DSN 2026 Nigerian Language ID (wav2vec2-base)

Fine-tuned facebook/wav2vec2-base (1 epoch) for 13-class spoken language identification: yor, ibo, hau, ego, anw, efi, ibb, idu, igl, nup, tiv, urh, other.

Training data: Data-Science-Nigeria/dsn-2026-lid-train (revision 87cf6d4b1c7599f97f44b6e6aade86438369a2b1), used under its license and attribution terms (see the dataset's ATTRIBUTION.md). No external data was used.

Input: 16 kHz mono audio, normalized with the included feature extractor. Validation (supplied split, no nup examples): accuracy 0.728, macro-F1 0.495.

Downloads last month
21
Safetensors
Model size
94.6M params
Tensor type
F32
ยท
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Model tree for Success44/dsn2026-lid-wav2vec2-base

Finetuned
(1026)
this model