Deepgram

What is Deepgram?

Deepgram is a voice AI platform that provides APIs for speech-to-text, text-to-speech, audio intelligence, and conversational voice agents. Developers and enterprises can use the services in real time or batch mode and deploy through Deepgram's cloud or self-hosted environments.

Its unified Voice Agent API combines transcription, language-model orchestration, speech synthesis, business logic, and external-system connections in one interface. Teams can also use individual APIs, embed Deepgram into another platform, or work with Deepgram on custom models for specialized workflows and compliance needs.

How to Use Deepgram

1. Create a Deepgram account and choose speech-to-text, text-to-speech, audio intelligence, or the unified Voice Agent API.
2. Test the selected model and configuration in the playground.
3. Integrate the API into an application and connect the required language-model, business-logic, and external-system components.
4. Run the workload in real time or batch mode, using cloud or self-hosted deployment as required.
5. Monitor results and tune models or voice-agent behavior for the target use case.

Core Features

• Speech-to-text API: Transcribes streaming or recorded audio.
• Text-to-speech API: Generates spoken audio for applications and agents.
• Unified Voice Agent API: Combines STT, TTS, model orchestration, and business logic.
• Audio intelligence: Extracts additional information from audio content.
• Multilingual conversation support: Flux detects spoken language and supports ten languages listed on the homepage.
• Flexible deployment: Supports real-time, batch, cloud, and self-hosted operation.

Use Cases

• Contact-center automation: Powers real-time voice agents for customer conversations.
• Speech analytics: Transcribes and analyzes calls or other audio at scale.
• Conversational applications: Adds speech input and generated voice responses to products.
• Medical and domain transcription: Supports specialized transcription workflows.
• Platform voice infrastructure: Lets partners embed enterprise voice capabilities in their own products.

Pricing

Deepgram offers free account signup. The homepage does not list included credits or usage prices for its APIs.

Frequently Asked Questions

Can Deepgram be self-hosted?

Yes. The homepage lists both cloud and self-hosted deployment.

Does the Voice Agent API require separate STT and TTS integrations?

No. Deepgram presents it as one API that unifies STT, TTS, model orchestration, and business logic.

Can developers test before integrating?

Yes. Deepgram provides an API playground for testing.