A new Hacker News project called PDF TO MP3 lets users turn documents into audiobooks by connecting their existing AI subscription. The service is free to use, and the tool works through a Model Context Protocol (MCP) server that users add to Codex through its settings menu.
Setup involves three steps. Users open Settings, go to Plugins, and add an MCP server using Streamable HTTP with the provided server address. They then authenticate and connect their PDF TO MP3 account. Next, they add a speech key, which covers only the cost of generating audio. Finally, they attach a PDF to their AI chat and ask the assistant to use the PDF TO MP3 MCP to turn it into an audiobook.
Billing is split between two sources. Document processing runs through the user's subscription and is already paid for. Speech generation is billed separately by OpenAI, based on usage. Users who prefer not to supply their own key can use the developer's API key instead.
The project is positioned as a way to reduce costs for people who already pay for an AI subscription and want to listen to written material rather than read it. Because speech is billed per use, the total cost depends on the length of the documents converted.
Source: Hacker News · Summarized by HeadlinesBriefing