Noise Subspace Fuzzy C-means Clustering for Robust Speech Re

Abstract. In this paper a fuzzy C-means (FCM) based approach for speech/non-speech discrimination is developed to build an effective voice activity detection (VAD) algorithm. The proposed VAD method is based on a soft-decision clustering approach built ove

Noise Subspace Fuzzy C-means Clustering for Robust Speech RecognitionJ.M. G´rriz1 , J. Ram´ 1 , J.C. Segura1 , o rez C.G. Puntonet2 , and J.J. Gonz´lez2 a1Dpt. Signal Theory, Networking and communications, University of Granada, Spain gorriz@ugr.es, WWW home page: http://www.ugr.es/~gorriz 2 Dpt. Computer Architecture and Technology, University of Granada, SpainAbstract. In this paper a fuzzy C-means (FCM) based approach for speech/non-speech discrimination is developed to build an e ective voice activity detection (VAD) algorithm. The proposed VAD method is based on a soft-decision clustering approach built over a ratio of subband energies that improves recognition performance in noisy environments. The accuracy of the FCM-VAD algorithm lies in the use of a decision function de ned over a multiple-observation (MO) window of averaged subband energy ratio and the modeling of noise subspace into fuzzy prototypes. In addition, time e ciency is also reached due to the clustering approach which is fundamental in VAD real time applications, i.e. speech recognition. An exhaustive analysis on the Spanish SpeechDat-Car databases is conducted in order to assess the performance of the proposed method and to compare it to existing standard VAD methods. The results show improvements in detection accuracy over standard VADs and a representative set of recently reported VAD algorithms.1IntroductionThe emerging wireless communication systems are demanding increasing levels of performance of speech processing systems working in noise adverse environments. These systems often bene ts from using voice activity detectors (VADs) which are frequently used in such application scenarios for di erent purposes. Speech/non-speech detection is an unsolved problem in speech processing and a ects numerous applications including robust speech recognition, discontinuous transmission, real-time speech transmission on the Internet or combined noise reduction and echo cancelation schemes in the context of telephony [1, 2]. The speech/non-speech classi cation task is not as trivial as it appears, and most of the VAD algorithms fail when the level of background noise increases. During the last decade, numerous researchers have developed di erent strategies for detecting speech on a noisy signal [3] and have evaluated the in uence of the VAD e ectiveness on the performance of speech processing systems [4]. Most of them have focussed on the development of robust algorithms with special attention on the derivation and study of noise robust features and decision rules [5–7, 3].