Manipulation of Signal-to-Noise Ratio to Compensate for Variations in Word Identification Scores Due to Change in Masker

Abstract

The type of masking noise is known to affect speech identification. Some maskers are known to have a greater masking effect on speech than others. Thus, the study aimed to investigate whether manipulating the signal-to- noise ratio (SNR) of a masker can compensate for variations in word identification scores obtained due to change in the type of masker. To investigate this, the scores obtained by 20 children on a speech identification test using an 8-talker babble was compared with that obtained on a word identification test in the presence of white noise. The former test was evaluated at 0 dB SNR using the `Speech perception-in-noise in Kannada' (SPIN-K) and the latter in three different SNRs (0 dB, -5 dB, & -10 dB) using the `Kannada Word identification-in-white noise' (WIWN-K). Speech babble was found to have a greater masking effect at 0 dB SNR, resulting in poorer speech identification scores than white noise. However, the speech identification scores obtained using white noise at -10 dB SNR was equivalent to that of scores obtained with speech babble at 0 dB SNR. The study highlights that the masking effect of continuous white noise can be made equivalent to the maskingeffect of an 8-talker speech babble by reducing the SNR.