IMPLEMENTING GRAPHICAL STRUCTURE BASED SPEECH-TO-TEXT SYSTEM FOR HINDI

Back to Page Authors: Divyendu Kumar Mishra

---

Keywords: Speech-to-text, Speech Recognition engine (SRE),, acoustic model, Diacritic, suprasegmental feature

---

Abstract: Speech is a natural gift that separates humans from other creatures and speech is the medium of communication between two human. There are two phenomenons in communication between two human first to give any message in the form of speech and second to receive any message in form of speech. In between these two processes it is important that receiver receives same message as sender send. Sender send the message in form of words and some linguistics symbols such as full stop (.), question mark (?) and exclamation mark (!) are hidden in that message. These linguistics symbols are not used in speaking form of any language but appear in written form of language. Humans can easily understand these linguistics symbols but sometimes when human gives input as a speech to the system and wants to receive written form of speech (text) as a output by the system then problem arise that system cannot understand these hidden linguistics symbols and cannot convert these symbols in text. This paper covers approach of using linguistics symbols in speech-to-text system and use of delete command for Hindi Language so that user can maintain the accuracy of speech-to-text system.

---