Glossary
What is
Deepgram?
Deepgram is the AI speech recognition engine that powers StreamTranslate — delivering real-time transcription with industry-leading accuracy.
Get Started Free
About Deepgram
Deepgram is an AI audio intelligence company founded in 2015. Their Nova-2 model is one of the fastest and most accurate speech recognition engines available, with sub-300ms latency and support for 50+ languages.
Deepgram is used by companies like Spotify, Twilio, and NASA for real-time transcription at scale.
Why StreamTranslate Uses Deepgram
- Speed: Sub-300ms transcription — critical for live stream sync
- Accuracy: Industry-leading word error rates for conversational speech
- Scale: Handles millions of concurrent audio streams
- Language Support: 50+ languages with native model training
- Reliability: 99.9% uptime SLA — streams never go down mid-session
Deepgram vs. Other STT Providers
Deepgram is widely regarded as the fastest and most accurate cloud STT provider for real-time use cases. Compared to competitors like Google Speech-to-Text or AWS Transcribe, Deepgram offers lower latency and better accuracy for streaming audio.
- Deepgram Nova-2: ~0.3s latency, best-in-class accuracy
- Google STT: ~0.8s latency, strong accuracy
- AWS Transcribe: ~1-2s latency, good accuracy
- Whisper (offline): accurate but not real-time capable
Pricing
- Stream Pass — $9.99: One full stream session, all languages
- Starter — $14.99/mo: 25 hours/month, single language
- Pro — $34.99/mo: 40 hours/month, dual language
- Unlimited — $79.99/mo: Unlimited hours, dual language
See full pricing →