This app allows to load and correct transcripts for media such as video or audio.
The transcript that the app was made for where AI derived: the AI provides both:
- transcript
- translation
both are expected as SRT file divided by speakers such as:
1
00:00:00,031 --> 00:00:17,811
[SPEAKER_06]: Pasamos inmediatamente al tema 9, ...
The inputs are assumed in an S3: currently they need to be uploaded manually
The dataloader component:
- puts these videos in a Mongodb
- adds the statements per speaker and debate in a Solr search engine
- serves as a backend to connect to these two databases
The frontend component:
- has a search page to search in the speakers statements
- offers a media player page to compare transcript and translation with the media and annotate or correct
The app is setup via docker-compose. See documentation for the details.