RATCHADA-STT Dataset

Overview

The dataset includes recordings from earnings calls of publicly traded companies in Thailand. Each audio file is accompanied by a transcription and metadata such as company name, reporting period, and other relevant details.

Dataset Info

• Total Duration:

  • Train: 26507.87 seconds (~ 7.36 hours)
  • Test: 8376.28 seconds (~ 2.33 hours)

• File Count:

  • Train: 10912 files
  • Test: 2804 files

Dataset Structure

The dataset consists of recordings from various financial presentations categorized into different subfolders based on event and company.

Metadata

• Features:

  • path: string
  • array: sequence of float32
  • label: string
  • start: float64
  • end: float64

• Dataset Size: 2.11 GB

Example Usage

Load the Ratchada-STT dataset

from datasets import load_dataset

dataset = load_dataset("ThinkingMachinesDataScience/Ratchada-STT")

Access train and test splits

train = dataset['train']
test = dataset['test']

Example usage with first sample

print(train[0])

License

This dataset is distributed under the MIT License.

Citation

If you use this dataset in your research or work, please cite it as:

@dataset{thinkingmachinesdatascience/ratchada-stt,
  title = {Ratchada-STT Dataset},
  publisher = {Hugging Face},
  year = {2024},
  author = {Thinking Machines Data Science},
  license = {MIT},
}

Contribution Guidelines

• Ensure your PR adheres to the repository's coding standards.
• Include relevant tests and ensure they pass.
• Update documentation if your contribution changes any functionality.
• Follow the Hugging Face Community Code of Conduct.