Deepgram
Connects to the Deepgram speech-to-text API. An agent can transcribe audio from a public URL, optionally with speaker diarization.
Add to an agent node's tools array with _node_type: "tool-deepgram". See Connector Reference — How connectors are used.
Authentication
| Field | Required | Description |
|---|---|---|
api_key | Yes (or deepgram_api_key) | Deepgram API key, sent as Authorization: Token <api_key>. |
deepgram_api_key | Yes (if api_key not set) | Used if api_key is not set. |
{
"name": "deepgram",
"_node_type": "tool-deepgram",
"api_key": "{{ secret.DEEPGRAM_API_KEY }}"
}
Tools
deepgram_transcribe_url
Transcribe audio from a URL using Deepgram Nova-2. Faster and cheaper than uploading.
Parameters
| Parameter | Type | Required | Description |
|---|---|---|---|
audio_url | string | Yes | Public URL to audio file (mp3, mp4, wav, m4a, etc.) |
model | string | No | nova-2 (default) | nova | enhanced | base |
language | string | No | BCP-47 language code, e.g. en-US (auto-detect when empty) |
diarize | boolean | No | Enable speaker diarization (labels speaker 0, 1, 2...) |
punctuate | boolean | No | Add punctuation (default true) |
Returns: JSON with text (full transcript), words (array of {word, start, end, confidence, speaker?}), confidence, duration, and provider ("deepgram").