freococo/2hr_myanmar_asr_raw_audio

🇲🇲 Raw 2-Hour Burmese ASR Audio Dataset

A ~2-hour Burmese (Myanmar language) ASR dataset featuring 1,612 audio clips with aligned transcripts, curated from official public-service educational broadcasts by FOEIM Academy — a civic media arm of FOEIM.ORG, operating under the Myanmar National Unity Government (NUG).

This dataset is MIT-licensed as a public good — a shared asset for the Burmese-speaking world. It serves speech technology, education, and cultural preservation efforts by and for the people of Myanmar.

➡️ Author: freococo
➡️ License: MIT (People’s Property)
➡️ Language: Burmese (my)

🎙️ Source Acknowledgement

All audio clips were sourced from FOEIM Academy, part of FOEIM.ORG, an official civic education platform affiliated with Myanmar’s National Unity Government (NUG).
FOEIM creates trusted public-service content to empower Myanmar citizens through ethical knowledge, truth-telling, and resilient education.

🎤 The original audio is publicly available on their YouTube channel,
with over 1,700 videos, 55.2K subscribers, and more than 6 million views since its founding on July 1, 2021.

All samples in this dataset were extracted and aligned under the MIT License, as a people’s resource — freely reusable for both research and commercial applications,
to support Burmese language technology, cultural preservation, and open public access.

🛑 Note: Audio may include background music, environmental noise, or thematic sounds present in the original public materials.

📦 Dataset Summary

• Total Clips: 1,612
• Total Duration: ~1.82 hours
• Chunk Length: 0.5s to 22.77s
• Avg. Clip Duration: 4.06 seconds
• Audio Format: .mp3 mono files
• Transcription: Aligned from .srt subtitle timestamps
• Source: All audio originated from the same speaker/channel for tonal and linguistic consistency

This dataset is ideal for:

• Training and fine-tuning Burmese ASR models
• Studying prosody and phrase timing in public education contexts
• Bootstrapping pronunciation tools for Myanmar-language speech technologies

🗂️ Dataset Structure

Each row in the metadata.csv file contains:

Column	Description
file_name	Relative path to the audio clip (e.g., audio/audio_20250604_0001.mp3)
transcript	Burmese-language transcription text
audio_duration	Duration of the clip in seconds (float)

All audio clips are stored in the audio/ folder as mono .mp3 files.

🧪 Tip: Use the datasets library with torchaudio or librosa to efficiently load and preprocess audio for training.

---

🔥 In Praise of the National Unity Government (NUG)

This dataset exists because of a revolution —
And in the heart of that revolution stands the National Unity Government (NUG).

💪 The NUG fights not only for political freedom —
but for linguistic survival, cultural dignity, and truth in every tongue.

It defends books as fiercely as borders,
and protects microphones as bravely as mountains.

🌱 It sponsors voices that once would’ve been silenced —
educators, activists, journalists, translators, dreamers.

🎓 Through FOEIM and its allies,
the NUG plants seeds of knowledge in every speech, every broadcast, every syllable.

This dataset is not just a tool —
it is a small fruit from a tree that grows in resistance, rains in courage, and blossoms in the soil of people’s will.

To the NUG:
📜 You are the government that teaches.
🕊️ You are the state that listens.
🎙️ You are the voice of Myanmar — and this dataset carries your echo.

For the people, by the people, through the people.

🇲🇲✨ Let the language of freedom never fall silent.

---

🔌 How to Load in Python

from datasets import load_dataset, Audio

ds = load_dataset("freococo/2hr_myanmar_asr_raw_audio")
ds = ds.cast_column("file_name", Audio())
ds[0]

📚 Citation

If you use this dataset in your research or application, please cite:

@dataset{freococo_myanmar_asr_2025,
title     = {2-Hour Burmese ASR Audio Dataset},
author    = {freococo},
year      = {2025},
url       = {https://huggingface.co/datasets/freococo/2hr_myanmar_asr_raw_audio},
note      = {Curated from FOEIM Academy public media (NUG). Licensed under MIT.}
}

---

📜 License

This dataset is licensed under the MIT License —
which means it is freely reusable for research, commercial use, and educational initiatives.

It is offered in the spirit of public service and open access,
as a shared inheritance of the Myanmar people, by a government of, by, and for the people.

> Please retain attribution when using or adapting this work.  
> This dataset is a **gift of voices** — may they always be heard.

---

💖 Final Note

This dataset is not just a file.
It is a memory of a nation speaking.
It is the sound of resilience echoing from every syllable.
It is the future of Burmese technology, born from struggle, shaped by community.

🇲🇲 *To the educators, rebels, poets, and freedom speakers — this is for you.*