Bam_ASR_Eval_500 is a curated evaluation dataset for Automatic Speech Recognition (ASR) models in Bambara (Bamanakan), a major language spoken in Mali and West Africa. This dataset comprises 500 audio recordings totaling approximately 36.69 minutes of annotated speech, designed specifically for benchmarking ASR systems. It focuses on real-world challenges in low-resource languages like Bambara, including spontaneous speech… See the full description on the dataset page: https://huggingface.co/datasets/RobotsMali/Bam_ASR_Eval_500.
