Automated subtitling by means of speech and language technology
Ston is an innovative procurement project that tries to partly automatize the subtitling workflow by means of speech and language technology. This requires different technological components:
- Speech detection: when are they speaking?
- Speaker clustering: who speaks when?
- Language detection: which language is being spoken?
- Speech recognizer: what is being said?
- Synchronization module for the alignment of script on video
- Post processing to stylize the subtitles according to the VRT standard
A cockpit is needed to be developed to steer all components, to adjust the configuration to the needs of every program format and verify the results.
The focus within the STON-project lies on the automatized subtitling of documentaries and online news content. The results of the project are also very valuable to other applications, such as the automatized annotating of content.