# PDF TO MP3

> Control the playback sentence by sentence. Get more time to think and understand the text much faster. Ask questions and get explanations for every sentence. Upload your documents and listen to them.

PDF TO MP3 uses Document Player to help people listen to documents while keeping their attention on the text. A listener can hear a sentence again, move to the next sentence at their own pace, turn on automatic playback, ask about the current sentence, or hear it explained in a different way.

New accounts start with a free plan that includes 3,000 words each month.

The website includes a hand-written example called "How Document Player works." It shows how the player works, how sentence playback helps with focus, and how the Ask and Explain tools help with hard text.

## Main pages

- [English home](https://pdftomp3.com/)
- [Pricing](https://pdftomp3.com/pricing/)
- [Examples](https://pdftomp3.com/examples/)
- [iPhone and Android app](https://pdftomp3.com/app/)
- [Connect your AI with MCP](https://pdftomp3.com/connect-ai/)

## Main facts

- People can upload a document or a link.
- The player can move through text one sentence at a time.
- Ask lets a person talk about the current sentence or ask for a translation.
- Explain gives a simpler explanation in the sentence's context.
- The player works on the web, iPhone, and Android.
- The phone apps can use AirPods and steering-wheel controls to move through sentences.
- People can download document and chapter audio as MP3 files.

## Connect your AI with MCP

The [setup guide](https://pdftomp3.com/connect-ai/) explains how to connect your AI. The connection name is PDFTOMP3-MCP. The MCP server address is `https://pdftomp3.com/mcp`. It uses Streamable HTTP. Add it in your AI app, then sign in to Document Player and approve the connection.

The `send_to_doc_player` tool sends text to the signed-in user's private browser reader. It takes `text`, an optional `title`, and an optional `request_id` to avoid duplicates when retrying. It returns a reader link. Open that link and press Play to listen sentence by sentence. Sending text does not start playback or open a window.

The setup page includes an example that plays without signing in. Listening to text sent by an AI needs a Document Player account; normal speech usage applies. Passages expire after 30 days. Each account can keep 100 passages, each with up to 100,000 characters.

## Codex skill plugin

The Connect your AI page also offers PDFTOMP3-Plugin as a ZIP for local Codex use. Add its marketplace folder, install pdftomp3-plugin, then ask Codex to set up PDFTOMP3-Plugin and connect your account. The plugin includes a setup skill, a reading skill, and a helper that calls the API directly. It needs Python 3.10 or later, with no extra packages. The helper saves sign-in tokens in a local plain text file in the user's data folder.

Codex reads the source and writes the spoken text in the chat. The helper sends chapters to the signed-in account and can request speech with the user's saved OpenAI or Gemini key. Preparing text starts no paid work. Users open the returned player link and press Play. This path needs no MCP connection in Codex.

## Canvas: listen without a PDFTOMP3 account

Canvas is a separate reader for text prepared by an AI. Read `/canvas/guide.html` for the public text API and `/canvas-api/info` for discovery. Create a canvas, upload final spoken text by chapter, and return its `url`. Never send a Gemini key to the Canvas API. The user enters it in the browser, which calls Gemini Live directly. Canvases do not expire. Shared links reveal text but never grant edit access. Speech starts on Play and prepares only the next two sentences. Word marks use estimates.

## Search files

- [Sitemap](https://pdftomp3.com/sitemap.xml)
- [Robot rules](https://pdftomp3.com/robots.txt)