How do you get the transcription from voice-call plugin set up with a realtime openAI gpt-realtime-1.5 model?
"voice-call": {
"config": {
"fromNumber": "+1111111111",
"inboundPolicy": "allowlist",
"allowFrom": [
"+1111111111"
],
"inboundGreeting": "Hi XX! How can I help?",
"outbound": {
"defaultMode": "conversation"
},
"provider": "twilio",
"publicUrl": "https://openclaw.xxx.com/voice/webhook",
"realtime": {
"enabled": true,
"provider": "openai",
"toolPolicy": "owner",
"instructions": "Speak briefly and conversationally. Call openclaw_agent_consult before using deeper tools.",
"providers": {
"openai": {
"apiKey": "sk-proj-xxx",
"model": "gpt-realtime-1.5",
"voice": "alloy"
}
}
},
"twilio": {
"accountSid": "xxx",
"authToken": "xxx"
},
"webhookSecurity": {
"allowedHosts": [
"openclaw.xxx.com"
]
}
},
"enabled": true
}