Logo Lanfrica
  • Accueil
  • Atlas
  • Analyses
  • Documentation
  • Sign in

© 2026 Lanfrica. Tous droits réservés. Tous les droits d'auteur des ressources affichées sur le site Web Lanfrica appartiennent aux détenteurs de droits d'auteur d'origine, sauf indication contraire explicite.

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

Domaine:

natural language processing

Type de record:

paperdatasetmodel
Créateur:
SalWiaAbdAts
Hôte:avatar
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. This version has been reviewed and accepted for presentation at the InterSpeech 2025 conference to be held in Rotterdam from 17 to 21 August. 5 pages and 3 tables

Visit

arxiv.org

Tasks

automatic speech recognitionspeech processing

Languages

Akan

Tags

Computation and Language

Similaires

Atyap Afwan_: Preserving Tyap Through Community-Driven Speech DataSpeech-to-text-data-collection/STT-data-collectionSpeech Data Collection for The Nupe LanguageShruti-II: A vernacular speech recognition system in Bengali and an application for visually impaired communitySpeakerPool: A remote speech data collection platformAfroDigits: A Community-Driven Spoken Digit Dataset for African Languages

Atyap Afwan_: Preserving Tyap Through Community-Driven Speech Data

This dataset contains 98 recordings (≈1.16 hours) of everyday Tyap speech from 10 community speakers, each paired with detailed transcripts and English translations.

Speech-to-text-data-collection/STT-data-collection

A data engineering pipeline that allows recording millions of Amharic and Swahili speakers reading d

Speech Data Collection for The Nupe Language

This dataset contains audio recordings of the Nupe language. It features 1,583 audio recordings comprising 2 hours, 40 minutes, and 32 seconds of speech data, with paired transcripts. The recordings feature 8 unique speakers representing three distinct Nupe accent

Shruti-II: A vernacular speech recognition system in Bengali and an application for visually impaired community

SpeakerPool: A remote speech data collection platform

The collection of speech production data for academic use has traditionally been carried out by reco

AfroDigits: A Community-Driven Spoken Digit Dataset for African Languages

The advancement of speech technologies has been remarkable, yet its integration with African languages remains limited due to the scarcity of African speech corpora. To address this issue, we present AfroDigits, a minimalist, community-driven dataset of spoken digi