{
  "slug": "realtime-voice-agent",
  "title": "A realtime voice agent",
  "promise": "The four parts of an agent you talk to: a realtime framework, speech recognition, a fast model and a voice.",
  "week": null,
  "url": "https://agentswitchboard.dev/stacks/realtime-voice-agent",
  "items": [
    {
      "name": "LiveKit Agents",
      "slug": "livekit-agents",
      "url": "https://agentswitchboard.dev/agents/livekit-agents",
      "iconUrl": "https://agentswitchboard.dev/icons/livekit-agents.png?v=14116536",
      "description": "Builds production realtime voice and video AI agents with WebRTC transport, turn detection, and STT/LLM/TTS pipelines. Python and Node framework by LiveKit.",
      "provider": "LiveKit",
      "categories": [
        "voice-messaging"
      ],
      "accessMethods": [
        "api",
        "cli"
      ],
      "verified": true,
      "official": true,
      "featured": false,
      "authType": "apiKey",
      "why": "Runs the conversation: streams audio over WebRTC, detects when the speaker is done, and chains speech-to-text, the model and text-to-speech."
    },
    {
      "name": "Deepgram",
      "slug": "deepgram",
      "url": "https://agentswitchboard.dev/agents/deepgram",
      "iconUrl": "https://agentswitchboard.dev/icons/deepgram.png?v=34d0f9eb",
      "description": "Speech-to-text and voice AI API with real-time streaming transcription, speaker diarization, and text-to-speech for voice agents.",
      "provider": "Deepgram",
      "categories": [
        "voice-messaging"
      ],
      "accessMethods": [
        "api",
        "cli"
      ],
      "verified": true,
      "official": true,
      "featured": false,
      "authType": "apiKey",
      "why": "Hears the user, with real-time streaming transcription and speaker diarization."
    },
    {
      "name": "Cerebras Inference",
      "slug": "cerebras-inference",
      "url": "https://agentswitchboard.dev/agents/cerebras-inference",
      "iconUrl": "https://agentswitchboard.dev/icons/cerebras-inference.png?v=874fceca",
      "description": "Serves open-weight LLMs such as Qwen and GPT OSS through an OpenAI-compatible API on wafer-scale hardware, for developers who need very fast inference.",
      "provider": "Cerebras",
      "categories": [
        "language",
        "infrastructure"
      ],
      "accessMethods": [
        "api"
      ],
      "verified": true,
      "official": true,
      "featured": false,
      "authType": "apiKey",
      "why": "Thinks fast: serves open-weight models such as Qwen and GPT OSS through an OpenAI-compatible API, built for very fast inference."
    },
    {
      "name": "Cartesia",
      "slug": "cartesia",
      "url": "https://agentswitchboard.dev/agents/cartesia",
      "iconUrl": "https://agentswitchboard.dev/icons/cartesia.png?v=d81b9acb",
      "description": "Delivers ultra-low-latency text-to-speech and speech-to-text through the Sonic model family, with a voice-agent framework and streaming WebSocket API.",
      "provider": "Cartesia",
      "categories": [
        "voice-messaging"
      ],
      "accessMethods": [
        "api"
      ],
      "verified": true,
      "official": false,
      "featured": false,
      "authType": "apiKey",
      "why": "Answers out loud with low-latency text-to-speech, in a custom voice cloned from a short sample."
    }
  ]
}