Fix stale dataset metadata

Remove old 8-shard parquet files

Append four Darija audio transcript batches

Remove old single parquet shard

Shard dataset and keep audio sentence only

Update README.md

Append Tanger audio chunks

Append new Darija ASR chunks (+66 nouveaux, total=163)

fix: README.md pour activer le Dataset Viewer HuggingFace

Add dataset card (README.md)

Add Darija ASR subset (0-100)

initial commit