man am i misunderstanding the docs?

new inference.STT({
        model: 'assemblyai/universal-streaming',
        language: 'en',
}),

How can I use this with my own livekit server? Just getting

APIConnectionError

when i pass in apiKey or apiSecret

inference.STT is using the inference gateway is LiveKit Cloud feature. In order for you to run your STT, you must use the plugin

import * as deepgram from '@livekit/agents-plugin-deepgram';  
  
// Create STT instance  
const stt = new deepgram.STT();