Deepgram
Free tierThe most accurate and cost-effective real-time APIs for speech-to-text, text-to-speech, and voice agents
Free tier available·All audiences·Powered by Deepgram·API available
Key strengths
Real-time speech-to-text with ultra-low latencyConversation-aware TTS with turn-taking and interruption handlingUnified Voice Agent API (STT + TTS + LLM orchestration in one call)Self-hosted deployment option45+ language supportAudio Intelligence and analytics
Free tier + paid plans
San Francisco, USA
Founded 2015
Self-hostable
Facts verified against deepgram.com on
Recent changes
- Plans updated: Free tier includes $200 credit then pay-as-you-go. Growth plan starts at $4K+/year with pre-paid credits (save up to 20%). Enterprise pricing available via sales. STT streaming from $0.0048/min (Nova-3), TTS and Voice Agent API usage-based.
Get alerts when Deepgram changes
Free. We'll email you when pricing, the free tier, API access or open-source status changes.
No ratings yet
Use Cases — Technical
Deepgram is well-suited for engineering teams building:
- Real-time transcription pipelines: Stream audio over WebSocket and receive low-latency transcripts; ideal for live captioning, agent-assist, and voice search
- Conversational AI / Voice Agents: Use the unified Voice Agent API to orchestrate STT + LLM + TTS in a single call, reducing round-trip latency and integration overhead
- Contact center automation: Native Amazon Connect integration enables voice bot and analytics deployments within existing CCaaS stacks
- Speech analytics: Audio Intelligence endpoints support diarization, summarization, sentiment analysis, and topic detection at scale
- On-premises / air-gapped deployments: Self-hosted option supports regulated industries (finance, healthcare, government) with strict data residency requirements
- Multilingual applications: 45+ language support allows a single API integration to serve global user bases
- Interruption-aware TTS: Built-in turn-taking and interruption handling removes the need for custom VAD logic in voice agent implementations
Alternatives
See all alternatives