Most eLearning or other multi-media localization projects involve transcription as the initial step leading toward a new voice recording in the target language or for the creation of subtitles for the video/film/animation.
Important considerations in the transcription component:
- Transcription is most often a two-step process → Audio/video content is transcribed as text in the source language first and then translated into a new foreign language script. This process produces the highest degree of accuracy
- A direct transcription of audio in one language to text in another language (single step) can be more affordable but lacks traceability and is prone to less QA scrutiny.
- Almost all multi-media applications require the production of time-stamped scripts so that precise, individual recordings or subtitles can be created.
- Speech-to- Text technologies can be fairly accurate and are certainly fast. However, the best of these technologies is only 85-90% accurate in English and still must be edited by an expert human linguist to achieve our quality standards.
For many projects, the human transcriber is ultimately faster in overall performance.