Deepgram Text-to-Speech

Deepgram serves two text-to-speech families on separate endpoints, and the voices do not overlap. Aura voices run only on /v1/speak. Flux TTS voices run only on /v2/speak. /v2/speak is an additional endpoint. /v1/speak is unchanged and remains supported.

Pick the family first

Need	Family	Endpoint
A voice agent that streams LLM output and must survive barge-in	Flux TTS	`wss://api.deepgram.com/v2/speak`
One-shot English audio: a file, an IVR prompt, a notification	Flux TTS batch	`POST https://api.deepgram.com/v2/speak`
Spanish, German, French, Dutch, Italian, or Japanese, one-shot	Aura-2	`POST https://api.deepgram.com/v1/speak`
Spanish, German, French, Dutch, Italian, or Japanese, streamed with manual flush control	Aura-2	`wss://api.deepgram.com/v1/speak`
An existing `aura-{voice}-en` voice you must keep	Aura-1	`/v1/speak`