I experienced a critical issue in production where 95 customer calls were unexpectedly disconnected due to the call.in-progress.error-vapifault-deepgram-transcriber-failed error. I understand that outages can happen, which is why fallback mechanisms are critical; however, neither the transcriber fallback nor the fallback number was invoked. This failure in redundancy defeated the purpose of having safeguards in place and resulted in all affected calls being dropped, causing significant disruption to our system.

To ensure your transcriber fallback works as intended, verify your configuration includes the

fallbackPlan

property. Here’s an example of a correct setup via API:

json
"transcriber": {
  "provider": "deepgram",
  "model": "nova-3",
  "language": "en",
  "fallbackPlan": {
    "transcribers": [
      { "provider": "assembly-ai", "speechModel": "universal-streaming-multilingual", "language": "en" },
      { "provider": "azure", "language": "en-US" }
    ]
  }
}

Double-check your assistant’s configuration to ensure fallbacks are listed and ordered correctly, as missing or misconfigured fallbacks will cause calls to end with an error if the primary provider fails. For more details, see the [transcriber fallback documentation](https://docs.vapi.ai/customization/transcriber-fallback-plan). Source: - [Transcriber fallback configuration](https://docs.vapi.ai/customization/transcriber-fallback-plan)

Hi, Sorry for the inconvenience caused. There was an an outage on Vapi’s end, which is resolved now. For more details and updates, please refer to: [https://status.vapi.ai/incident/853120](https://status.vapi.ai/incident/853120) Thank you & have a lovely day! Best regards, Oshi Raghav Customer Support Team | Vapi