Transcription is not only a useful tool for audiovisual translation, but also a task that is being increasingly performed by translators in different scenarios. This article presents the results of an experiment in which three transcription methods are compared: manual transcription, respeaking, and revision of a transcript generated by speech recognition. The emphasis is put on respeaking, which is expected to be a useful method to speed up the process of transcription and have a positive impact on the transcribers' experience. Both objective and subjective measures were obtained: on the one hand, the time spent on each task and the output quality based on the NER metrics and on the other, the participants' opinions before and after the task, namely the self-reported effort, boredom, confidence in the accuracy of the transcript and overall quality.
|Publication status||Published - 1 Jan 2017|
- Audiovisual translation
- Speech recognition