Open a voice session

GET

/api/v1/voice/session

const url = 'http://localhost:11044/api/v1/voice/session';
const options = {method: 'GET', headers: {'X-Embabel-Api-Key': '<X-Embabel-Api-Key>'}};

try {
  const response = await fetch(url, options);
  const data = await response.json();
  console.log(data);
} catch (error) {
  console.error(error);
}

curl --request GET \
  --url http://localhost:11044/api/v1/voice/session \
  --header 'X-Embabel-Api-Key: <X-Embabel-Api-Key>'

Returns a short-lived Deepgram token and the settings it is for. The durable key never leaves this appliance. Ask again for each socket.

Authorizations

apiKeyAuth
basicAuth

Responses

200

The credential and the settings it is for

*/*

The credential a browser opens its Deepgram audio socket with and the settings that go with it, answered by GET /voice/session

object

effects

required

Audio effect presets the user chose for spoken replies, such as warm, radio or echo. Empty for none

Array<string>

replacements

required

Transcription corrections, each written heard:intended, to pass to Deepgram so it hears this world’s vocabulary

Array<string>

speed

required

Speaking rate as a multiplier of Deepgram’s natural pace, where 1.0 is natural

number format: double

token

required

The Deepgram credential for this session: the user’s own key if their world stores one, else this installation’s or deployment’s. Ask again for each socket rather than caching it

string

transcriber

required

Deepgram speech-recognition engine to transcribe with, such as nova-3

string

voice

required

Deepgram voice that speaks replies, such as aura-2-phoebe-en: the user’s chosen voice, else the deployment’s default

string

409

This world has no Deepgram key — voice is not set up