Skip to main content

Start Stream

Overview​

The start_stream action starts sending call media to your secure WebSocket endpoint. Use it for live transcription, monitoring, or real-time analysis. Place it in the actions array returned by your answer_url. MiniVoice executes actions in order, so put start_stream before the action that depends on its result and after any prompt or setup the caller should experience first.

Business Use Case​

Use start_stream when live audio is needed during the call, such as agent assist, live transcription, or monitoring. The stream target must be a secure WebSocket URL.

JSON Example​

{
  "actions": [
    {
      "start_stream": {
        "stream_url": "wss://media.example.com/minivoice",
        "track": "both"
      }
    }
  ]
}

Parameters​

FieldTypeRequiredDescription
typestringYesMust be start_stream when using typed action format.
stream_urlstringYesSecure WebSocket URL beginning with wss.
trackstringNoinbound, outbound, or both.

Execution Flow​

  1. MiniVoice validates stream_url and track.
  2. It opens media streaming to your WebSocket endpoint.
  3. Your server receives media for the selected track.
  4. Use stop_stream when you want to end streaming.

Webhook Behaviour​

No action-specific webhook is sent for start_stream. Your WebSocket server should log connection and media receipt.

Success Example​

A successful answer_url response includes the action in a valid actions array:

{
  "actions": [
    {
      "start_stream": {
        "stream_url": "wss://media.example.com/minivoice",
        "track": "both"
      }
    }
  ]
}

Your application should store the MiniVoice call ID from webhooks or API responses so you can correlate the action with the call.

Failure Behaviour​

Non-secure URLs are rejected. Streaming can fail if your WebSocket endpoint is unavailable or closes the connection.

Expected Result​

MiniVoice opens a media stream to the configured wss URL and sends the selected track while the call continues.

Testing​

Run a local WebSocket receiver through a secure tunnel, start a call, and confirm the receiver sees a connection for the selected track. During testing, log every webhook payload and compare the call ID, action timing, and final call status with the behavior you expected.

Integration guidance

Place this action in a short answer_url response first, then combine it with other actions after you have confirmed the single-action behaviour. Always log the answer_url response your application returned for the call. If the caller experience differs from what you expected, compare the logged response with the webhook timeline for that call ID.

When this action is part of a larger IVR, keep each step explicit. For example, play a prompt before collecting input, redirect after collecting input, and transfer only after your application has selected the final destination. This makes the flow easier to test and easier to change later.

For production integrations, build a fallback path. A caller may hang up, enter no digits, reach a busy destination, or disconnect before the flow completes. Your webhook handler should update the call record with the last known step so support teams can understand what happened.

Copy/Paste Examples​

Real Request​

Return this from your answer_url:

{
  "actions": [
    {
      "start_stream": {
        "stream_url": "wss://media.example.com/minivoice",
        "track": "both"
      }
    }
  ]
}

Real Response​

A valid answer_url response is the JSON action payload itself. MiniVoice accepts the response when it contains an actions array:

{
  "actions": [
    {
      "start_stream": {
        "stream_url": "wss://media.example.com/minivoice",
        "track": "both"
      }
    }
  ]
}

Real Webhook Example​

{
  "event": "call.completed",
  "created_at": "2026-05-29T12:10:00Z",
  "call": {
    "id": "call_123",
    "customer_id": "cust_123",
    "application_id": "app_123",
    "status": "completed",
    "direction": "inbound",
    "from": "+15551230001",
    "to": "+15551230002"
  },
  "variables": {},
  "data": {}
}

Common Use Case​

Use start_stream when live audio is needed during the call, such as agent assist, live transcription, or monitoring. The stream target must be a secure WebSocket URL.

Common Failure Case​

{
  "error": {
    "code": "answer_url_error",
    "message": "invalid action payload"
  }
}

Invalid fields or malformed JSON cause MiniVoice to reject the answer_url payload. Log the exact response your server returned while testing.