Mobile SDK API

音声認識に使用する言語と音声データの入力を受け、その認識結果をモバイル環境でテキストに変換する Mobile SDK APIのユースケースを説明します。

Java

Android環境での、Javaベースの Mobile SDK APIのサンプルコードは次の通りです。

サンプルコードの保存先アドレス: https://github.com/NaverCloudPlatform/naverspeech-sdk-android
説明
- Main Activityクラス: SpeechRecognitionListenerを初期化し、その後にイベントを handleMessageから受けて処理します。
- SpeechRecognitionListenerを継承したクラス: 音声認識サーバへの接続、音声の伝達、認識結果発生などのイベントによる結果の処理方法を定義します。

// 1. Main Activityクラス
public class MainActivity extends Activity {
	private static final String TAG = MainActivity.class.getSimpleName();
	private static final String CLIENT_ID = "YOUR CLIENT ID"; // 「マイアプリケーション」で Client IDを確認し、ここに入力してください。
    private RecognitionHandler handler;
    private NaverRecognizer naverRecognizer;
    private TextView txtResult;
    private Button btnStart;
    private String mResult;
    private AudioWriterPCM writer;
    // Handle speech recognition Messages.
    private void handleMessage(Message msg) {
        switch (msg.what) {
            case R.id.clientReady: // 音声認識準備完了
                txtResult.setText("Connected");
                writer = new AudioWriterPCM(Environment.getExternalStorageDirectory().getAbsolutePath() + "/NaverSpeechTest");
                writer.open("Test");
                break;
            case R.id.audioRecording:
                writer.write((short[]) msg.obj);
                break;
            case R.id.partialResult:
                mResult = (String) (msg.obj);
                txtResult.setText(mResult);
                break;
            case R.id.finalResult: // 最終認識結果
            	SpeechRecognitionResult speechRecognitionResult = (SpeechRecognitionResult) msg.obj;
            	List<String> results = speechRecognitionResult.getResults();
            	StringBuilder strBuf = new StringBuilder();
            	for(String result : results) {
            		strBuf.append(result);
            		strBuf.append("\n");
            	}
                mResult = strBuf.toString();
                txtResult.setText(mResult);
                break;
            case R.id.recognitionError:
                if (writer != null) {
                    writer.close();
                }
                mResult = "Error code : " + msg.obj.toString();
                txtResult.setText(mResult);
                btnStart.setText(R.string.str_start);
                btnStart.setEnabled(true);
                break;
            case R.id.clientInactive:
                if (writer != null) {
                    writer.close();
                }
                btnStart.setText(R.string.str_start);
                btnStart.setEnabled(true);
                break;
        }
    }

    @Override
    protected void onCreate(Bundle savedInstanceState) {
        super.onCreate(savedInstanceState);
        setContentView(R.layout.activity_main);
        txtResult = (TextView) findViewById(R.id.txt_result);
        btnStart = (Button) findViewById(R.id.btn_start);
        handler = new RecognitionHandler(this);
        naverRecognizer = new NaverRecognizer(this, handler, CLIENT_ID);
        btnStart.setOnClickListener(new View.OnClickListener() {
            @Override
            public void onClick(View v) {
                if(!naverRecognizer.getSpeechRecognizer().isRunning()) {
                    mResult = "";
                    txtResult.setText("Connecting...");
                    btnStart.setText(R.string.str_stop);
                    naverRecognizer.recognize();
                } else {
                    Log.d(TAG, "stop and wait Final Result");
                    btnStart.setEnabled(false);
                    naverRecognizer.getSpeechRecognizer().stop();
                }
            }
        });
    }
    @Override
    protected void onStart() {
    	super.onStart(); // 音声認識サーバの初期化はこちら
    	naverRecognizer.getSpeechRecognizer().initialize();
    }
    @Override
    protected void onResume() {
        super.onResume();
        mResult = "";
        txtResult.setText("");
        btnStart.setText(R.string.str_start);
        btnStart.setEnabled(true);
    }
    @Override
    protected void onStop() {
    	super.onStop(); // 音声認識サーバ終了
    	naverRecognizer.getSpeechRecognizer().release();
    }
    // Declare handler for handling SpeechRecognizer thread's Messages.
    static class RecognitionHandler extends Handler {
        private final WeakReference<MainActivity> mActivity;
        RecognitionHandler(MainActivity activity) {
            mActivity = new WeakReference<MainActivity>(activity);
        }
        @Override
        public void handleMessage(Message msg) {
            MainActivity activity = mActivity.get();
            if (activity != null) {
                activity.handleMessage(msg);
            }
        }
    }
}

// 2. SpeechRecognitionListenerを継承したクラス
class NaverRecognizer implements SpeechRecognitionListener {
	private final static String TAG = NaverRecognizer.class.getSimpleName();
	private Handler mHandler;
	private SpeechRecognizer mRecognizer;
	public NaverRecognizer(Context context, Handler handler, String clientId) {
		this.mHandler = handler;
		try {
			mRecognizer = new SpeechRecognizer(context, clientId);
		} catch (SpeechRecognitionException e) {
			e.printStackTrace();
		}
		mRecognizer.setSpeechRecognitionListener(this);
	}
	public SpeechRecognizer getSpeechRecognizer() {
		return mRecognizer;
	}
	public void recognize() {
		try {
			mRecognizer.recognize(new SpeechConfig(LanguageType.KOREAN, EndPointDetectType.AUTO));
		} catch (SpeechRecognitionException e) {
			e.printStackTrace();
		}
	}
	@Override
	@WorkerThread
	public void onInactive() {
		Message msg = Message.obtain(mHandler, R.id.clientInactive);
		msg.sendToTarget();
	}
	@Override
	@WorkerThread
	public void onReady() {
		Message msg = Message.obtain(mHandler, R.id.clientReady);
		msg.sendToTarget();
	}
	@Override
	@WorkerThread
	public void onRecord(short[] speech) {
		Message msg = Message.obtain(mHandler, R.id.audioRecording, speech);
		msg.sendToTarget();
	}
	@Override
	@WorkerThread
	public void onPartialResult(String result) {
		Message msg = Message.obtain(mHandler, R.id.partialResult, result);
		msg.sendToTarget();
	}
	@Override
	@WorkerThread
	public void onEndPointDetected() {
		Log.d(TAG, "Event occurred : EndPointDetected");
	}
	@Override
	@WorkerThread
	public void onResult(SpeechRecognitionResult result) {
		Message msg = Message.obtain(mHandler, R.id.finalResult, result);
		msg.sendToTarget();
	}
	@Override
	@WorkerThread
	public void onError(int errorCode) {
		Message msg = Message.obtain(mHandler, R.id.recognitionError, errorCode);
		msg.sendToTarget();
	}
	@Override
	@WorkerThread
	public void onEndPointDetectTypeSelected(EndPointDetectType epdType) {
		Message msg = Message.obtain(mHandler, R.id.endPointDetectTypeSelected, epdType);
		msg.sendToTarget();
	}
}

ObjectiveC

iOS環境での、ObjectiveCベースの Mobile SDK APIのサンプルコードは次の通りです。

サンプルコードの保存先アドレス: https://github.com/NaverCloudPlatform/naverspeech-sdk-ios

import UIKit
import NaverSpeech
import Common
let ClientID = "YOUR_CLIENT_ID"
class AutoViewController: UIViewController {
    required init?(coder aDecoder: NSCoder) { // NSKRecognizer初期化に必要な NSKRecognizerConfigurationを作成
        let configuration = NSKRecognizerConfiguration(clientID: ClientID)
        configuration?.canQuestionDetected = true
        self.speechRecognizer = NSKRecognizer(configuration: configuration)
        super.init(coder: aDecoder)
        self.speechRecognizer.delegate = self
    }
    override func viewDidLoad() {
        super.viewDidLoad()
        self.setupLanguagePicker()
    }
    override func didReceiveMemoryWarning() {
        super.didReceiveMemoryWarning()
        if self.isViewLoaded && self.view.window == nil {
            self.view = nil
        }
    }
    override func viewDidLayoutSubviews() {
        super.viewDidLayoutSubviews()
        let x = languagePickerButton.frame.minX
        let y = languagePickerButton.frame.maxY
        self.pickerView.frame = CGRect.init(x: x, y: y, width: languagePickerButton.bounds.size.width, height: self.pickerView.bounds.size.height)
    }
    @IBAction func languagePickerButtonTapped(_ sender: Any) {
        self.pickerView.isHidden = false
    }
    @IBAction func recognitionButtonTapped(_ sender: Any) { // ボタンを押すと音声認識を開始
        if self.speechRecognizer.isRunning {
            self.speechRecognizer.stop()
        } else {
            self.speechRecognizer.start(with: self.languages.selectedLanguage)
            self.recognitionButton.isEnabled = false
            self.statusLabel.text = "Connecting......"
        }
    }
    @IBOutlet weak var languagePickerButton: UIButton!
    @IBOutlet weak var recognitionResultLabel: UILabel!
    @IBOutlet weak var recognitionButton: UIButton!
    @IBOutlet weak var statusLabel: UILabel!
    fileprivate let speechRecognizer: NSKRecognizer
    fileprivate let languages = Languages()
    fileprivate let pickerView = UIPickerView()
}

extension AutoViewController: NSKRecognizerDelegate { //NSKRecognizerDelegate protocolを実装

    public func recognizerDidEnterReady(_ aRecognizer: NSKRecognizer!) {
        print("Event occurred: Ready")
        self.statusLabel.text = "Connected"
        self.recognitionResultLabel.text = "Recognizing......"
        self.setRecognitionButtonTitle(withText: "Stop", color: .red)
        self.recognitionButton.isEnabled = true
    }
    public func recognizerDidDetectEndPoint(_ aRecognizer: NSKRecognizer!) {
        print("Event occurred: End point detected")
    }
    public func recognizerDidEnterInactive(_ aRecognizer: NSKRecognizer!) {
        print("Event occurred: Inactive")
        self.setRecognitionButtonTitle(withText: "Record", color: .blue)
        self.recognitionButton.isEnabled = true
        self.statusLabel.text = ""
    }
    public func recognizer(_ aRecognizer: NSKRecognizer!, didRecordSpeechData aSpeechData: Data!) {
        print("Record speech data, data size: \(aSpeechData.count)")
    }
    public func recognizer(_ aRecognizer: NSKRecognizer!, didReceivePartialResult aResult: String!) {
        print("Partial result: \(aResult)")
        self.recognitionResultLabel.text = aResult
    }
    public func recognizer(_ aRecognizer: NSKRecognizer!, didReceiveError aError: Error!) {
        print("Error: \(aError)")
        self.setRecognitionButtonTitle(withText: "Record", color: .blue)
        self.recognitionButton.isEnabled = true
    }
    public func recognizer(_ aRecognizer: NSKRecognizer!, didReceive aResult: NSKRecognizedResult!) {
        print("Final result: \(aResult)")
        if let result = aResult.results.first as? String {
            self.recognitionResultLabel.text = "Result: " + result
        }
    }
}
extension AutoViewController: UIPickerViewDelegate, UIPickerViewDataSource {
    func numberOfComponents(in pickerView: UIPickerView) -> Int {
        return 1
    }
    func pickerView(_ pickerView: UIPickerView, numberOfRowsInComponent component: Int) -> Int {
        return self.languages.count
    }
    func pickerView(_ pickerView: UIPickerView, titleForRow row: Int, forComponent component: Int) -> String? {
        return languages.languageString(at: row)
    }
    func pickerView(_ pickerView: UIPickerView, didSelectRow row: Int, inComponent component: Int) {
        languages.selectLanguage(at: row)
        languagePickerButton.setTitle(languages.selectedLanguageString, for: .normal)
        self.pickerView.isHidden = true
        if self.speechRecognizer.isRunning { //音声の認識中に言語が変更されると音声の認識を直ちに停止(cancel)
            self.speechRecognizer.cancel()
            self.recognitionResultLabel.text = "Canceled"
            self.setRecognitionButtonTitle(withText: "Record", color: .blue)
            self.recognitionButton.isEnabled = true
        }
    }
}
fileprivate extension AutoViewController {
    func setupLanguagePicker() {
        self.view.addSubview(self.pickerView)
        self.pickerView.dataSource = self
        self.pickerView.delegate = self
        self.pickerView.showsSelectionIndicator = true
        self.pickerView.backgroundColor = UIColor.white
        self.pickerView.isHidden = true
    }
    func setRecognitionButtonTitle(withText text: String, color: UIColor) {
        self.recognitionButton.setTitle(text, for: .normal)
        self.recognitionButton.setTitleColor(color, for: .normal)
    }
}

この記事は役に立ちましたか?

Mobile SDK API
- Java
- ObjectiveC