freococo/myanmar_spoken_corpus

Myanmar Spoken Corpus (Version 1.0)

Overview

Myanmar Spoken Corpus is a high-quality, open dataset of spoken Myanmar sentences designed to support NLP and ASR applications. The dataset focuses on providing clean and structured spoken language data for advancing Myanmar language technology.

Dataset Statistics

• Number of Rows:
  • Local Parquet file: 16,020,011 rows
  • Hugging Face Dataset Viewer: 15,728,640 rows
• File Size: 1.78 GB (Parquet format)

Notes:

• The difference in row counts is due to automatic processing by the Hugging Face platform, which may include deduplication or exclusion of invalid rows during dataset conversion.
• The original dataset with all rows can be downloaded directly from this repository.

Key Features

• 16 Million Spoken Sentences: Filtered and refined specifically for spoken Myanmar language.
• Cleaned and Standardized: Includes Unicode conversion, sentence segmentation, and deduplication.
• Free and Open: Released under a universal freedom license, suitable for commercial, academic, and personal use.

License

This dataset is released under a Universal Freedom License. You can use, modify, and share it freely for any purpose. However, crediting the dataset and its sources is appreciated.

Acknowledgments

This dataset is derived from publicly available sources, including:

• The Hugging Face FineWeb2 dataset, specifically its Myanmar subset.
• Original writers, speakers, and creators of the sentences.

The dataset has undergone significant manual and automated processing to create a distinct and valuable resource.

Dataset Structure

The dataset is stored in a Parquet format for efficient storage and usage. It includes the following fields:

• sentence_id: A unique identifier for each sentence.
• text: The spoken sentence text in Unicode Myanmar.

Disclaimer

This dataset may contain:

• Offensive or sensitive content: As it is derived from web-based sources.
• Incomplete or noisy sentences: Despite extensive cleaning, some issues may remain.

Users are encouraged to review and preprocess the dataset further if needed.

Future Plans

This is the first version of the dataset. Future updates will include:

• Written Myanmar sentences to expand its use cases.
• Larger datasets with higher-quality content for universal Myanmar language applications.

Usage Example

You can load the dataset using Python's pandas library:

import pandas as pd

# Load the dataset
df = pd.read_parquet("my_spoken_corpus.parquet")

# Display the first few rows
print(df.head())