Datasetpublic

tamm5y5m5/quran_nemotron_dual_v5_npl

Quran Nemotron Dual V5 Assets Reproducibility assets for the Arabic-only Hafs Quran Nemotron dual-output training pipeline. The acoustic RNNT target and the Quran-Lab tj1 phoneme CTC target are stored independently. This repository does not grant a new license over source recordings. Every row contains its source repository and pinned revision. Source-specific terms and the Quran-Lab NPL-1.1 non-profit/share-alike restrictions continue to apply. The bootstrap_v4 directory… See the full description on the dataset page: https://huggingface.co/datasets/tamm5y5m5/quran_nemotron_dual_v5_npl.

sourceHugging Faceotherupdated 25d agoView on Hugging Face

0likes1.3kdownloads

Dataset Card

Quran Nemotron Dual V5 Assets

Reproducibility assets for the Arabic-only Hafs Quran Nemotron dual-output training pipeline. The acoustic RNNT target and the Quran-Lab tj1 phoneme CTC target are stored independently.

This repository does not grant a new license over source recordings. Every row contains its source repository and pinned revision. Source-specific terms and the Quran-Lab NPL-1.1 non-profit/share-alike restrictions continue to apply.

The bootstrap_v4 directory contains the verified, speaker-disjoint v4 corpus converted to schema version 5. Phone recordings and other added sources are published only after their access, licensing, deduplication, and leakage gates complete.

Evaluation benchmark audio is never included in training shards.