metadata
license: apache-2.0
FunCineForge: A Unified Dataset Pipeline and Model for Zero-Shot Movie Dubbing in Diverse Cinematic Scenes
FunCineForge dataset pipeline is a fully open-source, locally deployed tool for producing multimodal speech datasets. It integrates batches of film or television data from the source into comprehensive data including text, speech, video, clues, timestamps, and other information for training our VTTS dubbing LLM. All pre-trained models have been uploaded to Hugging Face. You can access https://anonymous.4open.science/w/FunCineForge/ to get our dataset samples and demo samples.