Dealing with Variation in Audio Description Scripts

Audio description scripts represent a text type structured into several parts helping the speaker produce their recording. The heterogeneous composition and formatting of these makes it difficult to describe linguistic features audio (AD) in one go. Hence, seems useful implement them corpus tool enabling analysis specificities each AD section. In this paper, 69 episodes from German television show serve as sample explore method for dealing with variation when preparing processing. our article, we offer short overview existing research on script corpora variation, present dataset tools used prepare data. We then outline analysed treatments applied. last section discuss results. Our leads us conclude that modifying original data sake implementation (e.g. changing features) is weighty step which may have unforeseen consequences: formal conveys more meaning than expected. Lay summary cultural technique blind people enjoy going movies despite disability. An additional soundtrack provides information necessary understand essential action film. such soundtracks are mostly written by authors specialising audiovisual translation. were interested texts, called scripts, e.g. use bold print or words entirely capital letters. conducted study reported article because wanted know sorts can be identified scripts. also extent some varying should modified improve quality. That why collected TV (we got rights research). These texts 13 over period 6 years. broadcaster only stable parameter. With help statistics tools, found out vary lot author another even time same author. order unify features. However, understood reducing not any case good option: voice talents responsible recording need clear sufficient how proceed. addition, writing creative activity locked strict rules.