MSMichael Anthony Stone

Michael Anthony Stone

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Michael Anthony Stone has a total of 112 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Hearing Impairment Assessment in Infants Through Explainable Computer Vision Analysis of Facial Features

    by Samuele Pe, Anisa Visram, Iain Jackson, Michael Stone, Enea Parimbelli, Kevin Munro, Arianna Dagliati


  • Open Access

    Remote audiogram and notched-noise tests using Gaussian processes

    by Emanuele Perugia, Michael A. Stone, Karolina Kluk, Josef Schlittenlacher

    The study aimed to develop and validate remote hearing detection (audiogram) and discrimination (notched-noise) tests, which used Bayesian active-learning based Gaussian processes, as a binary discriminator between audible and inaudible sounds, and to choose the stimulus parameters. Forty-two participants (aged 48–75 years) performed the two tests remotely using their own equipment. The participants were recruited with pure-tone audiometry known a priori. The agreements between the true and estimated hearing thresholds were assessed using Bland–Altman plots and concordance correlation coefficients (CCC). The notched-noise test was used to derive auditory filter shape parameters, from which equivalent rectangular bandwidths (ERBs) were estimated. The ERBs were correlated with participants' hearing thresholds and compared to published data. The agreement between the true and estimated hearing thresholds ranged between poor and fair (CCC = 0.19 and 0.34), with an average bias (±limit of agreements) of 27.3 (±17.8) dB. This was attributed to the lack of calibration of the participants' equipment. The estimated ERBs followed the trend of published data and were significantly correlated with pure-tone audiometry at 1000 Hz (rs = 0.36). Our findings indicate that the remote detection and discrimination tests can collectively track the breakpoints, slopes of thresholds, and the width of auditory filters.

  • Open Access

    Leveraging Monaural Exposures to Reveal Early Effects of Noise: Evidence from Police Radio Ear-Piece Use

    by Hannah Guest, Paul Elliott, Martie van Tongeren, Joseph Laycock, Steven Thorley-Lawson, Michael A. Stone, Michael T. Loughran, Christopher J. Plack

    Research into the long-term effects of noise on hearing is often confounded by health and lifestyle differences between individuals. UK police radio ear-pieces are capable of emitting high sound levels and, crucially, are worn in one ear, allowing between-ear comparisons which control for individual-level confounding factors. Low volume-control settings are recommended to reduce risk to police hearing, yet actual usage patterns and auditory effects remain unexamined. This study used a large-scale survey ( N = 4,498) to assess ear-piece noise exposure and the associated hearing health. Most participants reported using high volume-control settings and 45.2% reported experiencing signs of temporary threshold shift (TTS) in the exposed ear. Estimated weekly-averaged noise exposures frequently exceeded the UK's 85 dBA Upper Exposure Action Value. Ear-piece use was associated with 73% (95% confidence interval [CI] 46–106%) increased risk of persistent tinnitus, which on mediation analysis appeared to be driven by a subset of users who experienced signs of TTS. Importantly, tinnitus location was associated with the side of exposure, suggesting tinnitus related to device use rather than to other factors. In contrast, Digits-In-Noise thresholds showed no relation with noise exposure; potential explanations include compensatory auditory training effects, but limitations of Digits-In-Noise data must also be considered. Findings highlight a need for further investigation into hearing risks in police personnel, including in-person auditory testing. Risk mitigation strategies might involve improved device design, training on safe use, and expanded hearing health surveillance. Given the potential for cumulative auditory damage, TTS may serve as an early warning sign, warranting attention in broader noise-exposed populations.


  • Open Access

    Automated Question-and-Answer-Based Speech-In-Noise Hearing Test Using a Large Language Model

    by Mohsen Fatehifar, Kevin J. Munro, Michael A. Stone, David Wong, Tim Cootes, Josef Schlittenlacher

    A conversational speech-in-noise test administered without human supervision is desirable to reduce the need to attend the clinic. This study aimed to create and evaluate a novel question-and-answer-based speech-in-noise test called ‘CHAT’, where stimuli were synthesised using text-to-speech (TTS), responses captured using automatic speech recognition (ASR), and responses evaluated using a large language model (LLM). Three tests were compared: the newly developed CHAT, the Adaptive Sentence List (ASL) and the Four Alternative Auditory Feature Test (FAAF). CHAT used synthetic speech while ASL and FAAF by default used human-recorded speech. Speech reception thresholds (SRT) of 20 participants with normal hearing and 20 with mild to moderate hearing loss were measured using these three tests. Validity was evaluated using receiver operating characteristic (ROC) analysis and correlations between SRTs and average pure-tone audiometry (PTA), across the three speech tests. Reliability was evaluated by comparing limits-of-agreement between SRTs achieved by repeated runs of CHAT and ASL tests using Bland-Altman analysis. Results revealed limits-of-agreement between repeated runs of CHAT and ASL were ±3.6 dB and ±3.1 dB, respectively. The area under the ROC curve was 0.97 for CHAT, 0.95 for ASL and 0.89 for FAAF. The correlation between SRT and PTA results was 0.86 for CHAT, 0.82 for ASL and 0.79 for FAAF. These findings indicate that CHAT demonstrates reliability and validity comparable to human-recorded ASL and FAAF, supporting the use of synthetic speech for speech-in-noise testing. Overall, the results suggest that AI tools can help create and administer conversational speech-in-noise tests.

  • Open Access

    Digits-In-Noise Hearing Test Using Text-to-Speech and Automatic Speech Recognition: Proof-of-Concept Study

    by Mohsen Fatehifar, Kevin J. Munro, Michael A. Stone, David Wong, Tim Cootes, Josef Schlittenlacher

    This proof-of-concept study evaluated the implementation of a digits-in-noise test we call the ‘AI-powered test’ that used text-to-speech (TTS) and automatic speech recognition (ASR). Two other digits-in-noise tests formed the baselines for comparison: the ‘keyboard-based test’ which used the same configurations as the AI-powered test, and the ‘independent test’, a third-party-sourced test not modified by us. The validity of the AI-powered test was evaluated by measuring its difference from the independent test and comparing it with the baseline, which was the difference between the Keyboard-based test and the Independent test. The reliability of the AI-powered test was measured by comparing the similarity of two runs of this test and the Independent test. The study involved 31 participants: 10 with hearing loss and 21 with normal-hearing. Achieved mean bias and limits-of-agreement showed that the agreement between the AI-powered test and the independent test (−1.3 ± 4.9 dB) was similar to the agreement between the keyboard-based test and the Independent test (−0.2 ± 4.4 dB), indicating that the addition of TTS and ASR did not have a negative impact. The AI-powered test had a reliability of −1.0 ± 5.7 dB, which was poorer than the baseline reliability (−0.4 ± 3.8 dB), but this was improved to −0.9 ± 3.8 dB when outliers were removed, showing that low-error ASR (as shown with the Whisper model) makes the test as reliable as independent tests. These findings suggest that a digits-in-noise test using synthetic stimuli and automatic speech recognition is a viable alternative to traditional tests and could have real-world applications.


  • Online Arabic and English Digits-in-Noise Tests: Effects of Test Language and At-Home Testing

    by Adnan Shehabi, Christopher J. Plack, Garreth Prendergast, Kevin J. Munro, Michael A. Stone, Joseph Laycock, Arwa AlJasser, Hannah Guest

    Purpose: The Digits-in-Noise (DIN) test is used widely in research and, increasingly, in remote hearing screening. The reported study aimed to provide basic evaluation data for browser-based DIN software, which allows remote testing without installation of an app. It investigated the effects of test language (Arabic vs. English) and test environment (lab vs. home) on DIN thresholds and test–retest reliability. It also examined the effects of test language on the correlations between DIN and audiometric thresholds. Method: Fifty-two bilingual adults with normal hearing aged 18–35 years completed Arabic and English diotic DIN tests (two sessions in the lab and two sessions at home via the web). Effects of language and environment on DIN thresholds were assessed via paired t tests, while intraclass and Pearson's/Spearman's correlation coefficients quantified test–retest reliability and relations to audiometric thresholds. Results: DIN thresholds were 0.74 dB higher (worse) for Arabic than English stimuli. Thresholds were 0.52 dB lower in the lab than at home, but the effect was not significant after correction for multiple comparisons. Intraclass and Pearson's correlation coefficients were too low for meaningful analysis due to the use of a normal-hearing sample with low between-subject variability in DIN and audiometric thresholds. However, exploratory analysis showed that absolute test–retest differences were low (< 1.2 dB, on average) for both languages and both test environments. Conclusions: Arabic DIN thresholds were a little higher than English thresholds for the same listeners. Employing home-based rather than lab-based testing may slightly elevate DIN thresholds, but the effect was marginal. Nonetheless, both factors should be considered when interpreting DIN data. Test–retest differences were low for both languages and environments. To support hearing screening, subsequent research in audiometrically diverse listeners is required, testing the reliability of DIN thresholds and relations to hearing loss.


  • Open Access

    The influence of age and hearing loss on thresholds measured using the TFS-AF test

    by Emanuele Perugia, Melanie Lough, Michael A. Stone

    The upper limit of sensitivity to binaural temporal fine structure (TFS) is around 1400 Hz. Higher values have been reported when keeping interaural phase differences fixed and varying the frequency adaptively, as in the adaptive-frequency TFS test [TFS-AF, Füllgrabe et al., Int J Audiol, 56, 926-935]. The probability of TFS-AF thresholds exceeding 1400 Hz, and the influence of age and hearing threshold were assessed. TFS-AF thresholds from 277 participants across five studies were re-analysed as a function of Age, low-frequency pure-tone average threshold (PTL; 250, 500, 1000 and 2000 Hz), and high-frequency pure-tone average threshold (PTH; 6000 and 8000 Hz). The 95% confidence intervals on TFS-AF thresholds were computed. The effects of Age and Hearing threshold on the TFS-AF thresholds were assessed using a multiple-regression model. The bootstrapped 95% confidence interval for this dataset around the median was 1030 to 1122 Hz, similar to that predicted by Monte Carlo simulation. This range was affected by Age and PTL. Regression modelling showed that the occurrence rate of TFS-AF thresholds exceeding 1400 Hz increased with younger Age and better PTL. TFS-AF thresholds exceeding 1400 Hz are more likely for lower ages and better hearing thresholds.

  • Open Access

    Optimisation of visual reinforcement audiometry: a scoping review

    by Anisa S. Visram, Iain R. Jackson, Ibrahim Almufarrij, Michael A. Stone, Kevin J. Munro

    Visual reinforcement audiometry (VRA) is a well-established behavioural test used to assess hearing in infants and young children. This scoping review aimed to summarise the evidence for different approaches to optimising and improving the effectiveness of VRA for clinical practice. A pre-registered scoping review was conducted. Fifty-nine original articles were included in the review. The review identified a number of factors which improved response behaviour, such as increased variety and complexity of visual reinforcers, short reinforcer durations, and providing breaks. Intermittent conditioning, where as few as 50% of conditioning trials were rewarded, did not have an impact on response behaviour, and neither did the (suprathreshold) presentation level used during conditioning. More responses were achieved for younger (around 12 months) than older (around 18-24 months) infants. Once infants were developmentally ready to condition to play audiometry, this allowed for a more comprehensive hearing evaluation. VRA is a successful behavioural hearing test for most infants of developmental age around 7-24 months, with well-established protocols describing its clinical implementation. Further evidence is needed to assess potential benefits of different reinforcers, different auditory stimuli (e.g. filtered familiar sounds), and technologies to assist response detection.


  • Open Access

    The impact of face coverings on audio-visual contributions to communication with conversational speech

    by I. R. Jackson, E. Perugia, M. A. Stone, G. H. Saunders

    The use of face coverings can make communication more difficult by removing access to visual cues as well as affecting the physical transmission of speech sounds. This study aimed to assess the independent and combined contributions of visual and auditory cues to impaired communication when using face coverings. In an online task, 150 participants rated videos of natural conversation along three dimensions: (1) how much they could follow, (2) how much effort was required, and (3) the clarity of the speech. Visual and audio variables were independently manipulated in each video, so that the same video could be presented with or without a superimposed surgical-style mask, accompanied by one of four audio conditions (either unfiltered audio, or audio-filtered to simulate the attenuation associated with a surgical mask, an FFP3 mask, or a visor). Hypotheses and analyses were pre-registered. Both the audio and visual variables had a statistically significant negative impact across all three dimensions. Whether or not talkers’ faces were visible made the largest contribution to participants’ ratings. The study identifies a degree of attenuation whose negative effects can be overcome by the restoration of visual cues. The significant effects observed in this nominally low-demand task (speech in quiet) highlight the importance of the visual and audio cues in everyday life and that their consideration should be included in future face mask designs.


  • Open Access

    Toward a Real-World Technical Test Battery for Remote Microphone Systems Used with Hearing Prostheses

    by Michael A Stone, Melanie Lough, Keith Wilbraham, Helen Whiston, Harvey Dillon

    Remote microphones (RMs) enable clearer reception of speech than would be normally achievable when relying on the acoustic sound field at the listener's ear (Hawkins, J Sp Hear Disord 49, 409–418, 1984). They are used in a wide range of environments, with one example being for children in educational settings. The international standards defining the assessment methods of the technical performance of RMs rely on free-field (anechoic) delivery, a rarely met acoustic scenario. Although some work has been offered on more real-world testing (Husstedt et al., Int J Audiol 61, 34–45. 2022), the area remains under-investigated. The electroacoustic performance of five RMs in a low-reverberation room was compared in order to assess just the RM link, rather than measurements at the end of the signal chain, for example, speech intelligibility in human observers. It pilots physical- and electro-acoustic measures to characterize the performance of RMs. The measures are based on those found in the IEC 60118 standards relating to hearing aids, but modified for diffuse-field delivery, as well as adaptive signal processing. Speech intelligibility and quality are assessed by computer models. Noise bands were often processed into irrelevance by adaptive systems that could not be deactivated. Speech-related signals were more successful. The five RMs achieved similar levels of good predicted intelligibility, for each of two background noise levels. The main difference observed was in the transmission delay between microphone and ear. This ranged between 40 and 50 ms in two of the systems, on the upper edge of acceptability necessary for audio-visual synchrony.