An automated language-independent speech-in-noise test for remote hearing testing

Aim: Speech-in-noise tests (SNTs) are valuable measures of hearing ability in real-life conditions and may promote awareness and detect age-related hearing impairment at early stage. The aim of this study was to develop a language-independent, automated, fast SNT viable for remote screening and testing. Methods: We used Vowel-Consonant-Vowel (VCV) stimuli and multiple-choice tasks. The rationale was to limit the influence of education, literacy, and native language and to make the task feasible via user-operated automated procedures (e.g., via smartphone or web). We analysed VCV intelligibility as a function of signal-to-noise-ratio (SNR) in five languages (English, French, German, Italian, Portuguese) by combining objective and subjective measures. We developed two adaptive SNTs, one based on conventional ±2dB SNR rules (SNT-CN) and one based on newly developed rules that use intelligibility steps on clustered VCV (SNT-CL). We assessed test performance in 25 normal-hearing young adults. Results: Among the five languages here studied, the intelligibility of English VCVs showed the best combination of reliability, intelligibility, and dynamic range as a function of SNR. SNT-CL and SNT-CN led to similar speech reception thresholds (SRTs): -15.4 and -15.5 dB SNR (p=0.89). Test duration with SNT-CL was significantly lower than with SNT-CN: 3'30'' vs 5'20'' (p<<0.05), i.e. about two minutes shorter. The SRTs from test and retest trials were significantly different with SNT-CN (p=0.003) and repeatable with SNT-CL (p=0.7). Preliminary data in uncontrolled settings suggested that the SNTs are robust to noise. Conclusions: The proposed SNT-CL provided reliable estimates of SRTs and was robust to ambient noise. Compared to SNT-CN, it showed better repeatability and significantly shorter duration. As such, the proposed test may be viable for remote testing. Further research is needed to validate it in multi-language settings and define specifications for web or mobile delivery.