t140llmReal-time text for LLM streams

Convert LLM streaming responses into T.140 real-time text for SIP, WebRTC, and (S)RTP.

Any LLM SDK

Pass OpenAI, Anthropic, Gemini, Mistral, Cohere, Ollama, or Vercel AI SDK streams straight in — EventEmitters and async iterables both work, no wrapping.

Real transports

WebSocket, RTP, SRTP, and Unix sockets, with FEC (RFC 5109), redundancy, and multiplexing of several LLM streams into one output.

Low latency

An event-driven send model delivers characters with sub-millisecond chunk-to-wire latency — built for lossy and time-sensitive links.