MulmoCast：用脚本生成多格式演示内容

AI-powered podcast & video generator.

MulmoCast: A Multi-Modal Presentation Tool for the AI-Native Era Quick Start Guide If you want to try our beta version, follow the instruction in the release note below. MulmoCast 0.1.x Beta Release Notes MulmoCast 0.1.x Beta版 リリースノート What is MulmoCast? Note: The video version of "What is MulmoCast?" is available on X, which is created by MulmoCast itself. MulmoCast is a next-generation presentation platform, purpose-built for a world where AI and humans collaborate to create and share ideas. Unlike traditional tools like PowerPoint or Keynote—both designed in the pre-AI era for human creators—MulmoCast is AI-native: reimagined from the ground up to work hand-in-hand with generative AI that understands and produces natural language, images, audio, and video. As AI takes on an increasingly central role in content creation, it needs more than raw capability—it needs an environment designed for its strengths. MulmoCast provides that environment, empowering AI to generate rich, coherent presentations across multiple formats, while keeping humans in the creative loop. At the core of MulmoCast is MulmoScript, a JSON-based intermediate language that functions like a screenplay or web markup. Generated by large language models, MulmoScript defines everything from narrative structure to visuals, making it easy to produce presentations that adapt seamlessly across formats. flowchart TD S_TextEditor["Creator"] <--> S_AIChat2("LLM") S_AIChat2 --> MulmoScript["MulmoScript"] MulmoScript --> MulmoCast["MulmoCast"] MulmoCast --> O_Video("Video") & O_Podcast("Podcast") & O_Slideshow("Slideshow") & O_PDF("PDF") & O_Manga("Manga") & O_SwipeAnime("Swipe Anime") MulmoScript@{ shape: div-proc} MulmoCast@{ shape: trap-t} S_TextEditor:::StoryInput S_AIChat2:::ScriptInput O_Video:::ScriptOutput O_Podcast:::ScriptOutput O_Slideshow:::ScriptOutput O_PDF:::ScriptOutput O_Manga:::ScriptOutput O_SwipeAnime:::ScriptOutput classDef StoryInput stroke-width:1px, stroke-dasharray:none, stroke:#46EDC8, fill:#DEFFF8, color:#378E7A classDef ScriptInput stroke-width:1px, stroke-dasharray:none, stroke:#374D7C, fill:#E2EBFF, color:#374D7C classDef ScriptOutput stroke-width:1px, stroke-dasharray:none, stroke:#FBB35A, fill:#FFEFDB, color:#8F632D Loading In today’s world, people consume information through many channels—on Zoom calls, in Slack threads, via videos on the go, or through podcasts during commutes. Presentations can no longer be confined to a single format or device. That’s why MulmoCast is built to be truly multi-modal, enabling output not only as slide decks but also as videos, podcasts, manga-style comics, and PDF documents. Whether you're building a podcast from a research report or turning a business pitch into a narrated video, MulmoCast transforms source material into dynamic, multi-format content with the help of state-of-the-art AI like ChatGPT and Claude. MulmoCast isn't just a tool. It’s a new foundation for communication—one where humans and AI co-create across every mode of expression. MulmoScript MulmoScript is a simple JSON/YAML format for describing multi-modal content. You can define speakers, text, images, and layout — all in one script. Here is the "Hello World" in MulmoScript. { "$mulmocast": { "version": "1.1" }, "beats": [ { "text": "Hello World" } ] } See MulmoScript Format for details on the structure. Installation npm install -g mulmocast You'll also need to install ffmpeg: # For macOS with Homebrew brew install ffmpeg # For other platforms # Visit https://ffmpeg.org/download.html You can also use Dockerfile which helps you install the pre-requisits. docker build -t mulmo-cli . You can use the Docker image like this: docker run -e OPENAI_API_KEY=<your_openai_api_key> -it mulmo-cli mulmo tool scripting -i -t children_book -o ./ -s story Configuration Create a .env file in your project directory with the following API keys: Required OPENAI_API_KEY=your_openai_api_key (Optional) For Google's image and TTS. GEMINI_API_KEY=your_google_gemini_api_key (Optional) For AI providers # For Anthropic Claude (htmlPrompt feature) ANTHROPIC_API_TOKEN=your_anthropic_api_token For htmlPrompt configuration, see docs/image.md. (Optional) For Movie models REPLICATE_API_TOKEN=your_replicate_api_key (Optional) For TTS models # For ElevenLabs TTS ELEVENLABS_API_KEY=your_elevenlabs_api_key (Optional) to access web in mulmo tool BROWSERLESS_API_TOKEN=your_browserless_api_token # to access web in mulmo tool (Optional) Local usage tracking MULMOCAST_DUMP_USAGE=1 # print JSON usage dump to stdout after each CLI action MULMOCAST_DUMP_USAGE=/tmp/usage.json # ...or write it to a file Token / per-second / per-char usage per provider:model. See docs/api.md for the full reference. Google Vertex AI For enterprise/production environments, or to use Veo movie models that are only available on Vertex AI, use Vertex AI with Application Default Credentials (ADC): # Install gcloud CLI and authenticate gcloud auth application-default login Configure in MulmoScript: { "imageParams": { "provider": "google", "model": "gemini-2.5-flash-image", "vertexai_project": "your-project-id", "vertexai_location": "us-central1" } } Parameter Description Default vertexai_project Google Cloud Project ID None (enables Vertex AI mode when set) vertexai_location Region us-central1 For detailed setup instructions, see Vertex AI Setup Guide. Azure OpenAI To use Azure OpenAI instead of OpenAI API: # For image generation IMAGE_OPENAI_API_KEY=<your-azure-openai-api-key> IMAGE_OPENAI_BASE_URL=https://<resource-name>.openai.azure.com/ # For TTS (Text-to-Speech) TTS_OPENAI_API_KEY=<your-azure-openai-api-key> TTS_OPENAI_BASE_URL=https://<resource-name>.openai.azure.com/ # For LLM (translate, scripting) LLM_OPENAI_API_KEY=<your-azure-openai-api-key> LLM_OPENAI_BASE_URL=https://<resource-name>.openai.azure.com/ LLM_OPENAI_API_VERSION=2025-04-01-preview # optional MulmoScript configuration (same as OpenAI): { "imageParams": { "provider": "openai", "model": "gpt-image-1