Deepgram provides AI-powered speech recognition and understanding services, offering APIs for real-time and pre-recorded audio transcription, text-to-speech, and audio intelligence.
Deepgram provides AI-powered speech recognition and understanding services, offering APIs for real-time and pre-recorded audio transcription, text-to-speech, and audio intelligence. On Nagent, Deepgram is exposed as a fully-configurable artificial intelligence integration that any agent can call — 8 actions, and API key authentication. No code is required to wire Deepgram into your workflow — connect it once via the External Integrations panel and reuse it across every agent you build.
Agent builders use Deepgram to automate the kinds of tasks artificial intelligence teams previously handled manually. Concrete examples — each one is a single agent step in Nagent — include:
Every action and trigger is paired with a structured input/output schema (visible in the sections below), so when you wire Deepgram into Helix — our agentic agent builder — the editor knows exactly what each step expects and produces. Configure once, deploy anywhere across your Nagent agents.
Every operation an agent can call against Deepgram, with input parameters and output schema. Drop these into any step of an agent built in Helix.
DEEPGRAM_GET_MODELRetrieve metadata for a specific Deepgram model by its UUID. Returns detailed model information including name, architecture, supported languages, version, and capabilities. Works for both STT (speech-to-text) and TTS (text-to-speech) models.
Input parameters
The specific UUID of the model to retrieve
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
DEEPGRAM_GET_MODELSRetrieve metadata on all public Deepgram models (speech-to-text and text-to-speech). Returns comprehensive model information including supported languages, architectures, versions, and capabilities. Set include_outdated to True to include deprecated versions.
Input parameters
If true, include all versions of every model; otherwise only the latest.
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
DEEPGRAM_GET_PROJECTSTool to list all Deepgram projects. Use after authenticating with your API key.
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
DEEPGRAM_GET_PROJECT_USAGE_SUMMARYRetrieves aggregated usage statistics for a Deepgram project including total audio duration, billable duration, number of requests, channels processed, and confidence/relevance scores. Returns both overall totals and breakdowns by model/accessor/tag. Use this to analyze API consumption, track costs, or monitor transcription quality metrics over time.
Input parameters
End date for the usage summary range. Accepts YYYY-MM-DD or ISO 8601 format (e.g., '2024-01-31' or '2024-01-31T23:59:59Z')
Filter usage by request tag. Returns only usage from requests tagged with this value
Filter usage by Deepgram model name (e.g., 'nova-2', 'general', 'whisper'). Returns only usage from requests that used this model
Start date for the usage summary range. Accepts YYYY-MM-DD or ISO 8601 format (e.g., '2024-01-01' or '2024-01-01T00:00:00Z')
Unique identifier of the Deepgram project
Filter usage by API key or user accessor ID. Returns only usage attributed to this specific accessor
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
DEEPGRAM_LIST_PROJECT_SCOPESTool to list all scopes for a specified Deepgram project. Use when you need to retrieve all permission scopes for a project.
Input parameters
Unique identifier of the Deepgram project to list scopes for.
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
DEEPGRAM_LIST_THINK_MODELSTool to list available think models for AI agent processing and voice agent configuration. Use when you need to see which think models are available for voice agents.
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
DEEPGRAM_SPEECH_TO_TEXT_PRE_RECORDEDTool to transcribe pre-recorded audio files into text. Use when converting a publicly accessible audio file URL to text. Primary transcript is at `results.channels\[0\].alternatives\[0\].transcript` in the response. Silent audio returns a valid empty transcript, not an error. Verify supported models and language codes via `DEEPGRAM_GET_MODELS` when uncertain.
Input parameters
Enhanced tier for higher-accuracy recognition.
Deepgram model to use (e.g., 'general', 'phonecall').
Enable speaker diarization if true.
List of keyterms to boost in the transcript. Supported by Nova-3 and Flux models only. Use simple terms without intensifiers (e.g., 'invoice', 'payment terms'). For other models, use 'keywords' instead.
Specific model version to use.
List of keywords to boost in the transcript. Supported by Nova-2, Nova-1, Enhanced, and Base models. Format: 'keyword:intensifier' (e.g., 'invoice:5'). For Nova-3 or Flux models, use 'keyterm' instead.
Language code for transcription (e.g., 'en-US').
Public URL of the audio file to transcribe. Must be a direct, publicly downloadable media file URL — not a preview page, auth-gated link, or expiring/short-lived URL.
Automatically punctuate transcript.
Number of alternative transcripts to return.
MIME type of the audio file.
Apply smart formatting (capitalization/punctuation).
Automatically detect language if true. Prefer explicit `language` over this when the audio language is known; auto-detection reduces accuracy for non-English content.
Return interim results if true.
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
DEEPGRAM_TEXT_TO_SPEECH_RESTTool to convert text into natural-sounding speech. Use when you need TTS audio from text inputs.
Input parameters
The text to be synthesized into speech.
Deepgram TTS model to use (e.g., 'aura-asteria-en').
Pitch adjustment multiplier (0.5–2.0).
Speech rate multiplier (0.25–4.0).
Desired voice within the selected model.
Model version identifier.
Audio encoding type (e.g., 'linear16', 'mp3').
Language code for synthesis (e.g., 'en-US').
Container format for audio (e.g., 'wav', 'mp3').
Desired sample rate for output audio in Hz (e.g., 16000).
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
No publicly available marketplace agent is found using this tool yet. There are 45 agents privately built on Nagent that already use Deepgram.
Build on Nagent
Connect Deepgram to any Nagent agent in minutes — no API key management, no boilerplate. Just configure and deploy.
The five questions agent builders ask before adopting a new integration.
Open the External Integrations panel inside Nagent (app.nagent.ai/externalIntegration), find Deepgram, and click "Connect Now." You'll authenticate with an API key — Nagent handles credential storage and refresh automatically. Once connected, Deepgram is available to any agent in your workspace.
No. Nagent provides no-code integration for every tool. Once Deepgram is connected, you configure its 8 actions directly in the agent builder UI — no API calls, no boilerplate, no schema management.
Helix — Nagent's agentic agent builder — lets you drop Deepgram steps into any workflow visually. Pick an action (e.g., one of those listed above), fill in the inputs (Helix knows the required vs. optional schema for each parameter), and connect it to upstream/downstream steps. Triggers run as the entry point of an agent, so when a Deepgram event fires, the agent kicks off automatically.
Every Deepgram action and trigger ships with a fully-typed schema — input parameters with name, type, required flag, and description, plus the output payload shape. The schemas are documented in the sections above. Helix uses these schemas to validate your configuration at build time and to type-check the data flowing between steps.
Yes. While Deepgram ships with 8 pre-built artificial intelligence actions, you can layer custom logic around them inside Helix — pre/post-processing steps, conditional branches, retries, or stitching Deepgram together with other connected tools. For deeper customization, talk to our team about Nagent's Agentic AI Lab — forward-deployed engineers who build Deepgram-based workflows tailored to your business.