PDF TO MP3 lets Codex and Claude subscribers turn documents into audiobooks, with speech billed separately

PDF TO MP3, a service lets people turn documents into audiobooks using an AI subscription they already pay for. It works with Codex or Claude. The speech itself is billed separately by Google or OpenAI.

The idea is simple. Your AI reads the document and writes the script of the audiobook. That script is sent to PDF TO MP3's server, which turns it into audio. The finished audiobook shows up in a library where you can listen in the app.

The service connects through MCP, short for Model Context Protocol, a way for AI tools to plug into outside services. According to the page, setup has three steps. First, connect your AI. In Codex, open Settings, then Plugins, then Add MCP server. Choose Streamable HTTP and paste the server address https://pdftomp3.com/mcp. Save, then choose Authenticate and connect your PDF TO MP3 account. The page also lists Claude as an option.

Second, add a speech key. This is an API key, a kind of password that lets a service bill your account for usage. The page says this key pays for speech only. You can pick OpenAI or Gemini as the speech provider. For Gemini, the page suggests enabling billing in the Gemini console to improve reliability.

Third, send your first document. Attach a PDF in your AI chat and ask: "Use the PDF TO MP3 MCP to turn this into an audiobook."

The page splits the costs in two. Your AI subscription covers document processing through MCP. Your API key covers speech generation, and OpenAI charges per usage. The page lists the PDF TO MP3 service itself as $0.

A demo video shows an example. The presenter attached a Google paper on recursive self-improvement in Codex and asked for an audiobook. Using voice mode, they asked the AI to explain every sentence of the paper in a simple way, with a male and a female speaker interacting. The video says you can be creative with the script, and that the AI can pick multiple voices for multiple speakers.