Research Cell: An International Journal of Engineering Sciences

Open Access Open Access  Restricted Access Subscription Access

Resource Creation for Sanskrit ASR (Automatic Speech Recognition)


Affiliations
1 School of Sanskrit and Indic Studies, 2Jawaharlal Nehru University, New Delhi, India., India
 

There are a few works on Automatic Speech Recognition (ASR) for Sanskrit. Generally, developing an ASR system is a time and cost-consuming multi-layered teamwork. An average of basic 90 to 100 hours of annotated speech corpus is required for the development of a basic ASR system. The present paper is part of a doctoral research work that proposes corpora creation to provide a reliable annotated speech dataset for future research on Sanskrit ASR.

Keywords

Resource Creation, Automatic Speech Recognition for Sanskrit, Indian Language, Sanskrit Manuscripts, Transcription.
User
Notifications
Font Size


  • Resource Creation for Sanskrit ASR (Automatic Speech Recognition)

Abstract Views: 239  |  PDF Views: 0

Authors

Devendr Kumar
School of Sanskrit and Indic Studies, 2Jawaharlal Nehru University, New Delhi, India., India
Girish Nath Jha
School of Sanskrit and Indic Studies, 2Jawaharlal Nehru University, New Delhi, India., India

Abstract


There are a few works on Automatic Speech Recognition (ASR) for Sanskrit. Generally, developing an ASR system is a time and cost-consuming multi-layered teamwork. An average of basic 90 to 100 hours of annotated speech corpus is required for the development of a basic ASR system. The present paper is part of a doctoral research work that proposes corpora creation to provide a reliable annotated speech dataset for future research on Sanskrit ASR.

Keywords


Resource Creation, Automatic Speech Recognition for Sanskrit, Indian Language, Sanskrit Manuscripts, Transcription.

References