You need to agree to share your contact information to access this dataset

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

Log in or Sign Up to review the conditions and access this dataset content.

All Lab Speech

Cleaned African-language speech with embedded, playable audio (HF Audio). One config per language (<lang> = transcribed, <lang>_manifest = audio-only); the audio column sits right after audio_id and plays in the dataset viewer. Splits (train/validation/test) come from the source split labels.

from datasets import load_dataset
ds = load_dataset("African-Languages-Lab/all-lab-speech", "afrikaans")

Columns

audio_id, audio (playable), transcript, duration_sec, language, dialect, speaker_id, gender, age, split, utterance_type, task, collection_method, transcript_token_count, source_data_url.

Configs (71)

afrikaans, afrikaans_manifest, amharic, amharic_manifest, arabic, bambara, bambara_manifest, bemba, bemba_manifest, berber, berber_manifest, chichewa, chichewa_manifest, english, english_manifest, ewe, ewe_manifest, fon, fula, fula_manifest, hausa, hausa_manifest, igbo, igbo_manifest, kanuri, kikuyu, kikuyu_manifest, kinyarwanda, kinyarwanda_manifest, krio, krio_manifest, lingala, lingala_manifest, luganda, luganda_manifest, malagasy, malagasy_manifest, ndebele, ndebele_manifest, oromo, oromo_manifest, sepedi, sesotho, sesotho_manifest, shona, shona_manifest, somali, somali_manifest, swahili, swahili_manifest, swati, swati_manifest, tigrinya, tigrinya_manifest, tsonga, tsonga_manifest, tswana, tswana_manifest, twi, twi_manifest, umbundu, venda, venda_manifest, wolof, wolof_manifest, xhosa, xhosa_manifest, yoruba, yoruba_manifest, zulu, zulu_manifest

Downloads last month
33