The purpose of speech is communication, and the research in this area primarily deals with processing, representation of speech signal to develop voice based interfaces for human-computer interaction. Such natural interfaces enable access to information via hands-free mode to literate, illiterate as well as vision impaired people. The objective of Speech Processing Lab is to conduct goal oriented basic research, and thus we address fundamental issues involved in building robust speech-to-text systems, natural sounding text-to-speech systems, spoken/audio information retrieval, language identification, speaker recognition and emotion recognition using speech.
Indian language Identification System
Language Identification, Deep Neural Networks.
Automatic language identification (LID) refers to the task of identifying the language of a spoken utterance.
Language identification (LID) has a wide range of applications in multilingual speech recognition system as a front-end in information services (customer care) and in providing many computer and telephone based services. Any spoken utterance contains information about the speaker, emotion of the speaker, channel, environment and other variable factors. Presence of such variability makes it challenging to identify a language invariant to such factors. Especially, in an Indian scenario, where almost every state has a language of its own and every language having hundreds of dialects, the task of identifying a language becomes more difficult. This technology aims to provide an Indian language identification system using Deep Neural Networks. In the current work, it considers 13 of the official Indian languages and built system using DNNs. The technologyis being extended to all 23 official indian languages.
Type of Work
Current State of work
Technology designed and implemented
1. DRDO, Mobile industry and call centre companies will be interested
1. An Investigation of Deep Neural Network Architectures for Language Recognition in Indian Languages, Interspeech 2016
2. Improved Language Identification in Presence of Speech Coding, Springer LNAI, MIKE 2015