Smelt FAQs

Question 1

What input formats does Smelt support for conversion?

Accepted Answer

Smelt supports diverse input formats including PDF, DOCX, PPTX, XLSX, images, web pages, audio files, EPUBs, and even YouTube videos, making it a comprehensive solution for multimodal data processing.

Question 2

What output formats does Smelt provide for LLM-optimized text?

Accepted Answer

Smelt offers flexible output options for LLM-ready text in Markdown, Semantic XML, DocTags, and Structured JSON, catering to various integration and data consumption needs for AI applications.

Question 3

What is Smelt and what is its primary purpose?

Accepted Answer

Smelt is a versatile Python tool designed to convert a wide array of raw documents, images, web pages, audio, and videos into clean, structured text specifically optimized for large language models (LLMs).

Question 4

How does Smelt integrate with AI clients like Claude Desktop?

Accepted Answer

Smelt integrates seamlessly as an MCP (Multi-Client Protocol) tool. AI clients such as Claude Desktop can directly call Smelt's `convert` function, enabling powerful on-demand document and data processing within their workflows.

Question 5

Can Smelt handle different conversion quality or specific file types?

Accepted Answer

Yes, Smelt features a pluggable multi-backend architecture with automatic fallback chains. Users can install various backends (e.g., MinerU for PDF, faster-whisper for audio) to optimize for quality, performance (CPU/GPU), or specific file types, ensuring reliable conversions.

Smelt

Smelt

主要功能

使用案例

主要功能

使用案例