SINGLE CHANNEL SPEECH ENHANCEMENT USING EVOLUTIONARY ALGORITHM WITH LOG-MMSE

Additive noise degrades speech quality and intelligibility. Speech enhancement reduces this to make more pleasant intelligible. It plays a significant role in recognition or speech-operated systems. In paper, we propose single-channel method which the log-minimum mean square error (log-MMSE) modified accelerated particle swarm optimization algorithm are used design filter for improving intelligibility of noisy speech. Accelerated (APSO) is single dimension position changed iteration while obtaining particle’s new position. Using algorithm, designed with multiple passbands notches enhancement. The converges faster compared standard (PSO) APSO giving optimum coefficients. enhance proposed improves perceptual estimation (PESQ) by 17.05% 5dB babble noise, 33.92 % car 14.96 airport 39.13 exhibition noise. average output PESQ these four types improved conventional methods There an 7.58 dB improvement segmental SNR types. minimum distortion.