---
title: "Enterprise voice AI: STT, TTS & agent APIs"
description: "Power enterprise voice solutions with Deepgram’s Speech-to-Text, Text-to-Speech, and Voice Agent APIs. Real-time, accurate, and built for scale."
canonical: "https://deepgram.com/"
---

# The future is talking.

## The whole voice AI stack in one platform.

### Audio input

Your customer or end user speaking naturally

### Speech-to-text (STT)

Natural speech converted to text in real time

### LLM orchestration

Coordinates across models to deliver consistent  voice and text outputs

### Business logic

Transparent decision-making for complex,  multi-step workflows

### External systems

Connects outside systems and data into one seamless flow

### Text-to-speech (TTS)

Brings text to life across 45+ languages and  100+ voice profiles

- **200K+** — years of audio processed
- **650K+** — developers building
- **5,000+** — organizations powered
- **99.99%** — Enterprise-grade reliability

## Build in minutes

### Compare voices side by side

Build a web page that reads one script aloud in five Deepgram Flux TTS voices so I can compare them side by side. Use batch POST /v2/speak.

### Give your chatbot a voice

Add spoken replies to my existing chatbot: send each LLM message to Deepgram Flux TTS over WebSocket and play it in the browser. Voice: flux-sienna-en.

### Stream voice from your LLM

Build a Python app that streams LLM replies into Deepgram Flux TTS over the /v2/speak WebSocket and plays audio as it arrives. Voice: flux-hannah-en.

### Book appointments by voice

Build a receptionist on the Deepgram Voice Agent API that books appointments through function calling, confirms details before booking, and speaks with Flux TTS.

## Unfiltered feedback

Hear how teams are putting Deepgram to work, in their own words.

We've relied on Deepgram's speech recognition to power real-time voice interactions at scale, so the ability to deploy TTS built on the same enterprise runtime is incredibly compelling. Having STT and TTS under one roof eliminates the integration gaps and latency issues that come with piecing together separate tools.

Nikhil Gupta

Founder and CTO, Vapi

At Cloudflare, we understand what it takes to deliver real-time experiences at global scale. Flux represents the next evolution, purpose-built conversational speech recognition that eliminates the complexity and latency bottlenecks of traditional pipelines.

Dane Knecht

CTO, Cloudflare

NiCE supports some of the world's most complex contact center environments, where voice AI quality has to hold up across long, unpredictable customer interactions. Deepgram Flux TTS gives us a text-to-speech model designed for nuance, with the ability to carry context and consistency across a multi-turn conversation.

Neeraj Verma

VP, NICE

## We’re building over here

We ship in public, take feedback seriously, and answer 
questions in Discord instead of a ticket queue.

Pricing

## Pricing built for real ROI

Every tier runs on the same backend. Only the capabilities flex.

### Pay as you go

No minimums, no expiration, and no credit card required.

**Free $200 credit** then pay as you go

### Growth

Save up to 20%

Pre-paid credits for the year, redeemed against actual usage.

**$4,000+** / year

### Enterprise

For businesses with large volumes, data or deployment requirements, or support needs.

**Custom pricing**

## Articles

- The real-time voice AI stack for agents: An architecture guide (AI Engineering & Research)
- Top voice AI providers compared for 2026 (AI Engineering & Research)
- Text-to-speech comes of age: Deepgram launches conversation-native speech (Announcements)
- Anam solves turn-taking for unscripted avatar conversation with Flux STT (AI Agents)

## Speak with a voice  AI expert

Get help building fast, natural voice experiences with speech-to-text, text-to-speech, and voice agents.
