Introduction

Deepgram supports a variety of deployment methods, including a self-hosted offering, which is an isolated service deployed to customer-requisitioned cloud instances or on-premises data centers. This guide will walk you through the process of self-hosting Deepgram in your own environment.

If you're an IT administrator deploying a self-hosted instance of Deepgram, read on for high-level guidance on setup, suggested configuration, periodic maintenance, and frequently asked questions. This series of guides aims to:

Describe all requirements and needed assets for the installation
Tell you how to configure your environment and set up your server for the installation
Show you how to install the actual Deepgram application
Provide a starting point for how to connect your own applications with your self-hosted Deepgram instance
Help you plan your server maintenance and security practices

Why Self-Host

Using Deepgram as a service has a variety of benefits. First, it's extremely fast to start developing with. Signing up, getting an API key, and getting your first voice AI response can take as little as a minute. Using Deepgram as a service also enables you to avoid all hardware, installation, configuration, backup, and maintenance-related costs.

With that being said, there are situations a self-hosted deployment might make sense. The most common use cases we have seen are when you have stringent performance or security requirements.

Performance

Certain use cases, like AI voicebots, have very sensitive latency and load requirements. If you need ultra-low latency with voice AI services colocated with your other services, self-hosting can meet these requirements.

Security

One of the common use-cases for self-hosting Deepgram is to satisfy security or data privacy requirements. In a typical self-hosted deployment, no audio, transcripts, or other identifying markers of the request content are sent to Deepgram. Your Deepgram components will only contact the Deepgram license server in order to validate the Deepgram components and models, as well as report usage information. The usage information reported for each request includes metadata such as audio duration or character count, features requested, and success response codes.

Furthermore, self-hosted deployments do not offer a method to persist request or response data on your own servers. Request and response data are not stored beyond the duration of the original API request.

Components

Before you deploy Deepgram, you’ll need to make effective design decisions about the components of your system, their relationships, and the interactions between components. Ideally, your architecture will meet your business needs, optimize both performance and security, and provide a strong technical foundation for future growth.

Deepgram provides a variety of components available for a self-hosted deployment. Many guides describe how to create a deployment using Deepgram’s required components, API and Engine. Some guides include details on additional components, such as the License Proxy in the diagram below. See Self-Hosted Add Ons for more details.

ℹ️
If you aren't certain which components your contract includes, please consult your Deepgram Account Representative.

Deepgram API

The Deepgram API interfaces with the Deepgram Engine to expose endpoints for your requests.

Deepgram Engine

The Deepgram Engine performs the computationally intensive task of voice AI. It also manages GPU devices and responds to requests from the API layer. Because the Deepgram Engine is decoupled from the Deepgram API, you can scale it independently from the API.

Licensing and Usage

Deepgram components register with the Deepgram License Server, represented above within the Deepgram Cloud, in order to verify licensing and report usage. API and Engine containers can be configured to connect directly to the licensing server, or to proxy their communication through the Deepgram License Proxy.

Prerequisites

To take advantage of our self-hosted product offering, you will need to enroll in a Deepgram Enterprise Plan. If you're interested, please contact us! You'll be promptly connected with a Deepgram Account Representative who will help you every step of the way, from proof-of-concept to a full production environment.

Your Deepgram Account Representative will guide you through the process of setting up:

a Deepgram product contract
a Deepgram Console account. Your contract will be tied to a specific Deepgram project, and you can manage your usage, credentials, billing, and more in the Deepgram Console.

Ahead of your planned self-hosted deployment, your Deepgram Account Representative will need:

your Deepgram Console account email address
your Deepgram Console Project ID

Providing this information will allow Deepgram to authorize your project for self-hosted usage, including access to container images and download links for AI models.

Detailed troubleshooting and on-demand support require an on-going support contract with Deepgram.

Initial Setup

To setup your deployment environment and successfully self-host Deepgram products for the first time, the guides in this "Self-Hosted Deployments" section will take you through step by step.

Starting with this guide, you can use the "What's Next" section at the bottom of each page to take you from one guide to another until you have a working environment. Unless specifically indicated, each step in a given guide is required - skipping steps will leave you with an incomplete or broken environment.

As an example workflow, if you have a bare-bones deployment using Docker on AWS for a Proof-of-Concept, you would complete the following guides in order:

At the end of these guides, you should be able to make an inference request, and our proof-of-concept is complete!