---
title: Stream
description: >-
  StreamAction forks call audio to a WebSocket endpoint.  The socket must
  advertise the audio.drachtio.org subprotocol or the PBX handshake fails.
summary: >-
  StreamAction forks call audio to a WebSocket endpoint.  The socket must
  advertise the audio.drachtio.org subprotocol or the PBX handshake fails.
date: 2026-06-27T00:00:00.000Z
type: voice-action
---


## Summary

StreamAction forks call audio to a WebSocket endpoint.  The socket must advertise the audio.drachtio.org subprotocol or the PBX handshake fails.

## Parameters

| Parameter | Type | JSON key | Required | Description |
|-----------|------|----------|----------|-------------|
| `URL` | `string` | `url` | Yes | URL is the wss:// (or ws://) fork endpoint. Required. |
| `Track` | `string` | `track` | No | Track selects which leg is forked: inbound_track, outbound_track, or both_tracks. Empty means the PBX default. |
| `Connect` | `*bool` | `connect` | No | Connect answers and HOLDS the leg for the fork's lifetime.  Load-bearing: without it the fork is fire-and-forget on a document that ends immediately, and the call hangs up after about a second. Every production caller sets it. Tri-state so an unset value can defer to the PBX default rather than silently meaning false. |
| `OutputSamplingRate` | `int` | `outputSamplingRate` | No | OutputSamplingRate is the PCM rate delivered TO the socket: 8000 or 16000. Frames are 16-bit little-endian signed PCM. |
| `BidirectionalSamplingRate` | `int` | `bidirectionalSamplingRate` | No | BidirectionalSamplingRate is the rate the socket RETURNS audio at; the PBX resamples to the channel codec. Set it whenever the socket plays audio back, or playback behaviour is undefined. |
| `InputSamplingRate` | `int` | `inputSamplingRate` | No | InputSamplingRate is an informational hint about the source rate. |
| `Name` | `string` | `name` | No | Name is a friendly label echoed back in status events. |
| `Record` | `*bool` | `record` | No | Record requests recording of the forked leg. |
| `HeaderMetadata` | `string` | `headerMetadata` | No | HeaderMetadata is a flat JSON object string delivered in the FIRST WebSocket TEXT frame as its headers field — NOT as HTTP headers on the handshake, despite the name. |
| `OpenMetadata` | `string` | `openMetadata` | No | OpenMetadata is an opaque UTF-8 string delivered in the first TEXT frame. |
| `StatusCallback` | `string` | `statusCallback` | No | StatusCallback receives stream-started / stream-stopped / stream-error. |
| `StatusEvents` | `string` | `statusEvents` | No | StatusEvents is a space-separated subset of "stream-started stream-stopped stream-error". |


## Example

```xml
<Stream url="wss://your-app.example.com/audio" connect="true" track="both_tracks" bidirectionalSamplingRate="16000" />`}
  sdk={`import { sauti } from "@sautikit/node";

res.json(sauti.stream({
  url: "wss://your-app.example.com/audio",
  connect: true,
  track: "both_tracks",
  bidirectionalSamplingRate: 16000,
}));
```

```json
{
  "actions": [
    {
      "stream": {
        "url": "wss://your-app.example.com/audio",
        "connect": true,
        "track": "both_tracks",
        "bidirectionalSamplingRate": 16000
      }
    }
  ]
}
```

## See also

- [Voice Actions overview](/developers/voice-actions)
- [VoiceFlow DSL reference](/developers/concepts/voice-actions)
- [Webhooks catalogue](/developers/webhooks)