Deepgram

Active

Country: United States (US)
Founded: 2015
Developer: Deepgram Inc.
Pricing: Freemium
Open source: No
Platforms: Web, Cloud, Self-hosted

Indexes

APIs & MCP Servers

Overview

Deepgram is an enterprise voice AI platform providing Speech-to-Text (STT), Text-to-Speech (TTS), and Voice Agent APIs. Founded in 2015, it enables developers to build voice-enabled applications with real-time and batch processing capabilities. The platform is designed for developers and enterprises seeking accurate, scalable, and cost-effective voice AI solutions that can be deployed in cloud or self-hosted environments.

Key Features

Speech-to-Text (STT) - Real-time and batch automatic speech recognition with support for multiple models including Nova-2 and Whisper.
Text-to-Speech (TTS) - Convert text into natural-sounding speech for voice applications and agents.
Voice Agent APIs - Build autonomous voice agents capable of handling conversations and complex interactions.
Multi-model Support - Access to various ASR models optimized for different domains and use cases, including air traffic control and general applications.
Real-time Processing - Stream audio data for immediate transcription and analysis with low latency.
Pre-recorded Audio Support - Batch processing of audio files with comprehensive transcription and metadata enrichment.
Natural Language Understanding - Transcripts enriched with NLU metadata for deeper speech analysis and insights.
Starter Apps - Pre-built code examples and templates for rapid integration and application development.

Pricing

Plan	Price	Includes
Pay-as-you-go	Usage-based	$200 free credit for new users, no credit card required, 5 concurrent requests limit with Whisper model
Paid Plan	Custom pricing	Higher concurrency limits (15 concurrent requests with Whisper), enterprise features, dedicated support
Free Tier	Free	Limited API calls, access to starter apps, community support

Platforms & Requirements

Deepgram operates as a cloud-based API platform accessible via web and REST APIs. It supports real-time streaming and batch processing across any platform capable of making HTTP requests. Self-hosted deployment options are available for enterprise customers. No specific OS requirements exist as it is accessed programmatically; integration is language-agnostic through API calls.

Integrations & Ecosystem

Make.com
REST API
WebSocket for real-time streaming
OpenAI Whisper model
Custom webhook integrations
Voice agent frameworks
Audio file upload (WAV, MP3, and other formats)
Third-party voice application platforms

Alternatives

App	Difference
Google Cloud Speech-to-Text	Google's offering provides strong integration with Google Cloud ecosystem but typically higher latency and different pricing model.
Amazon Transcribe	AWS service with deep AWS integration but less specialized for real-time voice agents and lower accuracy in specialized domains.
AssemblyAI	Focused primarily on speech recognition with strong accuracy but offers fewer TTS and voice agent capabilities compared to Deepgram.
Twilio	Broader communications platform with voice capabilities but less specialized in AI-driven speech recognition and voice agents.

Reputation

Deepgram is recognized as a leader in enterprise voice AI with particular strength in speech recognition accuracy across diverse domains. The platform is praised for its cost-effectiveness, real-time capabilities, and developer-friendly API design. Users appreciate the free credits for new developers and the quality of the Nova-2 model. Some limitations include rate limits on certain models like Whisper and the need for custom pricing for enterprise features, which may present barriers for smaller projects.

Sources (8)