PluelyPluelyDocs

Settings & Configuration

Add a Custom Speech-to-Text Provider

Custom STT Providers in the Dev Space lets you "Create and manage custom STT providers. Configure endpoints, authentication, and response formats." This is how you bring your own transcription provider for Listen mode instead of relying on Pluely's managed transcription.

Adding a provider

  1. Open the Dev Space and go to Custom STT Providers.

  2. Click to add a new provider.

  3. Paste your provider's request as a **Curl Command *** — described as "The curl command to use with the STT provider." This is the exact request Pluely sends your audio through, including your endpoint and authentication.

  4. Fill in the **Response Content Path *** — the path to the transcribed text inside your provider's JSON response. The example value shown is:

    text

    meaning the transcript text sits directly at a field named text in the response — a much flatter path than a typical chat completion response, since transcription APIs commonly return the transcript as a single top-level field.

  5. Set the Streaming toggle depending on whether your provider streams the transcript incrementally as audio comes in, or returns it all at once.

  6. Save your provider.

Understanding the Response Content Path for STT

Speech-to-text APIs vary in how they shape their JSON response, just like AI providers do. The Response Content Path tells Pluely exactly where to find the transcript text once the audio has been processed. The example,

   text

points straight at a top-level text field. If your provider nests the transcript differently — inside a results array, for instance — adjust the path so it matches your provider's actual response shape.

Editing and deleting

Use Edit Provider to update a custom STT provider's curl command, response content path, or streaming toggle if your provider changes. Use Delete Provider to remove one you no longer use.

Why bring your own STT provider

Custom STT Providers exist for the same reason as Custom Providers (AI): control. If you already have a transcription provider account, or you want Listen mode's audio to go somewhere specific that you configured yourself, this is where you wire it in — with your own endpoint, your own authentication, and your own response shape.

Last updated 2026-07-10 · pluely.com