Comparison of Spectral Subtraction Noise Reduction Algorithms

Abstract

Noise in media is any undesirable signal that masks relevant information content. The addition of noise to real-world data in any context is practically inevitable. Noise reduction algorithms in the past have tackled the problem, albeit compromising on adaptability to multiple real-world uses. Prevalent methods of noise reduction are either not adequately versatile and/or are time and resource extensive. Spectral subtraction provides a hybrid approach to noise reduction that incorporates versatility and efficient resource usage. This research tested the performance of two spectral subtraction noise reduction algorithms (stationary and non-stationary) across 5 categories of real-world noise (speech only, speech with natural noise, music, animal sounds, and noise only). The testing was done on the basis of normalized cross-correlation, i.e. similarity between the noise-reduced audio and the original recording in each case. The graphs and audios of the original recording and the noise reduced samples were also recorded and studied. Non-stationary spectral subtraction performed better in samples where human speech was the target: speech only and speech with natural noise. Stationary spectral subtraction performed better when denoising music and animal sounds. This anomaly in performance between the two algorithms was only noted in categories with no human speech. This may have been because the algorithms were trained to denoise human speech primarily. These results exemplify the performance and versatility of different spectral subtraction algorithms. The category-specific results can be used to employ specific spectral subtraction algorithms for specific tasks for optimum performance.