Purpose: The primary purpose of this study was to examine the efficacy and the effectiveness of Starkey Livio Artificial Intelligence hearing aids in tracking step count. A secondary purpose was to investigate the accuracy of the fall detection and alert system of Livio hearing aids in detecting fall maneuvers. Method: A participant wore Binaural Starkey Livio receiver-in-the-canal style hearing aids, a Sportline pedometer, and a Fitbit Charge 3 concurrently during both real-world and treadmill walking conditions. The real-world condition was conducted over a 5-day period. Step count for the treadmill protocol was assessed at six different treadmill speeds (2 mph, 2.5 mph, 3 mph, 3.5 mph, 4 mph, 4.5 mph, and 5 mph). The fall detection and alert system were assessed through falling maneuvers of activities of daily living. Results: In the real-world condition, Livio, Sportline, and Fitbit recorded steps within 1 SD of each other. In addition, Livio recorded the most accurate steps compared to actual physical steps taken. In the treadmill condition, Livio recorded the least number of steps across all walking paces compared to the rest of the devices. Also, Livio hearing aids detected majority of the engaged falling maneuvers. Conclusions: The Livio was found to be feasible, consistent, and sensitive in detecting steps and falls. Further research of higher sample size and recruitment of individuals with hearing loss are suggested.

Objective measures of speech quality are highly desirable in benchmarking and monitoring the performance of hearing aids (HAs). Existing HA speech quality indices such as the hearing aid speech quality index (HASQI) are intrusive in that they require a properly time-aligned and frequency-shaped reference signal to predict the quality of HA output. Two new reference-free HA speech quality indices are proposed in this paper, based on a model that amalgamates perceptual linear prediction (PLP), hearing loss (HL) modeling, and machine learning concepts. For the first index, HL-modified PLP coefficients and their statistics were used as the feature set, which was subsequently mapped to the predicted quality scores using support vector regression (SVR). For the second index, HL-impacted gammatone auditory filterbank energies and their second-order statistics constituted the feature set, which was again mapped using SVR. Two databases involving HA recordings were collected and utilized for the evaluation of the robustness and generalizability of the two indices. Experimental results showed that the index based on the gammatone filterbank energies not only correlated well with HA quality ratings by hearing impaired listeners, but also exhibited robust performance across different test conditions and was comparable to the full-reference HASQI performance.

Background Hearing aid prescriptive methods are a commonly recommended component of evidence-based preferred practice guidelines and are often implemented in the hearing aid programming software. Previous studies evaluating hearing aid manufacturers' software-derived fittings to prescriptions have shown significant deviations from targets. However, few such studies examined the accuracy of software-derived fittings for the Desired Sensation Level (DSL) v5.0 prescription. Purpose The purpose of this study was to evaluate the accuracy of software-derived fittings to the DSL v5.0 prescription, across a range of hearing aid brands, audiograms, and test levels. Research Design This study is a prospective chart review with simulated cases. Data Collection and Analysis A set of software-derived fittings were created for a six-month-old test case, across audiograms ranging from mild to profound. The aided output from each fitting was verified in the test box at 55-, 65-, 75-, and 90-dB SPL, and compared with DSL v5.0 child targets. The deviations from target across frequencies 250-6000 Hz were calculated, together with the root-mean-square error (RMSE) from target. The aided Speech Intelligibility Index (SII) values generated for the speech passages at 55- and 65-dB SPL were compared with published norms. Study Sample Thirteen behind-the-ear style hearing aids from eight manufacturers were tested. Results The amount of deviation per frequency was dependent on the test level and degree of hearing loss. Most software-derived fittings for mild-to-moderately severe hearing losses fell within ± 5 dB of the target for most frequencies. RMSE results revealed more than 84% of those hearing aid fittings for the mild-to-moderate hearing losses were within 5 dB at all test levels. Fittings for severe to profound hearing losses had the greatest deviation from target and RMSE. Aided SII values for the mild-to-moderate audiograms fell within the normative range for DSL pediatric fittings, although they fell within the lower portion of the distribution. For more severe losses, SII values for some hearing aids fell below the normative range. Conclusions In this study, use of the software-derived manufacturers' fittings based on the DSL v5.0 pediatric targets set most hearing aids within a clinically acceptable range around the prescribed target, particularly for mild-to-moderate hearing losses. However, it is likely that clinician adjustment based on verification of hearing aid output would be required to optimize the fit to target, maximize aided SII, and ensure appropriate audibility across all degrees of hearing loss.

Objective: Evaluation of the Sennheiser HDA 280-CL circumaural headphone for the determination of (1) equivalent threshold sound pressure levels (ETSPL) for 125–18,000 Hz.; (2) real ear attenuation (250–8000 Hz); (3) insertion loss (63–18,000 Hz); (4) frequency response (125–18,000 Hz); (5) total harmonic distortion (THD) (125–10,000 Hz); and, (6) linearity (11,200–18,000 Hz).Study Sample: Twenty-five normal hearing adults aged 18–25 participated in (1) and (2).Design: (1) Hearing thresholds were measured using the Sennheiser HDA 280-CL. Frequency specific ETSPL values were calculated in an artificial ear. (2) Sound field thresholds were measured with the ears open and covered with the headphone to obtain the real ear attenuation thresholds (REAT). These values were used to determine the maximum permissible ambient noise levels (MPANL). (3) A B&K HATS mannequin recorded the output levels of a broadband pink noise with the ears open and covered with the headphones. (4, 5) The frequency response, THD and linearity were measured in an artificial ear.Results: Values for ETSPL, REAT, MPANL, insertion loss, as well as measures of frequency response, THD and linearity are presented.Conclusions: The Sennheiser HDA 280-CL meets the requirements for audiometric testing and the values presented can be used for calibration.

Direct drive hearing devices, which deliver a signal directly to the middle ear by vibrating the tympanic membrane via a lens placed in contact with the umbo, are designed to provide an extension of audible bandwidth, but there are few studies of the effects of these devices on preference, speech intelligibility, and loudness. The current study is the first to compare aided speech understanding between narrow and extended bandwidth conditions for listeners with hearing loss while fitted with a direct drive hearing aid system. The study also explored the effect of bandwidth on loudness perception and investigated subjective preference for bandwidth. Fifteen adult hearing aid users with symmetrical sensorineural hearing loss participated in a prospective, within-subjects, randomized single-blind repeated-measures study. Participants wore the direct drive hearing aids for 4 to 15 weeks (average 6 weeks) prior to outcome measurement. Outcome measures were completed in various bandwidth conditions achieved by reducing the gain of the device above 5000 Hz or by filtering the stimuli. Aided detection thresholds provided evidence of amplification to 10000 Hz. A significant improvement was found in high-frequency consonant detection and recognition, as well as for speech in noise performance in the full versus narrow bandwidth conditions. Subjective loudness ratings increased with provision of the full bandwidth available; however, real-world trials showed most participants were able to wear the full bandwidth hearing aids with only small adjustments to the prescription method. The majority of participants had either no preference or a preference for the full bandwidth setting.

This study evaluated a deep-neural-network denoising system using model-based design, comparing it with adaptive filtering and beamforming across various noise types, SNRs, and hearing-aid fittings. A KEMAR manikin fitted with five audiograms was recorded in reverberant and non-reverberant rooms, yielding 1,152 recordings. Speech intelligibility was estimated using the HASPI from 1,152 KEMAR manikin recordings. Effects of processing strategy and acoustic factors were tested with model-based within-device design that account for repeated recordings per device/program and fitting. Linear mixed model results showed that the DNN with beamforming outperformed conventional processing, with strongest gains at 0 and +5 dB SNR, moderate benefits at −5 dB in low reverberation, and none in medium reverberation. Across SNRs and noise types, the DNN combined with beamforming yielded the highest predicted intelligibility, with benefits attenuated under moderate reverberation. Azimuth effects varied; because estimates were derived from a better-ear metric on manikin recordings. Additionally, this paper reports comparisons using metrics of sound quality, for an intrusive metric (HASQI) and the pMOS non-intrusive metric. Results indicated that model type interacted with processing and acoustic factors. HASQI and pMOS scores increased with SNR and were moderately correlated (r² ≈ 0.479), supporting the use of non-intrusive metrics for large-scale assessment. However, pMOS showed greater variability across hearing aid programs and environments, suggesting non-intrusive models capture processing effects differently than intrusive metrics. These findings highlight the promise and limits of non-intrusive evaluation while emphasizing the benefit of combining deep learning with beamforming to improve intelligibility and quality.