speech/io.github.maniramezan.kommon.speech/SpeechRecognitionConfiguration

SpeechRecognitionConfiguration

data class SpeechRecognitionConfiguration(val locale: Locale, val preferOnDevice: Boolean = true, val reportsAlternatives: Boolean = true, val requestsWordDetails: Boolean = true, val completeSilenceMillis: Long? = null, val maxAlternatives: Int = DEFAULT_MAX_ALTERNATIVES)

Settings for one speech recognition run.

Constructors

constructor(locale: Locale, preferOnDevice: Boolean = true, reportsAlternatives: Boolean = true, requestsWordDetails: Boolean = true, completeSilenceMillis: Long? = null, maxAlternatives: Int = DEFAULT_MAX_ALTERNATIVES)

Properties

completeSilenceMillis

val completeSilenceMillis: Long?

A hint for how long the speaker must be silent before the input is considered complete, or null to use the recognizer's default.

locale

val locale: Locale

The language to recognize.

maxAlternatives

val maxAlternatives: Int

The maximum number of transcriptions to request when reportsAlternatives is true; the best one plus alternatives.

preferOnDevice

val preferOnDevice: Boolean

Use on-device recognition when it is available, falling back to the system recognizer's offline preference otherwise.

reportsAlternatives

val reportsAlternatives: Boolean

Whether to report other plausible transcriptions.

requestsWordDetails

val requestsWordDetails: Boolean

Whether to request per-word confidence and timing where the platform supports it (Android 14 and later).