|
| 1 | +--- |
| 2 | +title: WebSocket Transport |
| 3 | +description: Stream audio directly via WebSockets for real-time, bidirectional communication |
| 4 | +slug: calls/websocket-transport |
| 5 | +--- |
| 6 | + |
| 7 | +# WebSocket Transport |
| 8 | + |
| 9 | +Vapi's WebSocket transport enables real-time, bidirectional audio communication directly between your application and Vapi's AI assistants. Unlike traditional phone or web calls, this transport method lets you stream raw audio data instantly with minimal latency. |
| 10 | + |
| 11 | +## Key Benefits |
| 12 | + |
| 13 | +- **Low Latency**: Direct streaming ensures minimal delays. |
| 14 | +- **Bidirectional Streaming**: Real-time audio flow in both directions. |
| 15 | +- **Easy Integration**: Compatible with any environment supporting WebSockets. |
| 16 | +- **Flexible Audio Formats**: Customize audio parameters such as sample rate. |
| 17 | +- **Automatic Sample Rate Conversion**: Seamlessly handles various audio rates. |
| 18 | + |
| 19 | +## Creating a WebSocket Call |
| 20 | + |
| 21 | +To initiate a call using WebSocket transport: |
| 22 | + |
| 23 | +```bash |
| 24 | +curl 'https://api.vapi.ai/call' \ |
| 25 | + -H 'authorization: Bearer YOUR_API_KEY' \ |
| 26 | + -H 'content-type: application/json' \ |
| 27 | + --data-raw '{ |
| 28 | + "assistant": { "assistantId": "YOUR_ASSISTANT_ID" }, |
| 29 | + "transport": { |
| 30 | + "provider": "vapi.websocket", |
| 31 | + "audioFormat": { |
| 32 | + "format": "pcm_s16le", |
| 33 | + "container": "raw", |
| 34 | + "sampleRate": 16000 |
| 35 | + } |
| 36 | + } |
| 37 | + }' |
| 38 | +``` |
| 39 | + |
| 40 | +### Sample API Response |
| 41 | + |
| 42 | +```json |
| 43 | +{ |
| 44 | + "id": "7420f27a-30fd-4f49-a995-5549ae7cc00d", |
| 45 | + "assistantId": "5b0a4a08-133c-4146-9315-0984f8c6be80", |
| 46 | + "type": "vapi.websocketCall", |
| 47 | + "createdAt": "2024-09-10T11:14:12.339Z", |
| 48 | + "updatedAt": "2024-09-10T11:14:12.339Z", |
| 49 | + "orgId": "eb166faa-7145-46ef-8044-589b47ae3b56", |
| 50 | + "cost": 0, |
| 51 | + "status": "queued", |
| 52 | + "transport": { |
| 53 | + "provider": "vapi.websocket", |
| 54 | + "websocketCallUrl": "wss://api.vapi.ai/7420f27a-30fd-4f49-a995-5549ae7cc00d/transport" |
| 55 | + } |
| 56 | +} |
| 57 | +``` |
| 58 | + |
| 59 | +## Audio Format Configuration |
| 60 | + |
| 61 | +When creating a WebSocket call, the audio format can be customized: |
| 62 | + |
| 63 | +| Parameter | Description | Default | |
| 64 | +|-------------|-------------------------|---------------------| |
| 65 | +| `format` | Audio encoding format | `pcm_s16le` (16-bit PCM) | |
| 66 | +| `container` | Audio container format | `raw` (Raw PCM) | |
| 67 | +| `sampleRate`| Sample rate in Hz | `16000` (16kHz) | |
| 68 | + |
| 69 | +Currently, Vapi supports only raw PCM (`pcm_s16le` with `raw` container). Additional formats may be supported in future updates. |
| 70 | + |
| 71 | +<Note> |
| 72 | +Vapi automatically converts sample rates as needed. You can stream audio at 8kHz, 44.1kHz, etc., and Vapi will handle conversions seamlessly. |
| 73 | +</Note> |
| 74 | + |
| 75 | +## Connecting to the WebSocket |
| 76 | + |
| 77 | +Use the WebSocket URL from the response to establish a connection: |
| 78 | + |
| 79 | +```javascript |
| 80 | +const socket = new WebSocket("wss://api.vapi.ai/7420f27a-30fd-4f49-a995-5549ae7cc00d/transport"); |
| 81 | + |
| 82 | +socket.onopen = () => console.log("WebSocket connection opened."); |
| 83 | +socket.onclose = () => console.log("WebSocket connection closed."); |
| 84 | +socket.onerror = (error) => console.error("WebSocket error:", error); |
| 85 | +``` |
| 86 | + |
| 87 | +## Sending and Receiving Data |
| 88 | + |
| 89 | +The WebSocket supports two types of messages: |
| 90 | + |
| 91 | +- **Binary audio data** (PCM, 16-bit signed little-endian) |
| 92 | +- **Text-based JSON control messages** |
| 93 | + |
| 94 | +### Sending Audio Data |
| 95 | + |
| 96 | +```javascript |
| 97 | +function sendAudioChunk(audioBuffer) { |
| 98 | + if (socket.readyState === WebSocket.OPEN) { |
| 99 | + socket.send(audioBuffer); |
| 100 | + } |
| 101 | +} |
| 102 | + |
| 103 | +navigator.mediaDevices.getUserMedia({ audio: true }).then(stream => { |
| 104 | + const audioContext = new AudioContext(); |
| 105 | + const source = audioContext.createMediaStreamSource(stream); |
| 106 | + const processor = audioContext.createScriptProcessor(1024, 1, 1); |
| 107 | + |
| 108 | + processor.onaudioprocess = (event) => { |
| 109 | + const pcmData = event.inputBuffer.getChannelData(0); |
| 110 | + const int16Data = new Int16Array(pcmData.length); |
| 111 | + |
| 112 | + for (let i = 0; i < pcmData.length; i++) { |
| 113 | + int16Data[i] = Math.max(-32768, Math.min(32767, pcmData[i] * 32768)); |
| 114 | + } |
| 115 | + |
| 116 | + sendAudioChunk(int16Data.buffer); |
| 117 | + }; |
| 118 | + |
| 119 | + source.connect(processor); |
| 120 | + processor.connect(audioContext.destination); |
| 121 | +}); |
| 122 | +``` |
| 123 | + |
| 124 | +### Receiving Data |
| 125 | + |
| 126 | +```javascript |
| 127 | +socket.onmessage = (event) => { |
| 128 | + if (event.data instanceof Blob) { |
| 129 | + event.data.arrayBuffer().then(buffer => { |
| 130 | + const audioData = new Int16Array(buffer); |
| 131 | + playAudio(audioData); |
| 132 | + }); |
| 133 | + } else { |
| 134 | + try { |
| 135 | + const message = JSON.parse(event.data); |
| 136 | + handleControlMessage(message); |
| 137 | + } catch (error) { |
| 138 | + console.error("Failed to parse message:", error); |
| 139 | + } |
| 140 | + } |
| 141 | +}; |
| 142 | +``` |
| 143 | + |
| 144 | +### Sending Control Messages |
| 145 | + |
| 146 | +```javascript |
| 147 | +function sendControlMessage(messageObj) { |
| 148 | + if (socket.readyState === WebSocket.OPEN) { |
| 149 | + socket.send(JSON.stringify(messageObj)); |
| 150 | + } |
| 151 | +} |
| 152 | + |
| 153 | +// Example: hangup call |
| 154 | +function hangupCall() { |
| 155 | + sendControlMessage({ type: "hangup" }); |
| 156 | +} |
| 157 | +``` |
| 158 | + |
| 159 | +## Ending the Call |
| 160 | + |
| 161 | +To gracefully end the WebSocket call: |
| 162 | + |
| 163 | +```javascript |
| 164 | +sendControlMessage({ type: "hangup" }); |
| 165 | +socket.close(); |
| 166 | +``` |
| 167 | + |
| 168 | +## Comparison: WebSocket Transport vs. Call Listen Feature |
| 169 | + |
| 170 | +Vapi provides two WebSocket options: |
| 171 | + |
| 172 | +| WebSocket Transport | Call Listen Feature | |
| 173 | +|-------------------------------------|------------------------------------| |
| 174 | +| Primary communication method | Secondary, monitoring-only channel | |
| 175 | +| Bidirectional audio streaming | Unidirectional (listen-only) | |
| 176 | +| Replaces phone/web as transport | Supplements existing calls | |
| 177 | +| Uses `provider: "vapi.websocket"` | Accessed via `monitor.listenUrl` | |
| 178 | + |
| 179 | +Refer to [Live Call Control](/calls/call-features) for more on the Call Listen feature. |
| 180 | + |
| 181 | +<Warning> |
| 182 | +When using WebSocket transport, phone-based parameters (`phoneNumber` or `phoneNumberId`) are not permitted. These methods are mutually exclusive. |
| 183 | +</Warning> |
| 184 | + |
0 commit comments