Skip to content

Speech Recognition Service API Documentation

Overview

This documentation covers the functionalities provided by the ISpeechRecognitionService interface, designed for handling speech recognition tasks. This service allows for asynchronous speech recording and processing, providing mechanisms to start and stop recordings, and to handle both real-time and post-processing recognition results.

Interface: ISpeechRecognitionService

ISpeechRecognitionService provides methods for starting and stopping speech recordings, both individually and globally. It extends IAsyncDisposable for proper asynchronous resource management.

Methods

StartRecordingAsync

Starts an asynchronous speech recording session with customizable options.

  • Signature 1: 
    Task<string> StartRecordingAsync(SpeechRecognitionOptions options, Action<SpeechRecognitionResult> callback, Action<string> stoppedCallback = null);
  • options: Configuration settings for the speech recognition session.
  • callback: A method to handle the recognition results.

  • stoppedCallback: An optional method invoked when the recording stops.

  • Signature 2: 

    Task<string> StartRecordingAsync(Action<SpeechRecognitionResult> callback, Action<string> stoppedCallback = null);

  • callback: A method to handle the recognition results.
  • stoppedCallback: An optional method invoked when the recording stops.

StopRecordingAsync

Stops a specific ongoing speech recording.

  • Signature: 
    Task StopRecordingAsync(string recordingId);
  • recordingId: The identifier of the recording session to be stopped.

StopAllRecordingsAsync

Stops all ongoing speech recordings.

  • Signature: 
    Task StopAllRecordingsAsync();

Class: SpeechRecognitionOptions

Configuration options for the speech recognition service.

Properties

  • CultureInfo: The culture information derived from the language setting, influencing recognition dialect and locale. Ignored during JSON serialization.
  • Lang: The language code for the speech recognition.
  • FinalResultsOnly: Specifies if only final recognition results should be returned.
  • Continuous: Indicates whether the recognition service should continue listening after the user stops speaking.