This dataset contains over 500 hours of verified and transcribed Hausa speech recordings developed for training and evaluating automatic speech recognition (ASR) systems.
Hausa ASR Model Fine-Tuning: Verified speech recordings paired with transcriptions can be used to fine-tune ASR models for recognizing spoken Hausa, supporting speech recognition development for a low-resource language.
Hausa Speech Recognition Evaluation: Audio recordings and their verified transcriptions can be used to evaluate ASR model performance by comparing generated transcripts with reference texts, provided that evaluation samples are kept separate from training data.