Hi - your documentation says custom transcribers are supported (https://docs.vapi.ai/customization/custom-transcriber), but there's no way to configure this in the UI for an assistant. When I set a custom transcriber via the API

bash
curl -X PATCH https://api.vapi.ai/assistant/{our-assistant-id} \
     -H "Authorization: Bearer {our-bearer-token}" \
     -H "Content-Type: application/json" \
     -d '{
  "transcriber": {
    "provider": "custom-transcriber",
    "server": {
      "url": "{our-websocket-url-for-transcriber}"
    }
  }
}'

This returns a 200 OK status code. However your dashboard immediately throws a validation error saying "transcriber.provider must be one of the following values: assembly-ai, azure, deepgram, 11labs, gladia, google, openai, speechmatics, talkscriber, cartesia". Either the UI needs to support custom transcribers or at least be aware that we have set one over the api.

Hi @James Shelley I can help you to solve the issue your facing here, this truly my area of specialization. Let's have a chat now. Thanks

I can help you get this working. The 200 OK means your API call is valid, but the dashboard validator isn’t yet aware of the custom-transcriber option, so it throws the error even though the backend accepts it. Are you looking to manage everything through the dashboard, or is API-only configuration acceptable for your use case? @James Shelley

Hello, I'm facing the same problem, I'd be happy if one of you could help me out. @Vercel @PRESTIGEX

I can help you with this issues but gonna be a paid task

Okay, I'll be in touch if i decided to proceed with you. Thank you

Once you set a custom-transcriber in the API, you will have to continue modifying your assistant via API since our dashboard is unable to handle API exclusive features. The assistant will still work if the websocket URL is correct and is a valid configuration if it is tied to a phone number, but calling it will be the main way to test it.