ChenShuo2004/cs-board
将参考声音和中文文案自动生成白板动画视频的本地 AI 工具。
What it solves
cs-board is a local AI-powered video production workstation designed to transform text and audio into whiteboard-style animations. It eliminates the manual effort of storyboarding, drawing, and syncing audio with visuals, allowing users to create professional-speaking videos from a simple script and a reference audio clip.
How it works
The system follows a pipeline: it clones the voice from a reference audio file, breaks down the text script into scenes, generates consistent illustrations based on selected visual templates or reference images, and then renders the final MP4 video with hand-drawn animations and subtitles.
Key components include:
- Voice Cloning: Integrates with IndexTTS services (Gradio or FastAPI).
- Visual Templates: Offers 12 distinct styles (e.g., Minimalist, Cyber Neon, 3D Clay) to control the look and feel.
- Content Orchestration: Uses LLMs (via OpenLux API) to decompose text into storyboards and information graphics.
- Dynamic Infographics: Aligns visual elements to a phrase-level timeline based on the actual narration timing to prevent visuals from appearing too early.
Who it’s for
Content creators, educators, and business analysts who need to create knowledge-sharing videos, brand storytelling, or commercial analysis without needing professional animation skills.
Highlights
- Local Execution: Keeps assets, keys, and task history on the local machine.
- Consistent Characters: Supports uploading reference images for specific characters and styles to maintain visual continuity across a video.
- Local Voice Cloning: Voice cloning happens via local services, ensuring audio privacy.
- Local Overlays: Adds Chinese text overlays to avoid the common issue of AI image models generating garbled text.
- Checkpoint-based Recovery: Saves progress at each stage (audio, storyboard, images), allowing users to adjust render settings without restarting the entire process.
- LAN Collaboration: Allows multiple computers on the same local network to share a production queue.
Related
- Project
- Project
- Project
- Project
- Project