This repository contains the second version of our Automatic Speech Recognition and Subtitle Generation model, with improved architecture and trained on 14000 hours of Flemish broadcast subtitled speech data.
It can generate both an exact verbatim transcription with annotation tags as well as a fully formatted and cleaned up subtitle transcription.
It outputs both modalities with separate decoders.
This repository contains the base variant of the model with 70M parameters.
This repository only hosts the pre-trained model itself and the configuration files.
To download this model, see the instructions
here.
Usage of this model, as well as our other ASR models, is integrated in
our Github codebase.
Please refer to the Github for installation.
This model can also be accessed through the
webservice of the NeLF Project. After requesting access, you can upload audio or video files and they will be transcribed according to the desired settings.
If you use this model, please cite the research paper: (Will be added shortly).
