The dataset contains Urdu Audio Clips and their transcriptions in Urdu. Clips are on average 10 seconds each. Some Clips have transcriptions: Could not Understand Audio specifying a null audio file or a file containing only music