SIP4AI is a VOIP softphone where AI is on the other end, not a human. Self-host it, register a SIP line, and let AI take the calls. Works with your existing phone system — just add a line.
Full SIP protocol support, multiple AI providers, function calling, and a complete HTTP API for programmatic control.
Answer incoming calls automatically or make outbound calls with task-based objectives. Full control over caller ID and duration limits.
AI can call your webhooks mid-conversation. Book appointments, check orders, or transfer calls — all through simple HTTP endpoints.
Play custom audio while executing webhooks. Keep callers engaged with professional hold messages during function processing.
Blind and attended transfers via SIP REFER. Hand off to human agents seamlessly with consultation mode support.
Full HTTP API for programmatic control. Initiate calls, monitor status, inject context mid-conversation, and receive webhook events.
μ-law 8kHz native support with ElevenLabs and Cartesia. Full codec conversion for all providers including PCMU, PCMA, and G.729.
Choose the provider that fits your needs. From OpenAI's advanced reasoning to ElevenLabs' native μ-law support.
GPT-4o powered realtime conversations with advanced reasoning capabilities. Built-in function calling and server-side VAD with configurable thresholds.
{
"provider": "openai",
"openai": {
"instructions": "You are a helpful AI...",
"voice": "shimmer",
"turn_detection": {
"type": "server_vad",
"threshold": 0.7,
"silence_duration_ms": 1000
}
}
}Native μ-law 8kHz support means zero audio conversion — the lowest possible latency. Configure your agent in the ElevenLabs dashboard with client-side tools.
{
"provider": "elevenlabs",
"elevenlabs": {
"agent_id": "your-agent-id",
"first_message": "Hello! How can I help?",
"system_prompt": "You are a helpful..."
}
}Google's Gemini 2.0 Flash with built-in function calling. Five distinct voices to choose from, perfect for Google ecosystem integration.
{
"provider": "gemini",
"gemini": {
"model": "models/gemini-2.0-flash-exp",
"voice": "Kore",
"first_message": "Hello!",
"system_prompt": "You are a helpful..."
}
}Flexible component selection with Nova-3 STT and configurable LLM. 13 Aura-2 voices available with excellent latency characteristics.
{
"provider": "deepgram",
"deepgram": {
"model": "aura-2-thalia-en",
"listen_model": "nova-3",
"think_provider": "open_ai",
"think_model": "gpt-4o-mini",
"greeting": "Hello!"
}
}Hybrid architecture combining Deepgram STT, OpenAI LLM, and Cartesia Sonic TTS. Premium voice synthesis with native μ-law support.
{
"provider": "cartesia",
"cartesia": {
"model": "sonic-2024-10-19",
"voice_id": "694f9389-aac1...",
"stt_provider": "deepgram",
"llm_provider": "openai",
"llm_model": "gpt-4o-mini"
}
}A single Go binary handles everything: SIP signaling, RTP media, audio conversion, and WebSocket connections to AI providers.
Phone network via SIP/RTP
Go binary handles all conversion
WebSocket to voice AI
Initiate calls, monitor status, inject context mid-conversation, and receive webhook events. Everything you need for programmatic control.
/api/calls
Initiate an outbound call with task and instructions
/api/calls/{id}
Get call status, duration, and results
/api/calls/{id}/inject
Inject context into an active conversation
/api/calls/{id}
Hang up an active call
POST /api/calls
{
"target": "+61400000000",
"task": "Confirm appointment for Tuesday",
"instructions": "Be friendly and brief",
"event_webhook": "https://your-app.com/webhook",
"max_duration_seconds": 300
}
// Response
{
"success": true,
"call_id": "550e8400-e29b-41d4-a716...",
"status": "initiating"
}Answer and route incoming calls automatically
Schedule appointments via natural conversation
Outbound AI calls for feedback collection
Proactive status calls to customers
Self-host on your infrastructure. Pay annually for the license, not per minute or per call.
For personal & non-commercial use
For freelancers & solo projects
For teams & growing companies
For large-scale deployments
Get started in seconds. One command to install, set your credentials, and you're ready to go.
# Install SIP4AI
curl -sSL https://sip4ai.com/install.sh | bash
# Set credentials
export SIP_USERNAME=your_user
export SIP_PASSWORD=your_pass
export OPENAI_API_KEY=sk-...
# Run
sip4ai