Websocket Streaming

Use the desired personality id under your account in the WebSocket URL.

wss://app.openhome.xyz/websocket/voice-stream/OPENHOME_API_KEY/PERSONALITY_ID

Here is an example:

wss://app.openhome.xyz/websocket/voice-stream/xyzsdsadasannfma/4727

​

OPENHOME_API_KEY

Checkout API Docs to get OPENHOME_API_KEY

​

PERSONALITY_ID

Checkout Get Personalities section to get your account’s personality ids.

• 4727: Represents the personality ID.
• Set the personality ID to 0 to skip this part. This will start the call with the default agent, OpenHome.

​

WebSocket Flow

​

Audio Data Format

Audio data sent to the WebSocket must adhere to the following specifications:

• Format: 16-bit PCM
• Sample Rate: 16000 Hz
• Encoding: Base64

Ensure that the audio is converted to this format before sending it to the WebSocket.

​

WebSocket Message Structure

​

Client to Server Messages

1. User to Server Text Messages

   {
     "data": "MESSAGE_CONTEXT",
     "type": "transcribed"
   }
   
   

2. User to Server Audio Messages

   {
   "data": "BASE64_ENCODED_AUDIO_MESSAGES",
   "type": "audio"
   }
   

​

Server to Client Messages

1. Server to User Text Messages

   {
       "data": {
           "content": "MESSAGE CONTENT",
           "live": true,
           "role": "assistant"
       },
       "type": "message"
   }
   

   live: true indicates live transcription or response logs.

   final: true provides finalized transcription and response messages.

2. Server to User Audio Messages

   {
       "data": "BASE64_ENCODED_AUDIO_MESSAGES",
       "type": "audio"
   }
   

​

EXAMPLES

Find More Examples on GitHub