Datasetpublic

SPEAK-PP/openslr-sinhala-spelling-correction-prediction-reference-60000

OpenSLR Sinhala Spelling Correction – Prediction / Reference This dataset contains Sinhala sentence pairs intended for training and evaluating spelling-correction models on ASR output. Column Description dyslexic_sentence Noisy / dyslexic sentence (model input – ASR hypothesis) clean_sentence Clean / correct sentence (ground truth) Splits Split Source file Rows (approx.) train openslr-60000.csv ~60 000 eval openslr-7000.csv ~7 000… See the full description on the dataset page: https://huggingface.co/datasets/SPEAK-PP/openslr-sinhala-spelling-correction-prediction-reference-60000.

sourceHugging Facecc-by-4.0updated 7mo agoView on Hugging Face

0likes21downloads

Dataset Card

OpenSLR Sinhala Spelling Correction – Prediction / Reference

This dataset contains Sinhala sentence pairs intended for training and evaluating spelling-correction models on ASR output.

Column	Description
`dyslexic_sentence`	Noisy / dyslexic sentence (model input – ASR hypothesis)
`clean_sentence`	Clean / correct sentence (ground truth)

Splits

Split	Source file	Rows (approx.)
train	openslr-60000.csv	~60 000
eval	openslr-7000.csv	~7 000

Usage

python

from datasets import load_dataset

ds = load_dataset("SPEAK-PP/openslr-sinhala-spelling-correction-prediction-reference")
print(ds["train"][0])
# {'dyslexic_sentence': '...', 'clean_sentence': '...'}

Citation

If you use this dataset please cite the original OpenSLR Sinhala corpus and the SPEAK-PP spelling-corrector project.