Abstract

Over the last few years, the audiovisual translation (AVT) market has been transformed by the advent of cloud tools and the integration of translation technologies into AVT workflows. More specifically, machine translation (MT) and automatic speech recognition (ASR) tools are changing the localisation process for films, TV series and other video materials. However, as these are very recent developments, there is very little research on the effectiveness of such tools and the potential productivity gains. This paper presents the ¡Sub! and ¡Sub!2 projects, which compared three cloud subtitling workflows with different degrees of automation in a series of experiments. The aim was to determine the workflow with the best quality/ turnaround time ratio in relation to the audiovisual material in question, i.e., science documentaries. The relevance of key findings is discussed in relation to training and to the varied set of skills and competences that trainee subtitlers need to acquire to operate in today’s rapidly evolving market.