Voxta docs

Deepgram (TTS)

Deepgram's Flux voices as Voxta's text-to-speech: low-latency, streamed English speech with speed and expressivity dials. The same module also does STT.

Deepgram provides both STT and TTS from one module and one API key. This page covers TTS. The STT page covers transcription. If you already installed Deepgram for one, the same module handles the other.

Voxta speaks with Deepgram's Flux voices (English, American and British accents) over a streaming socket, so audio starts while the line is still being generated. The older Aura voices are not used.

Setup

Get an API key

Sign up at console.deepgram.com and create a key under API Keys. New accounts get free credits.

Add to Voxta

Add Modules → Services → Deepgram → Install, then follow the install steps. Paste the API Key, then pick a Flux voice.

Voice settings

SettingNotes
Speed0.85–1.15, in steps of 0.05. The voice's natural prosody is kept.
Expressivity-2 (calm) to 2 (animated), default 0. Changes pitch range and pacing, not which emotion is expressed. In beta at Deepgram.

Emotion tags

Deepgram doesn't read anything in the text as a direction. With emotion tags on, it offers dials instead of feelings: slow, fast, excited and calm, which adjust speed and expressivity.

On this page