Skip to main content

​
DeepgramSTT

Deepgram speech-to-text (STT) provider. Creates a configured Deepgram STT instance for use in a voice agent.

​
Constructor

​
apiKey
string | null
Deepgram API key. Falls back to the DEEPGRAM_API_KEY environment variable, then null if neither is set.
​
model
DeepgramSTTModel | string
default:"nova-2"
Deepgram recognition model (e.g. 'nova-2', 'nova-3', 'enhanced').
​
language
string
default:"en-US"
BCP-47 language code for transcription (e.g. 'en-US', 'es', 'fr').
​
interimResults
boolean
default:"true"
Emit partial (interim) transcript results before a segment is finalized.
​
punctuate
boolean
default:"true"
Add punctuation to transcripts.
​
smartFormat
boolean
default:"true"
Apply smart formatting (dates, times, currency, etc.) to transcripts.
​
sampleRate
number
default:"48000"
Audio sample rate in Hz. Note: this is fixed at 48000 and has no effect; non-48000 values are ignored.
​
endpointing
number
default:"50"
Silence duration in milliseconds used to detect end of speech (endpointing).
​
fillerWords
boolean
default:"true"
Include filler words (e.g. “uh”, “um”) in transcripts.
​
keywords
string[] | null
default:"null"
Keywords to boost recognition for, optionally with intensifiers (e.g. 'Deepgram:2'). null to disable.
​
keyterm
string[] | null
default:"null"
Key terms to boost recognition for (newer keyterm prompting). null to disable.
​
profanityFilter
boolean
default:"false"
Remove profanity from transcripts.
​
numerals
boolean
default:"false"
Convert spoken numbers into numeric digits.
​
tag
string | string[] | null
default:"null"
Free-form tag(s) attached to the request for tracking/usage reporting. null to disable.
​
enableDiarization
boolean
default:"false"
Enable speaker diarization (label which speaker said what).
​
redact
string | string[] | null
default:"null"
Redaction categories to remove from transcripts (e.g. 'pci', 'ssn', 'numbers'). null to disable.
​
baseUrl
string
default:"wss://api.deepgram.com/v1/listen"
Deepgram streaming endpoint URL.

​
close

Release any resources held by the provider.

​
closeEmitter

Close the emitter: drop all handlers and stop emitting. Waits up to 2 seconds for in-flight async handlers to settle before returning.

​
emit

​
event
string
required

​
getRuntimeConfig

​
returns
Record<string, any>

​
listenerCount

​
event
string
required
​
returns
number

​
off

​
event
string
required
​
handler
EventHandler
required

​
on

​
event
string
required
​
handler
EventHandler
required
​
returns
EventHandler

​
once

​
event
string
required
​
handler
EventHandler
required
​
returns
EventHandler

​
onSttTranscript

Register a callback to receive transcripts as they are produced.
​
callback
SttTranscriptCallback
required

​
processAudio

Process a buffer of audio frames. Override in a custom provider.

​
removeAllListeners

​
event
string

​
streamTranscribe

Transcribe a streaming audio source, yielding STTResponses. Override in a custom provider.
​
returns
AsyncIterator<STTResponse>

​
DeepgramSTTModel

Supported Deepgram speech-to-text model names.