Multitask-National-Speech-Corpus (MNSC v1) is derived from IMDA's NSC Corpus.
MNSC is a multitask speech understanding dataset derived and further annotated from IMDA NSC Corpus. It focuses on the knowledge of Singapore's local accent, localised terms, and code-switching.
ASR: Automatic Speech Recognition
SQA: Speech Question Answering
SDS: Spoken Dialogue Summarization
PQA: Paralinguistic Question Answering
from datasets import load_dataset
data =… See the full description on the dataset page:
https://huggingface.co/datasets/MERaLiON/Multitask-National-Speech-Corpus-v1.
