Deepgram is a voice AI company whose developer APIs turn speech into text, turn text into speech, and run full voice agents.
In plain words
Developers send audio to Deepgram's API and get back a transcript. It can work on recorded files or on live audio as it streams in. A separate API does the reverse, reading text aloud as generated audio. Its Voice Agent API ties the pieces together: it listens, passes the words to a language model, and speaks the reply over one connection.
A quick example
Imagine a support line that answers calls with a voice agent. The caller's speech is transcribed live, a language model decides what to say, and a synthetic voice reads the answer back. The agent can also call outside tools mid-conversation, such as looking up an order.
Where you'll meet it
- Transcribing recorded calls, meetings and media files.
- Live captions and real-time transcription of streamed audio.
- Phone and web voice agents.
Related
Sources
- DocsPre-Recorded Audio: Getting Started, Deepgram Docs · read 29 Sept 2026
- DocsStreaming Audio Transcription: Getting Started, Deepgram Docs · read 29 Sept 2026
- DocsText-to-Speech: Getting Started, Deepgram Docs · read 29 Sept 2026
- DocsVoice Agent: Getting Started, Deepgram Docs · read 29 Sept 2026
Checked by WhatIsAI against the sources above · 29 Sept 2026. Spotted an error? Report it.