Usage guide
Setup, environment variables, and Python/JavaScript/Go usage examples.
InworldAITTS
Text-to-speech with Inworld AI voices, streamed over WebSocket for low latency. Good for expressive, character-style voices. Authenticates with INWORLD_API_KEY from your environment, or an explicitapi_key.
Fields
str
default:"inworld-tts-1"
Voice model to synthesize with, e.g. “inworld-tts-1” or “inworld-tts-1.5-max”. Defaults to ‘inworld-tts-1’.
str
default:"Hades"
Voice to speak with, e.g. “Hades”, “Sarah” or “Ashley”. Defaults to ‘Hades’.
float
default:"UNSET"
Sampling temperature, 0.0 to 2.0. Higher sounds more varied and expressive. Defaults to 0.8.
int
default:"UNSET"
Output audio sample rate in Hz. Defaults to 24000.
bool
default:"UNSET"
Stream over a bidirectional WebSocket. Turn off to use HTTP streaming requests instead. Defaults to True.
bool
default:"UNSET"
Let Inworld decide when to flush buffered text for lowest latency. WebSocket only; best when text arrives as full sentences. Defaults to True.
int | None
default:"UNSET"
Longest time in milliseconds Inworld waits before flushing buffered text. WebSocket only; unset means no limit. Defaults to None (not set).
int | None
default:"UNSET"
Number of buffered characters that triggers a flush, up to 1000. WebSocket only. Defaults to None (not set).
str | None
default:"UNSET"
“ON” expands text like “Dr.” or dates into spoken words, “OFF” reads it as written. Unset lets Inworld decide. Defaults to None (not set).
float | None
default:"UNSET"
Speaking rate multiplier, 0.5 to 1.5. Unset uses the voice’s natural rate. Defaults to None (not set).
float
default:"UNSET"
Seconds after which the WebSocket connection is reopened, to avoid server-side timeouts. Defaults to 300.0.