A Taxonomy-oriented Overview of Noise Compensation Techniques for Speech Recognition

Designing a machine that is capable for understanding human speech and responds properly to speech utterance or spoken language has intrigued speech research community for centuries. Among others, one of the fundamental problems to building speech recognition system is acoustic noise. The performance of speech recognition system significantly degrades in the presence of ambient noise. Background noise not only causes high level mismatch between training and testing conditions due to unseen environment but also decreases the discriminating ability of the acoustic model between speech utterances by increasing the associated uncertainty of speech. This paper presents a brief survey on different approaches to robust speech recognition. The objective of this review paper is to analyze the effect of noise on speech recognition, provide quantitative analysis of well-known noise compensation techniques used in the various approaches to robust speech recognition and present a taxonomy-oriented overview of noise compensation techniques.