New Delhi, India
minigranth@gmail.com

Speech Recognition System

Created with Sketch.

Artificial Intelligence Tutorial


[fblike]

Speech Recognition System : Introduction

  • Speech recognition is one of the major applications where the concept of Hidden Markov Model is used.

  • The goal of speech recognition is to identify a sequence of words produced by a speaker, given the acoustic signals.

  • A speech recognition system could be useful as, speech is the dominant modality for communications among humans.

 

Speech Recognition System : Working

  • A speech recognition system consists of five basic building blocks which can be used to determine the almost exact meaning of input speech.

  • Acoustic Models : An acoustic model uses concept of Hidden Markov Model to integrate signals to the phonemes. It is a data base consisting of probabilistic models that can correlate phonemes to input speech utterance signals.

 

This image describes the complete speech recognition system consisting of five major processes in between before generating the desired output.

Speech Recognition System

 

  • Dictionary: Dictionary is used for mapping words with their pronunciation using phonemes.

  • Phonemes: They are smallest speech units that can be used to make one word different from another word.

  • Language Model: It is used to collect phrases that are acceptable i.e. words from dictionary are arranged into phrases into language model.

  • The language model integrates together to form a meaningful sentence which finally results in recognition from phrases.