UGSpeechData - Audio speech dataset of 5 Ghanaian languages - Akan, Ewe, Dagbani, Dagaare, and Ikposo

The dataset comprises of 5000 hours speech corpus in Akan, Ewe, Dagbani, Daagare, and Ikposo. Each language includes 1000 hours of audio speech from indigenous speakers of the language and 100 hours of transcription.

Link(s) to Data Assets

• Local Audios + AUDIO ID.csv

AUDIO_ID.csv Description

Column	Description
IMAGE_URL	Provides the relative path to the images in the folder
IMAGE_SRC_URL	Provides the source path to the actual image online
AUDIO_URL	Provides the relative path to the local audio language in the Local Audio folder
ORG_NAME	Identifies the institution coordinating the audio collection
PROJECT_NAME	Provides the name of the project
SPEAKER_ID	Provides the ID number of the individual describing the image
LOCALE	Provides the local language IETF BCP 47 language tag of the audio file
GENDER	Provides the individual providing the audio description gender
AGE	Provides the individual providing the audio description age
DEVICE	Identifies the device from which the audio recording was done
ENVIRONMENT	Identifies the space within which the audio was recorded
YEAR	The year in which the audio was recorded

Note: Local IDs

Locale ID	Name
ak_gh	Akan
dga_gh	Dagbani
dag_gh	Dagaare
ee_gh	Ewe
kpo_gh	Ikposo

CITATION

Wiafe, I., Abdulai, J., Ekpezu, A. O., Dodzi, R., Atsakpo, E. D., Nutrokpor, C., Winful, F. B. P., & Solaga, K. K. (2023). UGSPEECHDATA (Version 1.0.0) [Data set]. https://github.com/isaacwiafe/speech_data_ug