.
The effect of selective loss of auditory nerve fibers on temporal envelope processing: a simulation study
Computer simulation supports the hypothesis of cochlear synaptopathy selectively damaging low-spontaneous rate auditory nerves and impacting temporal envelope processing, and shows that some tasks are more sensitive to this issue than others.
A) Latency of the frequency following response to natural speech estimated by the complex linear model. B) Magnitudes of the model coefficients at 11 ms. C) Phases of the model coefficients at 11 ms.
Correlates of linguistic processing in the frequency following response to naturalistic speech
Our findings suggest that the frequency following response tracking the fundamental frequency of voiced speech plays an active role in the rapid and continuous processing of spoken language.
Cortical tracking of a distractor speaker modulates the comprehension of a target speaker
Cortical tracking of the ignored speaker plays a functional role in speech processing.
Using active inference to model selective attention during cocktail party listening
We introduce a new generative model of selective attention during cocktail party listening, and treat selective attention as an inference problem.
A) Flowchart of the "Meniere Calendar" application, B) Final psychometric function fitted to data from a normal hearing participant (left) and a Meniere's patient (right). C) Screenshots of the application's environment
Systematic monitoring of Meniere’s disease: A smartphone-based approach for the periodical assessment of audiometric measures and fluctuating symptoms
This study developed an application software that periodically records the course of Meniere's disease patients through the performance of automated, binaural audiometric tests.
Grand averaged auditory cortical evoked potential waveforms of the groups (Control/Blue; Tinnitus/Orange; Hyperacusis/Purple) generated by 500 Hz and 2000 Hz Narrow-Band Noise (NBN) stimuli across intensities.
Behavioral and electrophysiological evaluation of loudness growth in clinically normal hearing tinnitus patients with and without hyperacusis
The hyperacusis group differs significantly from the control and tinnitus groups in behavioral tests, but not in electrophysiological tests.
Topics found with Latent Dirichlet Allocation on 100k Reddit posts. The top rectangle of each block is the topic's name assessed through the algorithm's output: the lemmatised words shown in the middle rectangle. The proportion of messages that mention the topic is displayed in the lower rectangle.
What can we learn about tinnitus from social media posts?
Exploiting spontaneous messages of Reddit users discussing tinnitus, this work identifies the main topics of interest, their heterogeneity, and how they relate to one another based on cooccurrence in users' discussions; to enhance patient-centered support.
Outline of the RCT
Outcomes and experiences of delivering an internet-based intervention for tinnitus during the COVID-19 pandemic
A RCT was undertaken on a US populations indicating the value of an CBT internet-intervention for both reducing tinnitus distress, tinnitus comorbidities and managing the anxiety associated with the pandemic.