Efficiency of Speech Alignment for Semi-automated Subtitling in Dutch

This paper describes the use of speech alignment to aid in the process of subtitling Dutch TV programs. The recognizer aligns the audio stream with an existing transcript. The goal is therefore not to transcribe but to generate the correct timing of every word. The system performs subtasks such as audio segmentation, transcript preprocessing, alignment and subtitle compression. The result is not perfect but good enough to gain efficiency when used by a professional subtitler as a starting point to refine and finalize the subtitles. In our tests, considerable time savings of 47 to 53% on average are obtained, such that the generation of subtitles for a 1 hour program, is lowered from between 4 and 7 hours to between 2.5 and 4 hours. This is all the more important in the context of an increased pressure from user groups on governments and broadcasters to reach 100% subtitled TV programs.