Deepgram: STT at ~$0.26/hour with word timestamps, diarization and sub-300ms streaming — the transcription engine for live audio products.
Deepgram
Fast, cheap speech-to-text built for realtime streaming
deepgram.com · $200 free credit; ~$0.26/hr
Capabilities
Capabilities
- Streaming + batch STT, diarization, word timings; TTS and voice-agent APIs too
- $200 free credit; ~$0.0043/min batch
Best for
- Realtime transcription clones (meeting notes, captions, voice UIs)
Not for
- Highest-accuracy offline transcripts of messy audio (compare Whisper/AssemblyAI)