European Parliament Interpreting Corpus (EPIC)

EPIC v2.0 is a parallel and trilingual (English, Italian, and Spanish) corpus of European Parliament (EP) speeches and their simultaneous interpretations. The data were collected from EP sessions in Feb–Apr, and July 2004, including the speeches of 175 speakers and an unknown number of interpreters. The current version of the EPIC (v2.0) contains 692,585 tokens (source: 247,385, target: 445,200) and 83 h 36 min 14 s of audiovisual recordings (source videos: 27 h 32 min 20 s, target audio: 56 h 3 min 54 s). It is fully transcribed, annotated, and aligned at the recording–transcript level.