Nigerian language ID (12 languages + other), compact Whisper encoder
Encoder of openai/whisper-base with a 10-second window, fine-tuned for audio classification.
Labels: yor ibo hau ego anw efi ibb idu igl nup tiv urh other
Training
- Data: only the provided dataset Data-Science-Nigeria/dsn-2026-lid-train
- External data: none
- Class-balanced sampling, noise and phone-band augmentation, speed perturbation, EMA weights
- Weights saved in fp32
