Usage guide
Setup, environment variables, and Python/JavaScript/Go usage examples.
NvidiaSTT
Speech-to-text with NVIDIA Riva Parakeet models hosted on NVIDIA’s cloud. Good for fast streaming English transcription. Authenticates with NVIDIA_API_KEY from your environment, or an explicitapi_key.
Fields
str
default:"UNSET"
Riva ASR model name, e.g. “parakeet-1.1b-en-US-asr-streaming-silero-vad-sortformer”. Defaults to ‘parakeet-1.1b-en-US-asr-streaming-silero-vad-sortformer’.
str
default:"UNSET"
Riva gRPC server address as host:port. Defaults to ‘grpc.nvcf.nvidia.com:443’.
str
default:"UNSET"
NVIDIA cloud function ID of the hosted model to call. Must match the model. Defaults to ‘1598d209-5e27-4d3c-8079-4751568b1081’.
str
default:"UNSET"
BCP-47 language code of the speech, e.g. “en-US”. Defaults to ‘en-US’.
int | None
default:"UNSET"
Sample rate of incoming audio in Hz. Leave unset to detect it automatically. Defaults to None (not set).
bool
default:"UNSET"
Connect to the server over TLS. Turn off only for a local unsecured Riva server. Defaults to True.
bool
default:"UNSET"
Mask profanity in transcripts. Defaults to False.
bool
default:"UNSET"
Add punctuation and capitalization to transcripts. Defaults to True.
bool
default:"True"
Whether partial transcripts are also sent to the client as the user speaks, not just the final one. Defaults to True.