metadata
language:
- bn
license: cc-by-4.0
configs:
- config_name: default
data_files:
- split: test
path: data/*.parquet
BanglaEval — evaluation materials for the BanglaBox paper
Anonymous artifact for double-blind review. This dataset contains every evaluation input, output and score
behind the paper's tables, so each reported number can be recomputed rather than taken on trust. The model,
code and normaliser are in Banglabox/chatterbox-bangla-tts; the corpus is Banglabox/bangla-corpus.
Repository layout
.
├── data/
│ └── eval-000.parquet
├── sentences_eval.csv
├── subjective_ratings.xlsx
├── stress_set/
├── register/
├── external_benchmarks/
└── scripts/
├── score_released.py
├── subjective_stats.py
└── bn_norm.py
Licence
CC BY 4.0 for the materials created by the authors. Baseline audio from commercial systems is included solely for research comparison. Identifying information is withheld for double-blind review.