Introduction to Speech recognition System

Introduction to speech recognition System 

Speech is the vocalized form of human 

communication. Speech is natural, easy, fast, hands-free and do not require technical knowledge. Human beings are comfortable with speaking directly with computers rather 

than depending on primitive interfaces such as keyboards and pointing devices. The primitive interfaces like keyboard 

and pointing devices require certain amount of skills for effective usage. Earlier human intract with computer throught the keyboard or mouse. When the advancement in technology keeps growing, the way of intracting changes drastically. From 1950s the speech recognition technology came into existence, a technology through which speech is converted to text. 

When human interacts with the atmosphere, they receive information through different modalities: sight, audio, smell 

and touch. Speech is one of the most important media of communication. With the increase in the development of 

technologies it has become important way for the exchange of information between human and computers. Automatic 

speech recognition (ASR) system converts speech input to a readable text output. For this various machine learning techniques are used. 

Interaction with computer through a convenient and user-friendly interface has always been an important technological issue. Machine-oriented interfaces restrict 

the computer usage to a minuscule fraction of the population, who are both computer literate and conversant with written English.

Speech recognition is also popular with the name automatic speech recognition. ASR takes speech signal and convert these signal to text accordingly to crossponding text. In speech recognition the speech signal is process and extraction of features is done. These features are then passed to the matching model to generate text. Basically all automatic speech recognizer have three modules:

a) Acoustic processing

b) Features extraction 

c) Recognition

In Sociolinguistics, accent is a manner of pronouncing a language. Anyone who speaks a language, does so in an accent. The way the native speakers of a language speak that language defines the standard pronunciation, and is generally considered to be the standard or reference accent for that language.

In India, there are various regional 

varieties due to which the speaking styles of human being vary because of regional influences. English spoken by the 

people does not differ from native English in vocabulary and grammar but in pronunciation. Hence a desired speech recognition system needs to be developed.

Indian English (IE) derives from British English, but they differ in many aspects. It varies from speech to word, and is a typical 

variety of English. At present, it is hard to see researches on continuous speech recognition (CSR) of lesser-known English

varieties such as Indian English.Indian English has developed some distinctive features of its own with phonological features being the most remarkable. Compared with British English (or American English) which has a large amount of annotated speech data, IE is a relatively low-resourced language. What's more, the performance of existing English CSR systems perform unsatisfactorily when dealing with IE spontaneous conversations.