Deepgram

What is Deepgram?

Deepgram is a voice AI platform for developers that offers speech-to-text, text-to-speech, audio intelligence and a Voice Agent API through one set of APIs. Its Nova-3 models transcribe recorded and live audio in dozens of languages with speaker diarization, smart formatting, keyterm prompting and automatic language detection. Flux models are built for voice agents and add end-of-turn detection and interruption handling. The Voice Agent API joins speech recognition, LLM orchestration and speech synthesis in one WebSocket connection, with the option to bring your own LLM or TTS provider. Deepgram runs as a managed service with US, EU and Australia endpoints, or in-VPC and self-hosted for compliance needs. Contact centers, voice app builders and enterprises are the main users. Pay As You Go starts with $200 in free credit and no card, Growth uses prepaid annual credits from $4K per year, and Voice Agent pricing is $4.50 per hour.

Use cases

• Transcribe recorded calls and meetings
• Caption live audio in real time
• Build a voice agent that listens and speaks
• Generate streaming speech for conversations
• Analyze conversations for sentiment and topics
• Self-host speech models for data control

Who it's for

• Developers
• Contact center teams
• Voice app builders
• Enterprises

Why use Deepgram

• One Voice Agent API replaces stitching together separate speech, language and voice services.
• Built-in turn-taking and barge-in handling make conversations feel less robotic.
• Pay As You Go starts with $200 in free credit, no credit card, no minimums and no expiration.
• Deployment choices from managed cloud to self-hosted help with data residency and compliance needs.
• Batch and real-time modes cover both recorded audio and live conversations in one platform.

Deepgram features

• Nova-3 speech-to-text for batch and streaming audio
• Flux models with built-in turn detection and barge-in handling
• Speaker diarization, smart formatting and keyterm prompting
• Automatic language detection across dozens of languages
• Flux TTS with 36 English voices across seven accents
• Voice Agent API combining STT, LLM orchestration and TTS
• Bring-your-own LLM and TTS options
• Audio intelligence: summarization, sentiment, topics and intent
• Custom and industry-tuned models such as Nova-3 Medical
• Managed, dedicated, in-VPC or self-hosted deployment

Deepgram integrations

• Five9
• Genesys
• Twilio
• Vonage
• Amazon Connect
• AudioCodes

How to get started with Deepgram

1. Sign up at deepgram.com for the free credit and an API key, with no credit card required.
2. Choose an endpoint: pre-recorded or live speech-to-text, text-to-speech, or the Voice Agent API.
3. Pick a model, such as Nova-3 for transcription or Flux when the app has to manage conversational turns.
4. Send audio or open a WebSocket from your application using an SDK or the REST API.
5. Add telephony through a partner, then scale to Growth credits or self-hosting as volume rises.

Deepgram limitations

• Developer-oriented, so you integrate through APIs and SDKs
• Does not provide the reasoning LLM; you attach a managed or your own provider
• Flux models run only on the /v2/listen endpoint
• Concurrency limits vary by plan, and support on the self-serve plans is community and Discord

Deepgram pricing

Yes, Deepgram has a free plan.

Deepgram plans

• Pay As You Go

  Free $200 Credit

  then pay-as-you-go

  • No minimums, no expiration
  • No credit card required
  • All endpoints in public models
  • Up to 50 concurrent REST STT requests

• Growth

  $4K+ /year

  prepaid annual credits, save up to 20%

  • Credits redeemed against usage
  • Up to 225 concurrent WSS STT streams
  • Up to 60 for TTS and Voice Agent

• Enterprise

  Custom

  contact sales

  • Large volumes and custom models
  • Dedicated or self-hosted deployment
  • SLAs

Plans as listed on the official pricing page on 24 Oct 2026. Prices can change and may vary by country or exclude tax. Check the official pricing page.

Deepgram alternatives

The main Deepgram alternatives listed on Debuly are ElevenLabs, Cartesia, Rime, Hume AI, Talkscriber and Qwen3-TTS. The table compares pricing and platforms; order is A–Z.

1. 

   Cartesia

   Voice AI platform with Sonic text-to-speech, Ink speech-to-text and managed voice agents, priced by credits and minutes.

   Audio and voiceDeveloper platformsAutomation and agents
   FreemiumVisit Cartesia (official site)
2. 

   ElevenLabs

   AI audio platform for text to speech, voice cloning, dubbing, transcription, music and conversational voice agents, via app or API.

   Audio and voiceAutomation and agentsVideoDeveloper platforms
   FreemiumVisit ElevenLabs (official site)
3. 

   Hume AI

   Expressive text-to-speech and empathic voice agent APIs that let developers direct emotion, tone and pacing in generated speech.

   Audio and voiceDeveloper platformsAutomation and agents
   FreemiumVisit Hume AI (official site)
4. 

   Qwen3-TTS

   Open-source text-to-speech model family from Qwen with voice cloning, voice design from a text description and low-latency streaming.

   Audio and voiceDeveloper platforms
   Open sourceVisit Qwen3-TTS (official site)
5. 

   Rime

   Text-to-speech and speech-to-text API for voice agents, with low-latency Mist and Coda voices, 10 TTS languages and self-hosting.

   Audio and voiceDeveloper platforms
   Free trialVisit Rime (official site)
6. 

   Talkscriber

   Voice AI platform with speech-to-text and text-to-speech APIs, plus Omnix, a live co-pilot for insurance and financial sales calls.

   Audio and voiceAutomation and agentsProductivity
   FreemiumVisit Talkscriber (official site)

Deepgram FAQ

Sources & verification

Facts on this page come from the official sources below, each with the date we read it. Written by Debuly; the vendor did not review this page.

Image credits

• Logo: from deepgram.com · fetched 23 Oct 2026 · Vendor trademark, used only to identify the product; removed on request
• Cover: none collected yet, so a generated tile is shown.

Spot an error? Report an issue with this listing. We also remove a logo or image on the owner's request.