add text labels and split into clean and unclean splits
The names of the files should correspond to the verse that’s being recited which is technically the label.
For example: [surahNum]_[ayahNum]_[id].wav
2_1_2321423.wav corresponds to the text in Surah 2, Verse 1.
The PR updates the labels column with the text for each audio file.
If an audio file follows a valid filename format and is not corrupt, it becomes part of the clean split. The labels correspond to the surah/ayah num of the audio file and that the labels were not validated further.
Otherwise, it becomes part of the unclean split. Additionally an is_clean label is added to each data point for better clarity.
To update the documentation using HuggingFace’s dataset template card, after merging this PR into main, the dataset card can be created: https://huggingface.co/docs/datasets/en/dataset_card