ATC Asia Corpus: 3-Hour Air Traffic Control Audio from Taiwan

811 audio clips totaling 3 hours of real-world Air Traffic Control radio communications from Taiwan's major airports. This public subset is designed for fine-tuning Automatic Speech Recognition models in the Asia-Pacific aviation domain and was created by author hungC. The dataset was last updated on March 8, 2026.

Use Cases

• Fine-tuning Automatic Speech Recognition models based on real-world aviation radio communications.
• Training domain-specific language models for aviation terminology and phraseology.
• Benchmarking speech recognition performance in noisy, real-world operational environments.
• Studying communication patterns and accents in Asia-Pacific air traffic control.

Strengths

• Contains 811 real-world audio clips with an average length of 13.4 seconds.
• Provides 3 hours of domain-specific speech data from a defined geographic region (Taiwan).

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• The dataset is a 3-hour subset; the full corpus requires separate access.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

hungC on Hugging Face.

Collection Method

Likely recorded from real-world Air Traffic Control radio communications.

Time Range

null

Freshness

Last updated 2026-03-08 14:14:14; freshness should be verified.

Geography

Taiwan's major airports.