Development of Frugal Speech Corpus for Low Resource Indian Languages

Abstract

A speech corpus (or spoken corpus) is a database of audio files and text transcriptions. In speech technology, speech corpora are used, among other things, to create acoustic models (which can then be used with a speech recognition engine). Speech corpus is central element for training an acoustic model used in a speech recognition engine. In Linguistics, spoken corpora are used to do research into Phonetics, Conversation analysis, Dialectology and other _elds. Creation of speech corpus is a laborious, expensive and time-consuming task. Recording of speech _les is done manually by many speakers. Then do the transcriptions which is the process of converting the speech to it's corresponding text. Typically there are two types of Speech Corpora: Read Speech - for example, speakers are asked to read Book excerpts Broadcast news Lists of words Sequences of numbers Spontaneous Speech - for examples Dialogs - between two or more people Narratives - a person telling a story Map-tasks - one person explains a route on a map to another Appointment-tasks - two people try to find a common meeting time based on individual schedules Building speech recognition application for resource deficient languages is a challenge because of unavailability of speech corpus. This work proposes a mechanism to develop an inexpensive speech corpus for low resource Indian languages by exploiting existing collections of online speech data to build a frugal speech corpus.