Datasets:

Datasmartly
/

audios-tamazight2-with-speakers

Tasks:

Audio Classification

Modalities:

Formats:

Size:

Libraries:

Dataset card Data Studio Files Files and versions

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.

Datasmartly/audios-tamazight2-with-speakers

Dataset enrichi automatiquement avec speaker_name et speaker_id (cohérents globalement par split) via:

Embeddings : speechbrain ECAPA-TDNN @16 kHz
Clustering global des segments >= 0.5s (KMeans, K=auto via silhouette, max 10)
Post-assignation des segments courts (>= 0.1s) par similarité cosinus aux centroïdes (padding jusqu'à 0.8s).

Source: Datasmartly/audios-tamazight2 Splits: train

Downloads last month: 15