About the corpus

Each entry begins as an off-air recording of an international AM shortwave broadcast, captured through a remote receiver at a known frequency, time and location. The audio is transcribed automatically into timestamped passages, in the language actually spoken on air. Each transcript is then translated into English and summarized with the help of AI.

Limits

Transcription of shortwave reception is imperfect: fading, interference and accents produce errors. Treat every passage as an indication and double-check it before quoting it.