ATC ASR: 1,000-10,000 Utterance Pairs for Aviation Speech Recognition

Comprising between 1,000 and 10,000 audio-transcript pairs for Air Traffic Control speech recognition, compiled by user jacktol in 2025. It merges the UWB ATC Corpus and the ATCO2 1-Hour Test Subset into a fine-tuning-ready format. The records consist of cleanly segmented 16kHz .wav files paired with text utterances.

Use Cases

• Fine-tuning ASR models for aviation-specific terminology and callsigns
• Evaluating speech-to-text accuracy on 16kHz ATC audio segments
• Training acoustic models to handle high-noise cockpit and tower communication environments

Strengths

• Derived from two real-world ATC corpora (UWB and ATCO2)
• Cleanly segmented at the utterance level for immediate fine-tuning
• Standardized 16kHz audio sampling rate across all files

Limitations

• Small scale (1K-10K records) compared to general-purpose ASR datasets
• Domain-specific vocabulary is restricted to aviation and air traffic control terminology
• Potential for high background noise or radio interference typical of ATC environments

Provenance

Source

UWB ATC Corpus and ATCO2 1-Hour Test Subset

Collection Method

Curated and segmented from existing real-world ATC corpora

Freshness

Last updated August 2025.