Information mining from speech signal

Mining information from speech signal as the ultimate goal of data mining is concernedwith science, technology, and engineering of discovering patterns and extracting potentiallyuseful or interesting information automatically or semi-automatically from speech data. With theadvent of inexpensive storage space and faster processing over the past decade, data miningresearch start to penetrate these new grounds in the areas of speech and audio processing. A typicalspeech can be broadly defined as speech with emotional content, speech affected by alcohol anddrugs speech from speakers with disabilities, and various kinds of pathological speech. Individualspeech can vary because of different timing and amplitude of the movement of speecharticulators. Physical mechanism of speech undergoes changes; which can affect the nasal cavityresonance and mode of vibration of the vocal cords. A series of environmental variables likebackground noise, reverberation and recording condition have also to be taken into account. Inessence every speech production is unique and this uniqueness makes the automatic speechprocessing quite difficult.