Play.ht MCP. Convert Text to Professional Audio Files Fast

Look, this Play.ht MCP Server handles turning plain text into professional audio files using their neural voice engine. It's built to let your AI client do the heavy lifting—you just point it at the server.

To get started, you first need to check what voices are available. You'll use the get_voices tool; this calls up a structured list of every single voice Play.ht has in its library. It gives you metadata for each one, including their unique IDs and what languages they support. This is how you figure out which voice fits your project.

Once you've got the right Voice ID, you can actually generate the audio. You call convert_tts, feeding it three things: the text you want spoken, that specific Voice ID, and any parameters for quality or format. The tool then starts processing, turning that written script into either an MP3 or a WAV file.

Since these aren't instantaneous jobs, they run in the background. You don't just call convert_tts and assume you got the final file; it's a multi-step process. After submitting your text for conversion, the server gives you a unique request ID. This is key because it tells your AI client what to watch for next.

If you need to know if the audio job finished or if something went wrong, you use get_tts_status. You just pass that unique request ID into this tool, and it checks the status—it'll tell you if the job is still pending, if it's done, or if it failed. This lets your agent wait for confirmation before trying to pull down the final audio file.

It’s basically a three-step loop: first, check get_voices for options; second, run convert_tts with text and voice ID; and third, constantly monitor get_tts_status using the returned request ID until you can grab your finished MP3 or WAV file.

Using this setup means you don't have to manually manage API calls. Your AI client handles the whole sequence. It grabs the list of voices first, making sure it knows all the available IDs and languages before sending a single character of text for conversion. When that job is submitted via convert_tts, your agent gets back that unique tracker ID.

You can then feed that ID into get_tts_status repeatedly. This process keeps your workflow tight because you're never guessing if the file is ready; you just ask the server, and it tells you exactly where it stands.

If you need to build out video narrations, this handles everything from script text to final audio file download. Developers can integrate realistic speech synthesis directly into their own apps without having to deal with manual API scheduling or polling. Accessibility teams find it useful because they can quickly turn large documents or reports into clear, audible speech for people who rely on that format.

It's a complete pipeline: discover voices, submit text, track status, and get the file.