videoloc/seamless-basic
Updated
•
14
AI & ML interests
VIDEOLOC uses multimodal AI to improve automated subtitle generation. By combining audio and text encoders with quality estimation models, the system predicts editing effort and reranks ASR hypotheses to select the best candidates.
Recent Activity
Organization Card
VIDEOLOC
Welcome to the VIDEOLOC Hugging Face organisation. VIDEOLOC is a comprehensive system for video subtitle localisation combining ASR generation, multimodal quality estimation (Time-to-Edit prediction), and n-best reranking.
This organisation hosts the trained quality estimation models used in the VIDEOLOC pipeline. Additional resources, including the full codebase, datasets, and research papers, are being released progressively.
VIDEOLOC is funded by the European Union under the Italian PNRR, developed in collaboration with Pi School and FBK.
datasets
0
None public yet