Skip to main content
POST
Mint a Deepgram streaming STT session

Authorizations

x-access-token
string
header
required

JWT token passed in x-access-token header

Body

application/json

All fields optional. The endpoint only mines requestArgs for keyterm hints — anything else is ignored.

requestArgs
object

Realtime context echoed from the client. Used to derive Deepgram keyterm hints (assistant name, conversation title, explicit keyterms[]).

Response

STT session minted.

provider
enum<string>
Available options:
deepgram
access_token
string

Short-lived Deepgram bearer token. The browser passes this via the WebSocket subprotocol when opening the WSS stream.

expires_in
number

Token TTL in seconds. 0 in dev raw-key mode.

wss_url
string

Deepgram listen WSS URL the client should connect to.

Example:

"wss://api.deepgram.com/v1/listen"

model
string

Deepgram STT model in use (institution-tuned).

eot_threshold
number

End-of-turn confidence threshold.

eot_timeout_ms
number

End-of-turn timeout in ms.

eager_eot_threshold
number

Optional. Present only when speculative early-start is enabled (>0). Lower confidence at which Flux emits EagerEndOfTurn/TurnResumed so the client can start answering early.

local_vad
object

Client-side energy-VAD config for always-on fast barge-in (cut the avatar on local speech onset instead of waiting for Deepgram StartOfTurn). Always present on Anam sessions; client-only — does not affect the WSS URL.

encoding
enum<string>

Fixed PCM encoding the client AudioWorklet emits.

Available options:
linear16
sample_rate
enum<number>

Fixed sample rate (16 kHz mono int16).

Available options:
16000
keyterms
string[]

Resolved keyterm list — institution name/ainame plus the derived/explicit terms.