Advancing automatic speech recognition for low-resource ghanaian languages: Audio datasets for Akan, Ewe, Dagbani, Dagaare, and IkposoScience Data Bank

Audio datasets are fundamental to the development of automatic speech-recognition (ASR) systems. However, the availability of a large corpus of audio datasets in low-resource languages (LRLs) is limited. This study addresses this gap by introducing audio speech datasets for five low-resource languag...

Full description

Saved in:
Bibliographic Details
Main Authors: Isaac Wiafe, Jamal-Deen Abdulai, Akon Obu Ekpezu, Raynard Dodzi Helegah, Elikem Doe Atsakpo, Charles Nutrokpor, Fiifi Baffoe Payin Winful, Kafui Kwashie Solaga
Format: Article
Language:English
Published: Elsevier 2025-08-01
Series:Data in Brief
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S2352340925006043
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849249364049920000
author Isaac Wiafe
Jamal-Deen Abdulai
Akon Obu Ekpezu
Raynard Dodzi Helegah
Elikem Doe Atsakpo
Charles Nutrokpor
Fiifi Baffoe Payin Winful
Kafui Kwashie Solaga
author_facet Isaac Wiafe
Jamal-Deen Abdulai
Akon Obu Ekpezu
Raynard Dodzi Helegah
Elikem Doe Atsakpo
Charles Nutrokpor
Fiifi Baffoe Payin Winful
Kafui Kwashie Solaga
author_sort Isaac Wiafe
collection DOAJ
description Audio datasets are fundamental to the development of automatic speech-recognition (ASR) systems. However, the availability of a large corpus of audio datasets in low-resource languages (LRLs) is limited. This study addresses this gap by introducing audio speech datasets for five low-resource languages spoken in Ghana and parts of Togo. Specifically, it presents a 5000-hour speech corpus in Akan, Ewe, Dagbani, Dagaare, and Ikposo. Each language corpus includes 1000 h of validated audio speech recorded by their indigenous speakers. These audio recordings are spoken descriptions of 1000 culturally relevant images collected using a custom Android mobile application. To enhance the dataset’s utility in ASR and linguistic research 10 % of the audio recordings for each language were randomly selected and transcribed, resulting in approximately 100 h of transcription per language. This dataset represents a critical resource for preserving and documenting Ghanaian languages. It holds the potential for advancing speech and language technologies in these languages. Creating this audio dataset is the first step towards bridging the technological gap between high- and low-resource languages. Ethical guidelines were strictly followed throughout the data collection process and participants were given incentives for lending their voices to this study.
format Article
id doaj-art-07719d37eaea485dba01b925dcec26c4
institution Kabale University
issn 2352-3409
language English
publishDate 2025-08-01
publisher Elsevier
record_format Article
series Data in Brief
spelling doaj-art-07719d37eaea485dba01b925dcec26c42025-08-20T03:57:36ZengElsevierData in Brief2352-34092025-08-016111188010.1016/j.dib.2025.111880Advancing automatic speech recognition for low-resource ghanaian languages: Audio datasets for Akan, Ewe, Dagbani, Dagaare, and IkposoScience Data BankIsaac Wiafe0Jamal-Deen Abdulai1Akon Obu Ekpezu2Raynard Dodzi Helegah3Elikem Doe Atsakpo4Charles Nutrokpor5Fiifi Baffoe Payin Winful6Kafui Kwashie Solaga7Corresponding author.; Department of Computer Science, University of Ghana, Legon-Accra, GhanaDepartment of Computer Science, University of Ghana, Legon-Accra, GhanaDepartment of Computer Science, University of Ghana, Legon-Accra, GhanaDepartment of Computer Science, University of Ghana, Legon-Accra, GhanaDepartment of Computer Science, University of Ghana, Legon-Accra, GhanaDepartment of Computer Science, University of Ghana, Legon-Accra, GhanaDepartment of Computer Science, University of Ghana, Legon-Accra, GhanaDepartment of Computer Science, University of Ghana, Legon-Accra, GhanaAudio datasets are fundamental to the development of automatic speech-recognition (ASR) systems. However, the availability of a large corpus of audio datasets in low-resource languages (LRLs) is limited. This study addresses this gap by introducing audio speech datasets for five low-resource languages spoken in Ghana and parts of Togo. Specifically, it presents a 5000-hour speech corpus in Akan, Ewe, Dagbani, Dagaare, and Ikposo. Each language corpus includes 1000 h of validated audio speech recorded by their indigenous speakers. These audio recordings are spoken descriptions of 1000 culturally relevant images collected using a custom Android mobile application. To enhance the dataset’s utility in ASR and linguistic research 10 % of the audio recordings for each language were randomly selected and transcribed, resulting in approximately 100 h of transcription per language. This dataset represents a critical resource for preserving and documenting Ghanaian languages. It holds the potential for advancing speech and language technologies in these languages. Creating this audio dataset is the first step towards bridging the technological gap between high- and low-resource languages. Ethical guidelines were strictly followed throughout the data collection process and participants were given incentives for lending their voices to this study.http://www.sciencedirect.com/science/article/pii/S2352340925006043Speech-to-textSpeech synthesisLow-resource languagesNatural language processingText-to-speechSpeech datasets
spellingShingle Isaac Wiafe
Jamal-Deen Abdulai
Akon Obu Ekpezu
Raynard Dodzi Helegah
Elikem Doe Atsakpo
Charles Nutrokpor
Fiifi Baffoe Payin Winful
Kafui Kwashie Solaga
Advancing automatic speech recognition for low-resource ghanaian languages: Audio datasets for Akan, Ewe, Dagbani, Dagaare, and IkposoScience Data Bank
Data in Brief
Speech-to-text
Speech synthesis
Low-resource languages
Natural language processing
Text-to-speech
Speech datasets
title Advancing automatic speech recognition for low-resource ghanaian languages: Audio datasets for Akan, Ewe, Dagbani, Dagaare, and IkposoScience Data Bank
title_full Advancing automatic speech recognition for low-resource ghanaian languages: Audio datasets for Akan, Ewe, Dagbani, Dagaare, and IkposoScience Data Bank
title_fullStr Advancing automatic speech recognition for low-resource ghanaian languages: Audio datasets for Akan, Ewe, Dagbani, Dagaare, and IkposoScience Data Bank
title_full_unstemmed Advancing automatic speech recognition for low-resource ghanaian languages: Audio datasets for Akan, Ewe, Dagbani, Dagaare, and IkposoScience Data Bank
title_short Advancing automatic speech recognition for low-resource ghanaian languages: Audio datasets for Akan, Ewe, Dagbani, Dagaare, and IkposoScience Data Bank
title_sort advancing automatic speech recognition for low resource ghanaian languages audio datasets for akan ewe dagbani dagaare and ikpososcience data bank
topic Speech-to-text
Speech synthesis
Low-resource languages
Natural language processing
Text-to-speech
Speech datasets
url http://www.sciencedirect.com/science/article/pii/S2352340925006043
work_keys_str_mv AT isaacwiafe advancingautomaticspeechrecognitionforlowresourceghanaianlanguagesaudiodatasetsforakanewedagbanidagaareandikpososciencedatabank
AT jamaldeenabdulai advancingautomaticspeechrecognitionforlowresourceghanaianlanguagesaudiodatasetsforakanewedagbanidagaareandikpososciencedatabank
AT akonobuekpezu advancingautomaticspeechrecognitionforlowresourceghanaianlanguagesaudiodatasetsforakanewedagbanidagaareandikpososciencedatabank
AT raynarddodzihelegah advancingautomaticspeechrecognitionforlowresourceghanaianlanguagesaudiodatasetsforakanewedagbanidagaareandikpososciencedatabank
AT elikemdoeatsakpo advancingautomaticspeechrecognitionforlowresourceghanaianlanguagesaudiodatasetsforakanewedagbanidagaareandikpososciencedatabank
AT charlesnutrokpor advancingautomaticspeechrecognitionforlowresourceghanaianlanguagesaudiodatasetsforakanewedagbanidagaareandikpososciencedatabank
AT fiifibaffoepayinwinful advancingautomaticspeechrecognitionforlowresourceghanaianlanguagesaudiodatasetsforakanewedagbanidagaareandikpososciencedatabank
AT kafuikwashiesolaga advancingautomaticspeechrecognitionforlowresourceghanaianlanguagesaudiodatasetsforakanewedagbanidagaareandikpososciencedatabank