jaketae/storyteller

Orchestrates a full generative pipeline where GPT produces multi-sentence narratives, Stable Diffusion generates sentence-aligned imagery, and TTS models create synchronized audio narration—composited into MP4 video with subtitles. Exposes both CLI and Python APIs with flexible device placement (CPU/CUDA/MPS), dtype control, and attention-slicing optimizations for memory-constrained environments. Supports pluggable model selection across all three modalities, enabling experimentation with different GPT, diffusion, and TTS checkpoints.

535 stars. No commits in the last 6 months.