Abstract:
Aim and Objectives: The present study aimed to measure the efficacy of a novel noise reduction method on speech recognition scores in the presence of background noise (cafeteria and traffic) in individuals with sensorineural hearing impairment.
Methods: Twenty-three adults diagnosed with moderate and moderate to moderately severe
sensorineural hearing loss were participated in the study. The speech recognition scores (SRS) were measured in Kannada, from three different noise reduction algorithms, i.e., traditional noise reduction algorithm, novel noise reduction algorithm, and no noise reduction, at three different levels of SNRs (-5, 0 & 5 dB) under cafeteria and traffic noise.
Results and Discussion: The results showed that at −5 dB and 0 dB SNR, the novel noise
reduction algorithm had a better SRS compared to the no-noise reduction condition, at −5 dB, 0 dB, and 5 dB SNR, no noise reduction had a better SRS than the traditional noise reduction algorithm and at 5 dB SNR, there was no much difference in SRS between the novel noise reduction algorithm and the no noise reduction condition for both the types of noises. Across all three SNR levels, the novel noise reduction algorithm had a better SRS than the traditional noise reduction algorithm. Traffic noise had a better SRS than the cafeteria noise.
Conclusion: The present study showed that the novel noise reduction algorithm significantly
improved speech recognition in noisy conditions, in low and moderate SNRs and this algorithm has potential implication in real-time signal processing hearing aids to improve SNRs.