What is Deepgram used for?

Deepgram offers speech-to-text, text-to-speech and voice agent APIs for transcription and conversational voice applications. It is a good fit for building conversational voice agents, transcribing calls and meetings, and adding speech output to applications.

How much does Deepgram cost?

Deepgram is a paid product with no free plan. Usage-based speech AI billed per minute. Pre-recorded Nova-3 monolingual transcription is $0.0043/min on Pay As You Go and $0.0036/min on Growth; custom models are quoted by sales.

Is Deepgram open source?

No. Deepgram is proprietary (closed-source) software. Open-source alternatives to Deepgram include whisper.cpp, Whisper and LocalAI.

Can I self-host Deepgram?

Yes. Although Deepgram is closed source, it can be self-hosted on your own servers, and the vendor also offers a hosted version.

What are some alternatives to Deepgram?

Deepgram competes with AssemblyAI, Speechmatics and ElevenLabs. For open-source options, see Enlisted's ranked list of open-source Deepgram alternatives.

Deepgram: features, pricing and alternatives

About Deepgram

Deepgram is a commercial voice AI company that provides APIs for turning speech into text, text into speech, and building spoken agents. It is aimed at developers and enterprises that need transcription or voice interfaces inside their own products.

The platform covers four areas: speech-to-text for real-time and batch transcription, text-to-speech with expressive voices, a voice agent API, and audio intelligence for extracting insight from conversations. Its Flux models are described as handling turn-taking, interruptions and conversational context. The voice agent API combines speech recognition, language model orchestration and speech synthesis behind one interface, so developers do not have to stitch separate components together.

Deepgram runs as a hosted cloud service, and the homepage states that deployment can also be self-hosted. There is a free sign-up path, a playground for trying the models and a pricing page.

Key features

Real-time and batch speech-to-text
Text-to-speech with expressive voices
Unified voice agent API
Turn-taking and interruption handling
Audio intelligence for call insights
Cloud or self-hosted deployment

Good fit for

→Building conversational voice agents
→Transcribing calls and meetings
→Adding speech output to applications

Tags: speech-to-text; text-to-speech; voice-ai; transcription; voice-agents; api; ai-infrastructure

Deepgram: questions and answers

What is Deepgram used for?: Deepgram offers speech-to-text, text-to-speech and voice agent APIs for transcription and conversational voice applications. It is a good fit for building conversational voice agents, transcribing calls and meetings, and adding speech output to applications.
How much does Deepgram cost?: Deepgram is a paid product with no free plan. Usage-based speech AI billed per minute. Pre-recorded Nova-3 monolingual transcription is $0.0043/min on Pay As You Go and $0.0036/min on Growth; custom models are quoted by sales.
Is Deepgram open source?: No. Deepgram is proprietary (closed-source) software. Open-source alternatives to Deepgram include whisper.cpp, Whisper and LocalAI.
Can I self-host Deepgram?: Yes. Although Deepgram is closed source, it can be self-hosted on your own servers, and the vendor also offers a hosted version.
What are some alternatives to Deepgram?: Deepgram competes with AssemblyAI, Speechmatics and ElevenLabs. For open-source options, see Enlisted's ranked list of open-source Deepgram alternatives.