TOOLS
Deepgram
Real-time speech-to-text API for voice products
VoiceFreemiumLLM-drafted
Part of this entry was LLM-drafted and is being polished.
Deepgram is the developer-popular real-time ASR API — Nova-3 model is class-leading on accuracy and latency. Used as the STT backbone for many voice agent products. Free credits to start, pay-as-you-go.
Editor's verdict
Default real-time STT for voice agent builders — Nova-3 is genuinely strong on streaming use cases. Pricing is fair. Caveat: for purely batch transcription of recorded files (podcasts, meetings), AssemblyAI offers more 'understanding' features (summary, topics, sentiment) at similar cost.
Use cases
- real-time speech-to-text
- voice agent stt
- transcription api
Reviews
No reviews yet. Be the first.
Last updated: 2026-04-29