Skip to content
Microsoft Teams Bridge for Deepgram Voice Agents (Python)

Microsoft Teams Bridge for Deepgram Voice Agents (Python)

Welcome! deepgram-msteams-bridge (PyPI: deepgram-msteams-bridge) puts a Deepgram Voice Agent on a real Microsoft Teams call.

It is a small Python service (and importable asyncio library, built on aiohttp) that sits between two WebSockets:

flowchart LR
    Teams["Microsoft Teams<br/>call"]
    StandIn["StandIn media bridge<br/>(hosted service)"]
    Bridge["this bridge<br/>(yours)"]
    DG["Deepgram Voice Agent<br/>Nova STT + LLM + Aura TTS<br/>+ turn-taking"]
    Teams <--> StandIn
    StandIn -- "HMAC WS, PCM 16k" --> Bridge
    Bridge -- "WS, linear16 @ 16k" --> DG

The hosted StandIn media bridge (standin.komaa.com) joins the Teams call and dials into your bridge, one authenticated WebSocket per call. The bridge opens one Deepgram Voice Agent session per call and relays between them. The StandIn wire is base64 PCM 16 kHz mono and the Voice Agent session is pinned to linear16 at 16 kHz both ways, so the hot path is copy-only: no resampling, no re-encoding, no transcoding.

What it does

Section titled “What it does”
  • Realtime voice - the caller talks to your Deepgram agent and hears it reply. Turn-taking, VAD and interruption are the Voice Agent’s own; the bridge maps the caller’s barge-in onto the Teams side and ghost-drops stale audio so nothing plays after an interruption.
  • No dashboard to configure - the bridge configures each session itself from environment variables: STT model (nova-3), LLM (Deepgram-managed open_ai/anthropic, or any BYO provider via a think endpoint), voice (aura-2-thalia-en), prompt, greeting.
  • Per-call personalization - caller name, tenant and direction are injected into the agent prompt; a deterministic greeting doubles as a spoken AI disclosure.
  • Built-in agent tools - end_call, express (avatar emotion), show_image (image on the bot’s tile, SSRF-guarded), look (vision). Declared automatically as client-side functions in every session.
  • Extensible tools - register your own function tools the bridge executes in-process. See Extending the Agent’s Tools.
  • Live context without interruptions - participant counts, DTMF digits and active-speaker changes ride UpdatePrompt as a bounded rolling context section.
  • Two call governors - StandIn-side tier cutoffs get a spoken goodbye; a bridge-side MAX_CALL_MINUTES hard cap protects your Deepgram budget. Goodbyes are the exact text on both paths, and both are backstopped.
  • Hardened transport - replay-proof HMAC handshake, connection and payload caps, duplicate-call rejection, dead-peer detection, SettingsApplied-gated audio, graceful drain, and a *.deepgram.com host allowlist so your API key can never be sent elsewhere.

Use the sidebar to navigate. Start with Getting Started, or jump to the Configuration Reference or Wire Protocol. There is also a runnable example project in the repository showing a custom vision hook and a custom lookup_order tool - walked through in Run the Example.