README.md

metadata

dataset_info:
  features:
    - name: filename
      dtype:
        audio:
          sampling_rate: 16000
    - name: 'Y'
      dtype: string
  splits:
    - name: train
      num_bytes: 53095303942.242
      num_examples: 1635599
  download_size: 53282183764
  dataset_size: 53095303942.242

Malaya-Speech Speech-to-Text dataset

This dataset combined from semisupervised Google Speech-to-Text and private datasets.

Processing script https://github.com/mesolitica/malaya-speech/blob/master/pretrained-model/prepare-stt/prepare-malay-stt-train.ipynb

This repository is to centralize the dataset for https://malaya-speech.readthedocs.io/