Speech Recognition

Listen for speech and recognize what was said using the settings for this operation.

Behavior and notes

• When saveUtteranceAudio is true, include audioRecording and use audioRecording.recordingOutput to configure the saved utterance.

Developer guidance

Speech Recognition selects the recognition engine and locale, controls whether recognition uses a live stream or a file, and optionally saves utterance audio. The task object controls recognition options such as partial results and punctuation. Recording-side settings belong in audioRecording.

Example in context

Illustrative JSON excerpt (not a complete configuration document):

{
  "speechRecognition": {
    "speechRecognitionLocale": "en-US",
    "task": {
      "reportPartialResults": true,
      "autoPunctuate": false
    },
    "saveUtteranceAudio": false
  }
}

Recognition task options are nested under task, not directly under speechRecognition.

Properties