Authors' note: This research was funded by the National Institute of Deafness and Communication Disorders of the National Institutes of Health under Award Number 5R21DC016241-02. Hearing aids are a central part of managing adults with hearing loss. Despite evidence supporting improved listening abilities and quality of life as a result of hearing aid use,1 device uptake remains poor. People with hearing loss typically take action only after an average of six to 12 years from the initial identification of the loss.2,3 Various factors hinder help-seeking and hearing aid uptake, including personal readiness, finances,4 and stigmatization.5 Having amplification alternatives within a consumer-driven service delivery model has the potential to increase a person's options and initial access to hearing care. Traditionally, a hearing aid is prescribed after evaluation by a licensed professional. In 2016, the U.S. Food and Drug Administration (FDA) published a non binding guidance document no longer enforcing medical assessment before the provision of amplification,6 thus allowing alternative self-test diagnostics and hearing devices.IStock/elenabs2. internet, telehealth, hearing loss.Figure 1: Views on listening experience while using smartphone hearing aid apps (n=5), internet, telehealth, hearing loss.Table 1: Latency (milliseconds;ms) from Original Sound Source to Output via Wired Earbuds for Android Smartphones and Applications, internet, telehealth, hearing loss.Table 2: Latency (milliseconds;ms) from Original Sound Source to Output from Wired Earbuds for iPhone 6 Across Applications, internet, telehealth, hearing loss.Table 3: Signal-to-Noise Ratio Improvement (dB SNR) Across Hearing Aid Applications, Noise Suppression (NS), or Program Settings for Android Smartphones, internet, telehealth, hearing loss.Table 4: Signal-to-Noise Ratio Improvement (dB SNR) Across Hearing aid Applications, Noise Suppression (NS)/ Noise Reduction (NR), or Program Settings for iPhone 6, internet, telehealth, hearing loss.Table 5: Cost of Smartphone Hearing Aid Apps for Android and iOS (USD), internet, telehealth, hearing loss.The ubiquitous nature of smartphones and the rise of hearing aid applications (apps) are promising, widely accessible, and affordable amplification alternatives to traditional hearing aids for certain people.7,8 To date, 61 percent of the global population are mobile internet users, with an expected growth to 79 percent by 2025.9 Amid rapid technological advancements, a large number of smartphone hearing aid apps are now available. However, there is limited research evidence on these apps’ performance and benefits compared with conventional hearing aids. One study investigated two smartphone hearing aid apps and showed comparable benefit to a conventional hearing aid in terms of amplification and speech-in-noise improvement.7 Due to constant developments in smartphone technology and creation of new hearing aid apps, this study investigated electroacoustic and self-reported performance across a range of apps and smartphone manufacturers. METHODS The study received Institutional Board Approval from the Health Science Ethics Committee at the University of Pretoria in South Africa. Two investigators evaluated the apps on both Google Play and the Apple App Store. The apps had to: be downloadable and function without internet connection; be simple to use without specialist knowledge; function through inexpensive wired earbuds or headphones; and produce reasonable quality sound as assessed through an informal listening check. Four apps on Google Play and the Apple App Store were selected (Table 1) and evaluated for (1) objective sound quality (latency and signal-to-noise ratio [SNR] improvement) and (2) subjective listening experience. In terms of sound quality, the latency or time delay of amplified signals was measured across three Android smartphones and one iPhone to evaluate the performance of the apps (Table 1). Latency was measured using a Rion NL-52 sound level meter and G.R.A.S 46AG-4 CCP occluded ear simulator, presented and recorded using Audacity® software (version 2.20). Click stimulus with one second between clicks was used to measure round-trip latency (in milliseconds) from the original sound source (loudspeaker) to the output using manufacturer-supplied wired Samsung earbuds. In addition, latency using the premium version of the Petralex app was measured between the wired earbuds and a set of wireless headphones (LG HBS with insert earbuds) for the Samsung S7 and iPhone 6. In all instances, the smartphones were placed 20 cm 0° azimuth from the loudspeaker. The smartphone apps were kept to default settings. SNR improvement was measured using three higher-end smartphones (the Samsung S6, Samsung S7, and iPhone 6) using various hearing aid apps and their respective noise suppression or program options (Table 2). Spoken digit triplets (0 to 9 separated by 500 ms silence) were presented in speech-weighted masking noise at 0 dB SNR, with the noise and digit energy kept constant at 70 dB for all recordings. The presented sound was received via the embedded headset microphone, and the smartphone output was recorded on a laptop computer via the artificial ear. Before SNR gains could be calculated, the smartphones were calibrated. To this end, the differences in noise energy, where no noise suppression was used, was calculated. These differences were used to calibrate the output of each smartphone/app-pair to 0 dB for no noise suppression. After calibration, the energy for each smartphone/app-pair was calculated for noise only, as well as for noise plus digits. This was done for no noise suppression and other available noise suppression settings. The SNR gains achieved by the various noise suppression settings were calculated by subtracting the noise energy from the noise plus digits energy in each case. For the listening evaluation, five participants with normal hearing (PTA0.5-4kHz ≤ 15 dB HL) were recruited from students at the University of Pretoria's department of speech-language pathology and audiology. Participants were asked to join a one-on-one conversation with the investigator while using the free-trial version of the Petralex app (available on both Android and iOS) on an iPhone 6, coupled with wired earbuds. Participants were instructed to complete the incorporated self-hearing test on the app and adjust settings to their comfort. However, on the trial version, settings for noise suppression and own-voice suppression options could not be activated. The participants were then asked to participate in the conversation, which was maintained for 10 minutes (timed) by asking and answering two sets of questions (one for each conversation partner) on various topics (e.g., travel, work, music). Afterwards, specific feedback related to the listening experience (Fig. 1) was captured using a five-point Likert scale. In addition, participants were able to provide feedback using the app. RESULTS Sound Quality. In terms of latency, all apps on the iPhone 6 surpassed those on the three Android smartphones, with shorter latencies depending on the hearing aid app used (Tables 1 and 2). The shortest latency was recorded for Super Hearing Aid using the iPhone 6. Latency varied for all hearing aid apps on the Android smartphones. The shortest latency on Android was obtained using Petralex on the Samsung S7. Furthermore, there was a substantial difference in latency between wired and wireless earbuds. Wired earbuds had shorter latencies of 58 ms and 20 ms for the Samsung S7 and iPhone 6, respectively, as opposed to 580 ms and 145 ms using the wireless headset. Some apps provided noise suppression or settings for different listening environments. SNR improvement (dB SNR) for the various apps with their respective settings are provided in Tables 3 and 4. Fennex on the iPhone 6 produced the best SNR improvement of 18.1 dB SNR when noise reduction was activated. For Android, Petralex performed the best when set at full noise suppression, with 3.9 dB SNR improvement. However, it was still considerably lower than the Petralex app on the iPhone 6, with 14.3 dB SNR improvement. In some instances, when no noise suppression or reduction strategies were set, SNRs became worse, for example, using Petralex on the Samsung S6 (-2.1 dB), S7 (-1.6 dB), and iPhone 6 (-1 dB SNR). In general, apps with better sound quality performance were more expensive (Table 5). Subjective Listening Experience. Participants indicated whether they agreed or disagreed with five statements pertaining to their listening experience using the Petralex app (Fig. 1) that had the best electroacoustic performance (Tables 1-4). The majority (4/5) either agreed or strongly agreed that the conversation was easy to follow while using the app. However, three participants said that they would not use the app in difficult listening environments, and all five participants indicated that they would prefer to use a hearing aid as opposed to the app (if they had hearing loss). Where participants provided their own views on the app, three said that their own voice was either too loud or echoed, and that environmental sounds were over amplified. DISCUSSION Overall, apps on the iPhone 6 with wired earphones had the shortest latency and highest SNR improvement when using the Fennex and Petralex apps. Android phones had longer latencies and lower SNR improvement on all apps, and performance varied between devices. Research on tolerable processing delays in digital hearing aids has shown that latencies as low as 20 to 30 ms are perceived as disturbing by people with mild-to-moderate hearing loss.10 Even among individuals with normal hearing, speech production is affected when delays exceed 30 ms.11 The coupling of a wireless headset to both Android and iPhone devices produced longer latencies that were well over the acceptable (<20 ms) range. Most apps, however, include disclaimers that apps should be used with wired earbuds to avoid signal delay, although connection to wireless headsets are possible (e.g., Fennex, Petralex, Super Hearing Aid, Hearing Aid Master). All apps improved SNR, but, across Android apps, improvements were small (0.48 – 3.90 SNR) compared with iOS (0.84 – 18.11 SNR). Only apps on iOS had latencies that approximated acceptable signal delay and showed significant SNR improvement. The iOS apps used in the study of Amlani, et al., (2013),7 which showed speech recognition performance and amplification comparable with an audiologist-fitted hearing aid, were no longer available. None of the participants indicated a preference for using an app over a hearing aid. Following conversations while using the app was rated favorably by the participants, but they expressed concerns about aspects such as own-voice and background noise amplification. Of note, the free-trial versions of apps like Petralex, used for subjective listening evaluation in this study, did not allow for settings like own-voice suppression and noise reduction. The option to access more premium settings (e.g., own voice suppression) and better SNR improvement would depend on a person's willingness to pay for the application. Hearing aid apps have the potential to increase global access to amplification at a reduced cost, increase awareness of hearing loss, and reduce stigma related to wearing hearing aids.7 In addition, periodic app updates could ensure that the user always has the latest software and features available. However, one problem could be that self-administered diagnostic hearing tests incorporated into the app to determine the amount of amplification needed are usually some variant of international gold-standard pure-tone audiometry. Remote pure-tone testing may have variable accuracy for different degrees of hearing loss, devices, or earphones.12 Although the better-performing apps (Fennex and Petralex) use hearing aid algorithms to apply amplification, inaccurate hearing results could lead to inappropriate over- or underamplification. In addition, the performance of hearing aid apps is not uniform across devices or platforms (iOS or Android). In conclusion, some hearing aid apps, such as Fennex running on iOS, could potentially benefit people with hearing loss, but, at least for now, should be coupled with wired headsets for optimal performance. For Android smartphones, increased signal delays should be expected. Further developments to minimize latency and provide valid self-hearing test procedures would be helpful to improve the apps.