openclaw-voice-gpt-realtime

Orchestrate real-time voice calls between Twilio and OpenAI Realtime.

4|Updated Feb 15, 2026

Search Tags:#voice #real-time #transcripts #twilio #ivr #ai-calling #openai-realtime

Ready to run

npx skills add https://github.com/ConnorCallison/openclaw-voice-gpt-realtime --skill openclaw-voice-gpt-realtime

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: openclaw-voice-gpt-realtime
Source: https://github.com/ConnorCallison/openclaw-voice-gpt-realtime/tree/main
Command: npx skills add https://github.com/ConnorCallison/openclaw-voice-gpt-realtime --skill openclaw-voice-gpt-realtime

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Orchestrate real-time voice conversations between users and businesses using OpenAI Realtime.

Core Features & Use Cases

End-to-end real-time voice calls: A single model handles speech-to-speech with minimal latency for tasks like restaurant reservations, store inquiries, and support calls.
Inbound and outbound call handling: Outbound tasks initiate calls; inbound calls can be accepted with configurable policies and greetings; DTMF navigation and voicemail handling are supported.
Structured outcomes & transcripts: Call results include confirmations, timestamps, and full transcripts for auditing and analytics.
Safety & governance: Prompts are wrapped with safety rules to prevent deception and protect user data.
Extensible architecture: WebSocket bridge between Twilio and OpenAI Realtime, optional assets/scripts/references; easy integration with OpenClaw.

Quick Start

Install the plugin, configure credentials, start a tunnel, verify status, and place a test call.

Frequently Asked Questions about openclaw-voice-gpt-realtime

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I build real-time AI voice calls using Twilio and OpenAI Realtime?▼

Real-time AI voice calls are orchestrated by bridging Twilio and OpenAI Realtime via WebSocket, handling speech-to-speech interactions with minimal latency for tasks like restaurant reservations and support calls.

Can I handle inbound calls and outbound calls with OpenAI Realtime?▼

Inbound calls and outbound calls are fully supported, allowing configurable greetings, DTMF navigation, and voicemail detection across real-time voice sessions.

What's the best way to get structured outcomes and transcripts from AI voice calls?▼

Structured outcomes and transcripts are generated automatically per call session, providing confirmations, timestamps, and full conversational logs for auditing and analytics.

Does this real-time voice approach support DTMF navigation and voicemail detection?▼

DTMF navigation and voicemail detection are supported natively, enabling automated IVR traversal and proper handling of unanswered outbound calls.

How do I securely bridge Twilio audio to OpenAI Realtime WebSocket sessions?▼

A WebSocket bridge connects Twilio and OpenAI Realtime, requiring secure public URL validation to ensure safe session management and data transmission.

What safety guardrails are needed for automated AI calling systems?▼

Prompts are wrapped with safety rules to prevent deception and protect user data, ensuring governance over automated real-time voice conversations.