Setup Speech Processor dependencies

Editor

• Add a language model provider component to the scene.

  • If you want to load language model from StreamingAssets, add the Streaming Assets Language Model Provider component. Specify the path to the language models relative to the StreamingAssets folder and select the default language from the Language pop-up menu.

    To add a new language model, download it (e.g. from here), extract it to the Unity StreamingAssets directory and specify path to its content and its language in Streaming Assets Language Model Provider settings.

    

  • If you want to download remote zipped language model, add the Remote Language Model Provider component. Specify the address of the remote zip containing language model files and select the default language from the Language pop-up menu. If the contents of the language model (folders am, conf, graph etc.) are located in the root of the zip, leave the Entry field empty, otherwise specify the path to the contents of the language model inside the zip.

    

• Add speech source component to the scene.

  • If you want to use microphone, add Microphone Speech Source component.

    

  • If you want to use audio clip, add Audio Clip Speech Source component and assign an audio clip to the Clip field. Use uncompressed mono audio (go to audioclip import settings and set Force To Mono to true, Load Type to Decompress On Load).

    

  • If you want to use AudioListener as a speech source, add AudioListener Speech Source component and specify channel from which you want the audio data to be streamed.

Explicit initialization

Speech processor implicitly initializes its dependencies when started. To initialize dependency explicitly, use SpeechProcessorDependency.Initialize() method. For example:

using System;
using System.Collections;
using Recognissimo;
using UnityEngine;

public class ExplicitInitializationExample : MonoBehaviour
{
    // SpeechSource inherits SpeechProcessorDependency base class.
    [SerializeField]
    private SpeechSource speechSource;

    private IEnumerator Start()
    {
        yield return speechSource.Initialize(HandleTaskStarted, HandleInitializationFail);
    }

    private void HandleTaskStarted(string taskName, bool isLongRunning)
    {
        // Print only for coroutines.
        if (isLongRunning)
        {
            Debug.Log($"Starting task {taskName}");
        }
    }

    private void HandleInitializationFail(string taskName, Exception exception)
    {
        Debug.Log($"Task {taskName} failed with error {exception.Message}");
    }
}