MRTV Voices: Over 130,000 Burmese News Audio Clips with Transcripts

MRTV Voices is a large-scale Burmese speech dataset built from publicly available news broadcasts and programs aired on Myanma Radio and Television (MRTV). It contains over 130,000 short audio clips, approximately 117 hours of audio, with aligned transcripts derived from auto-generated subtitles. The dataset was created by freococo and was last updated on 2025-06-23.

Use Cases

• Train automatic speech recognition models based on formal Burmese news audio.
• Fine-tune speech synthesis models based on clear pronunciation and pacing from official broadcasts.
• Analyze linguistic patterns in government and official media communications.
• Develop audio alignment or subtitle generation tools for Burmese content.

Strengths

• Contains over 130,000 audio clips, providing a substantial volume of data.
• Offers approximately 117 hours of audio, enabling extensive model training.
• Includes aligned transcripts, which are essential for supervised speech tasks.
• Source material features formal Burmese with clear enunciation, likely providing high-quality audio samples.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Data may reflect the geographic and editorial bias inherent to the state-run media source MRTV.
• Transcripts are derived from auto-generated subtitles, which may contain errors.

Provenance

Source

Myanma Radio and Television (MRTV), the official state-run media channel of Myanmar.

Collection Method

Built from publicly available news broadcasts and programs; transcripts derived from auto-generated subtitles.

Freshness

Last updated 2025-06-23 11:49:56; freshness should be verified.

Geography

Myanmar