INTERNATIONAL JOURNAL OF SCIENTIFIC DEVELOPMENT AND RESEARCH International Peer Reviewed & Refereed Journals, Open Access Journal ISSN Approved Journal No: 2455-2631 | Impact factor: 8.15 | ESTD Year: 2016
open access , Peer-reviewed, and Refereed Journals, Impact factor 8.15
A Kiswahili Dataset for the Development of a Text-to-Speech System
Authors Name:
Kelvin Kiptoo Rono
, Ciira wa Maina , Elijah Mwangi
Unique Id:
IJSDR2208117
Published In:
Volume 7 Issue 8, August-2022
Abstract:
A Text-to-Speech (TTS) system requires adequate data to efficiently build an intelligent and natural system. Based on open-source, free-access data availability, languages are classified into high-resource and low-resource. Kiswahili language, which has a vast population of speakers, is still classified as a low-resource language. It is a low-resource language since there is a limited dataset for building Natural language processing (NLP) systems. This article presents a Kiswahili dataset that contributes to the required language processing resources to build a TTS system or other NLP tasks. A TTS system based on an Artificial Neural Network (ANN) is the current state of the art, and this dataset has been successfully used to build a TTS system. The dataset consists of 7,108 audio clips from a single speaker, with each audio clip varying from 1s to 12.5s. The total audio length is approximately 16 hrs. The dataset created thus meets all the requirements for developing a Kiswahili ANN-based project.
Keywords:
Text-to-Speech (TTS) system, Deep Learning, Natural Language Processing, Kiswahili Dataset, Artificial Neural Networks (ANNs), Language Modelling
Cite Article:
"A Kiswahili Dataset for the Development of a Text-to-Speech System", International Journal of Science & Engineering Development Research (www.ijsdr.org), ISSN:2455-2631, Vol.7, Issue 8, page no.788 - 791, August-2022, Available :http://www.ijsdr.org/papers/IJSDR2208117.pdf
Downloads:
000337212
Publication Details:
Published Paper ID: IJSDR2208117
Registration ID:201482
Published In: Volume 7 Issue 8, August-2022
DOI (Digital Object Identifier):
Page No: 788 - 791
Publisher: IJSDR | www.ijsdr.org
ISSN Number: 2455-2631
Facebook Twitter Instagram LinkedIn