Introduction:
Speech perception in noisy environments is a crucial skill that often declines with age and varies across languages. Considering the need for a language-specific assessment tool, the present study aimed to develop the Speech-in-Noise Test–Hindi (SiNT-H) and to evaluate speech recognition performance and phonetic error patterns between younger and middle-aged adults at varying signal-to-noise ratio (SNR) levels.

Material and methods:
The study included 27 young and 27 middle-aged native Hindi speakers with hearing thresholds ≤25 dB HL. The study was conducted in two phases. Phase I involved the development of the SiNT-H, its administration to young adults, analysis of performance across SNRs, and assessment of test–retest reliability. Phase II compared SiNT-H scores between young and middle-aged adults at different SNR levels using four-talker babble noise. Additionally, a detailed phonetic error analysis was carried out across SNR conditions.

Results:
The SiNT-H demonstrated no significant ear-wise or list-wise differences. Females outperformed males, with younger females showing better performance than middle-aged females, indicating a gender–age interaction. Performance significantly declined with decreasing SNRs, confirming the test’s sensitivity to noise. Phonetic analysis revealed increased errors at lower SNRs, particularly in voicing, manner, place, and aspiration features. At 0 dB SNR, voicing, manner, and aspiration errors were more prevalent among middle-aged adults, reflecting age-related processing declines.

Conclusions:
The findings highlight the importance of incorporating language-specific acoustic-phonetic features in speech-in-noise assessment and underscore the influence of age on speech perception in noisy environments. The SiNT-H appears to be a sensitive tool for evaluating age-related changes in speech recognition in noise among Hindi speakers.