---
license: cc-by-4.0
dataset_info:
  features:
  - name: audio
    dtype: audio
  - name: segment_timestamp
    dtype: string
  - name: word_timestamp
    dtype: string
  splits:
  - name: train
    num_bytes: 77463357.233
    num_examples: 1377
  download_size: 72126754
  dataset_size: 77463357.233
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---

# TUFS Bahasa Malaysia Speech Dataset (ms)

This dataset originated from [TUFS Open Language Resources](https://www.coelang.tufs.ac.jp/mt/vmod/) provided by the Tokyo University of Foreign Studies (TUFS).
It contains audio recordings and transcriptions of Bahasa Malaysia speech materials intended for language learning and research.

## Dataset Summary

The dataset includes:
- Audio files in `.mp3` format
- Machine-readable transcriptions in a structured prompt format
- Metadata such as duration and unique identifiers

Each entry represents a single audio utterance paired with its corresponding transcript in the Bahasa Malaysia language.