A Cookbook for Community-driven Data Collection of Impaired Speech in LowResource Languages

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Salihs, Sumaya Ahmed, Wiafe, Isaac, Abdulai, Jamal-Deen, Atsakpo, Elikem Doe, Ayoka, Gifty, Cave, Richard, Ekpezu, Akon Obu, Holloway, Catherine, Tomanek, Katrin, Winful, Fiifi Baffoe Payin
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866912462213742592
author Salihs, Sumaya Ahmed
Wiafe, Isaac
Abdulai, Jamal-Deen
Atsakpo, Elikem Doe
Ayoka, Gifty
Cave, Richard
Ekpezu, Akon Obu
Holloway, Catherine
Tomanek, Katrin
Winful, Fiifi Baffoe Payin
author_facet Salihs, Sumaya Ahmed
Wiafe, Isaac
Abdulai, Jamal-Deen
Atsakpo, Elikem Doe
Ayoka, Gifty
Cave, Richard
Ekpezu, Akon Obu
Holloway, Catherine
Tomanek, Katrin
Winful, Fiifi Baffoe Payin
contents This study presents an approach for collecting speech samples to build Automatic Speech Recognition (ASR) models for impaired speech, particularly, low-resource languages. It aims to democratize ASR technology and data collection by developing a "cookbook" of best practices and training for community-driven data collection and ASR model building. As a proof-of-concept, this study curated the first open-source dataset of impaired speech in Akan: a widely spoken indigenous language in Ghana. The study involved participants from diverse backgrounds with speech impairments. The resulting dataset, along with the cookbook and open-source tools, are publicly available to enable researchers and practitioners to create inclusive ASR technologies tailored to the unique needs of speech impaired individuals. In addition, this study presents the initial results of fine-tuning open-source ASR models to better recognize impaired speech in Akan.
format Preprint
id arxiv_https___arxiv_org_abs_2507_02428
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle A Cookbook for Community-driven Data Collection of Impaired Speech in LowResource Languages
Salihs, Sumaya Ahmed
Wiafe, Isaac
Abdulai, Jamal-Deen
Atsakpo, Elikem Doe
Ayoka, Gifty
Cave, Richard
Ekpezu, Akon Obu
Holloway, Catherine
Tomanek, Katrin
Winful, Fiifi Baffoe Payin
Computation and Language
This study presents an approach for collecting speech samples to build Automatic Speech Recognition (ASR) models for impaired speech, particularly, low-resource languages. It aims to democratize ASR technology and data collection by developing a "cookbook" of best practices and training for community-driven data collection and ASR model building. As a proof-of-concept, this study curated the first open-source dataset of impaired speech in Akan: a widely spoken indigenous language in Ghana. The study involved participants from diverse backgrounds with speech impairments. The resulting dataset, along with the cookbook and open-source tools, are publicly available to enable researchers and practitioners to create inclusive ASR technologies tailored to the unique needs of speech impaired individuals. In addition, this study presents the initial results of fine-tuning open-source ASR models to better recognize impaired speech in Akan.
title A Cookbook for Community-driven Data Collection of Impaired Speech in LowResource Languages
topic Computation and Language
url https://arxiv.org/abs/2507.02428