Interface STTOptions

interface STTOptions {
    apiKey?: string;
    audioChunkDuration?: number;
    baseURL?: string;
    enableLogging?: boolean;
    httpSession?: STTHTTPSession;
    includeLanguageDetection?: boolean;
    includeTimestamps?: boolean;
    keyterms?: string[];
    languageCode?: string;
    model?: string;
    modelId?: string;
    noVerbatim?: boolean;
    sampleRate?: STTRealtimeSampleRates;
    secondaryLanguages?: string[];
    serverVad?: null | VADOptions;
    tagAudioEvents?: boolean;
    useRealtime?: boolean;
}

Properties

apiKey? audioChunkDuration? baseURL? enableLogging? httpSession? includeLanguageDetection? includeTimestamps? keyterms? languageCode? model? modelId? noVerbatim? sampleRate? secondaryLanguages? serverVad? tagAudioEvents? useRealtime?

Properties

Optional apiKey

apiKey?: string

Optional audioChunkDuration

audioChunkDuration?: number

Duration of each outgoing realtime audio chunk in milliseconds. Must be a positive integer. Defaults to 50. Larger chunks reduce message frequency but increase buffering latency. Flushes send any shorter remaining chunk before committing. Only used for Scribe v2 realtime.

Optional baseURL

baseURL?: string

Optional enableLogging

enableLogging?: boolean

Optional httpSession

httpSession?: STTHTTPSession

Optional includeLanguageDetection

includeLanguageDetection?: boolean

Whether committed transcripts report the detected language. Defaults to true when no languageCode is set and false otherwise. Only supported for Scribe v2 realtime.

Optional includeTimestamps

includeTimestamps?: boolean

Optional keyterms

keyterms?: string[]

Keywords or phrases to bias transcription towards. Scribe v2 batch accepts up to 1000 keyterms of at most 50 characters each; realtime accepts up to 50 keyterms of at most 20 characters each. Usage incurs additional costs.

Optional languageCode

languageCode?: string

Optional model

model?: string

Optional modelId

modelId?: string

Deprecated

Use model instead.

Optional noVerbatim

noVerbatim?: boolean

Optional sampleRate

sampleRate?: STTRealtimeSampleRates

Optional secondaryLanguages

secondaryLanguages?: string[]

Language codes to constrain speech prediction to in addition to languageCode. Useful for bilingual applications where audio switches between a primary and a limited set of secondary languages. Only supported for Scribe v2 realtime. When omitted, the model predicts from its full set of supported languages.

Optional serverVad

serverVad?: null | VADOptions

Optional tagAudioEvents

tagAudioEvents?: boolean

Optional useRealtime

useRealtime?: boolean

Settings

Member Visibility

Theme

On This Page

apiKeyaudioChunkDurationbaseURLenableLogginghttpSessionincludeLanguageDetectionincludeTimestampskeytermslanguageCodemodelmodelIdnoVerbatimsampleRatesecondaryLanguagesserverVadtagAudioEventsuseRealtime