Free Essays, Term Papers, Research Papers at Anti Essays

1
CHAPTER 1 INTRODUCTION 1.1 VOICE ACTIVITY DETECTION: The voice activity detection represents an important part of the speech processing and it is the subject of contemporary research. The Voice Activity Detectors are used typically in speech enhancement algorithms, which are based on the estimations of the noise characteristics. Other Voice Activity Detectors application can be found in communication area, e.g. in VoIP (Voice over Internet Protocol). To conserve bandwidth, Voice Activity Detector prevents “silence packets” from being transmitted. It was also shown that usage of the Voice Activity Detectors can increase accuracy of the speech recognizer. The amount of observed pause samples is usually very high, thus its probability is higher than the probability of other speech segments. This can rationalise the separate recognition of pauses and the utilization of the voice activity detector. Voice is the basic mode of communication that conveys the contents of the speech and helps to identify the production source. With a growth of technology in electronic communication, voice communication has changed from simple telephonic system to mobile and internet systems. Impairments to generated speech signal occur chiefly due to background noise at source and noise introduced in a channel. It is often possible to lose clarity and intelligibility of speech because of noise pickup. Detection of voice segments from noisy speech is necessary step at a first stage of these systems. Detected voice segments can be preceded further to offer accurate results in these systems.
We can describe Voice Activity Detection as an algorithm or procedure which determines if the voice being analyzed is human speech or background noise. Generally the characteristics of the signal segments are needed, that emphasize the difference between speech and noise segments. These characteristics are formulated in terms of parameters of the signal. Commonly used parameters are...