Skip to main content
Already building on a voice framework? Use SLNG’s STT and TTS models through it with a native plugin instead of calling the gateway directly. Each integration translates the platform’s native protocol into SLNG calls, so you add caching, regional routing, and provider flexibility without rewriting your agent. If you run a custom stack instead of a framework, point your requests at the drop-in gateway.

​
Region and world-part overrides

All integrations route through the SLNG platform, which supports the X-Region-Override and X-World-Part-Override headers to pin requests to a specific region (for data residency) or to a broader geographic zone like eu or na. The Cognigy and Jambonz bridges also accept ?region= and ?world-part= query parameters. Use these when the platform does not let you set custom HTTP headers. If you send both, the header wins. The LiveKit plugin wraps this in a region_override parameter on the STT and TTS classes, so you set it once when you instantiate the model. The Pipecat plugin exposes the same routing through region_override and world_part_override on its STT and TTS services. See Region and world-part overrides for the full region list and header behavior.

​
Available integrations

LiveKit

Use SLNG models inside LiveKit Agents with the livekit-plugins-slng adapter.

Pipecat

Use SLNG models inside Pipecat pipelines with the pipecat-slng STT/TTS services.

Cognigy

Connect Cognigy voice bots to SLNG STT and TTS models.

Jambonz

Route Jambonz calls through SLNG for speech processing.

​
Route an agent’s LLM

Already running an agent on a hosted platform? Point its LLM at the SLNG LLM Router, an OpenAI-compatible chat endpoint, to add response caching with a base-URL swap and without rewriting your agent.

ElevenLabs

Use the router as the Custom LLM for an ElevenLabs Conversational AI agent.

Vapi

Set the router as a custom OpenAI-compatible model in Vapi.

Retell

Route a Retell agent’s reasoning through the LLM Router.