Home | History | Annotate | Download | only in speech
      1 /*
      2  * Copyright (C) 2010 The Android Open Source Project
      3  *
      4  * Licensed under the Apache License, Version 2.0 (the "License"); you may not
      5  * use this file except in compliance with the License. You may obtain a copy of
      6  * the License at
      7  *
      8  * http://www.apache.org/licenses/LICENSE-2.0
      9  *
     10  * Unless required by applicable law or agreed to in writing, software
     11  * distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
     12  * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
     13  * License for the specific language governing permissions and limitations under
     14  * the License.
     15  */
     16 
     17 package android.speech;
     18 
     19 import android.annotation.SdkConstant;
     20 import android.annotation.SdkConstant.SdkConstantType;
     21 import android.app.Service;
     22 import android.content.Intent;
     23 import android.content.pm.PackageManager;
     24 import android.os.Bundle;
     25 import android.os.Handler;
     26 import android.os.IBinder;
     27 import android.os.Message;
     28 import android.os.RemoteException;
     29 import android.util.Log;
     30 
     31 import java.lang.ref.WeakReference;
     32 
     33 /**
     34  * This class provides a base class for recognition service implementations. This class should be
     35  * extended only in case you wish to implement a new speech recognizer. Please note that the
     36  * implementation of this service is stateless.
     37  */
     38 public abstract class RecognitionService extends Service {
     39     /**
     40      * The {@link Intent} that must be declared as handled by the service.
     41      */
     42     @SdkConstant(SdkConstantType.SERVICE_ACTION)
     43     public static final String SERVICE_INTERFACE = "android.speech.RecognitionService";
     44 
     45     /**
     46      * Name under which a RecognitionService component publishes information about itself.
     47      * This meta-data should reference an XML resource containing a
     48      * <code>&lt;{@link android.R.styleable#RecognitionService recognition-service}&gt;</code> tag.
     49      */
     50     public static final String SERVICE_META_DATA = "android.speech";
     51 
     52     /** Log messages identifier */
     53     private static final String TAG = "RecognitionService";
     54 
     55     /** Debugging flag */
     56     private static final boolean DBG = false;
     57 
     58     /** Binder of the recognition service */
     59     private RecognitionServiceBinder mBinder = new RecognitionServiceBinder(this);
     60 
     61     /**
     62      * The current callback of an application that invoked the
     63      * {@link RecognitionService#onStartListening(Intent, Callback)} method
     64      */
     65     private Callback mCurrentCallback = null;
     66 
     67     private static final int MSG_START_LISTENING = 1;
     68 
     69     private static final int MSG_STOP_LISTENING = 2;
     70 
     71     private static final int MSG_CANCEL = 3;
     72 
     73     private static final int MSG_RESET = 4;
     74 
     75     private final Handler mHandler = new Handler() {
     76         @Override
     77         public void handleMessage(Message msg) {
     78             switch (msg.what) {
     79                 case MSG_START_LISTENING:
     80                     StartListeningArgs args = (StartListeningArgs) msg.obj;
     81                     dispatchStartListening(args.mIntent, args.mListener);
     82                     break;
     83                 case MSG_STOP_LISTENING:
     84                     dispatchStopListening((IRecognitionListener) msg.obj);
     85                     break;
     86                 case MSG_CANCEL:
     87                     dispatchCancel((IRecognitionListener) msg.obj);
     88                     break;
     89                 case MSG_RESET:
     90                     dispatchClearCallback();
     91                     break;
     92             }
     93         }
     94     };
     95 
     96     private void dispatchStartListening(Intent intent, final IRecognitionListener listener) {
     97         if (mCurrentCallback == null) {
     98             if (DBG) Log.d(TAG, "created new mCurrentCallback, listener = " + listener.asBinder());
     99             try {
    100                 listener.asBinder().linkToDeath(new IBinder.DeathRecipient() {
    101                     @Override
    102                     public void binderDied() {
    103                         mHandler.sendMessage(mHandler.obtainMessage(MSG_CANCEL, listener));
    104                     }
    105                 }, 0);
    106             } catch (RemoteException re) {
    107                 Log.e(TAG, "dead listener on startListening");
    108                 return;
    109             }
    110             mCurrentCallback = new Callback(listener);
    111             RecognitionService.this.onStartListening(intent, mCurrentCallback);
    112         } else {
    113             try {
    114                 listener.onError(SpeechRecognizer.ERROR_RECOGNIZER_BUSY);
    115             } catch (RemoteException e) {
    116                 Log.d(TAG, "onError call from startListening failed");
    117             }
    118             Log.i(TAG, "concurrent startListening received - ignoring this call");
    119         }
    120     }
    121 
    122     private void dispatchStopListening(IRecognitionListener listener) {
    123         try {
    124             if (mCurrentCallback == null) {
    125                 listener.onError(SpeechRecognizer.ERROR_CLIENT);
    126                 Log.w(TAG, "stopListening called with no preceding startListening - ignoring");
    127             } else if (mCurrentCallback.mListener.asBinder() != listener.asBinder()) {
    128                 listener.onError(SpeechRecognizer.ERROR_RECOGNIZER_BUSY);
    129                 Log.w(TAG, "stopListening called by other caller than startListening - ignoring");
    130             } else { // the correct state
    131                 RecognitionService.this.onStopListening(mCurrentCallback);
    132             }
    133         } catch (RemoteException e) { // occurs if onError fails
    134             Log.d(TAG, "onError call from stopListening failed");
    135         }
    136     }
    137 
    138     private void dispatchCancel(IRecognitionListener listener) {
    139         if (mCurrentCallback == null) {
    140             if (DBG) Log.d(TAG, "cancel called with no preceding startListening - ignoring");
    141         } else if (mCurrentCallback.mListener.asBinder() != listener.asBinder()) {
    142             Log.w(TAG, "cancel called by client who did not call startListening - ignoring");
    143         } else { // the correct state
    144             RecognitionService.this.onCancel(mCurrentCallback);
    145             mCurrentCallback = null;
    146             if (DBG) Log.d(TAG, "canceling - setting mCurrentCallback to null");
    147         }
    148     }
    149 
    150     private void dispatchClearCallback() {
    151         mCurrentCallback = null;
    152     }
    153 
    154     private class StartListeningArgs {
    155         public final Intent mIntent;
    156 
    157         public final IRecognitionListener mListener;
    158 
    159         public StartListeningArgs(Intent intent, IRecognitionListener listener) {
    160             this.mIntent = intent;
    161             this.mListener = listener;
    162         }
    163     }
    164 
    165     /**
    166      * Checks whether the caller has sufficient permissions
    167      *
    168      * @param listener to send the error message to in case of error
    169      * @return {@code true} if the caller has enough permissions, {@code false} otherwise
    170      */
    171     private boolean checkPermissions(IRecognitionListener listener) {
    172         if (DBG) Log.d(TAG, "checkPermissions");
    173         if (RecognitionService.this.checkCallingOrSelfPermission(android.Manifest.permission.
    174                 RECORD_AUDIO) == PackageManager.PERMISSION_GRANTED) {
    175             return true;
    176         }
    177         try {
    178             Log.e(TAG, "call for recognition service without RECORD_AUDIO permissions");
    179             listener.onError(SpeechRecognizer.ERROR_INSUFFICIENT_PERMISSIONS);
    180         } catch (RemoteException re) {
    181             Log.e(TAG, "sending ERROR_INSUFFICIENT_PERMISSIONS message failed", re);
    182         }
    183         return false;
    184     }
    185 
    186     /**
    187      * Notifies the service that it should start listening for speech.
    188      *
    189      * @param recognizerIntent contains parameters for the recognition to be performed. The intent
    190      *        may also contain optional extras, see {@link RecognizerIntent}. If these values are
    191      *        not set explicitly, default values should be used by the recognizer.
    192      * @param listener that will receive the service's callbacks
    193      */
    194     protected abstract void onStartListening(Intent recognizerIntent, Callback listener);
    195 
    196     /**
    197      * Notifies the service that it should cancel the speech recognition.
    198      */
    199     protected abstract void onCancel(Callback listener);
    200 
    201     /**
    202      * Notifies the service that it should stop listening for speech. Speech captured so far should
    203      * be recognized as if the user had stopped speaking at this point. This method is only called
    204      * if the application calls it explicitly.
    205      */
    206     protected abstract void onStopListening(Callback listener);
    207 
    208     @Override
    209     public final IBinder onBind(final Intent intent) {
    210         if (DBG) Log.d(TAG, "onBind, intent=" + intent);
    211         return mBinder;
    212     }
    213 
    214     @Override
    215     public void onDestroy() {
    216         if (DBG) Log.d(TAG, "onDestroy");
    217         mCurrentCallback = null;
    218         mBinder.clearReference();
    219         super.onDestroy();
    220     }
    221 
    222     /**
    223      * This class receives callbacks from the speech recognition service and forwards them to the
    224      * user. An instance of this class is passed to the
    225      * {@link RecognitionService#onStartListening(Intent, Callback)} method. Recognizers may call
    226      * these methods on any thread.
    227      */
    228     public class Callback {
    229         private final IRecognitionListener mListener;
    230 
    231         private Callback(IRecognitionListener listener) {
    232             mListener = listener;
    233         }
    234 
    235         /**
    236          * The service should call this method when the user has started to speak.
    237          */
    238         public void beginningOfSpeech() throws RemoteException {
    239             if (DBG) Log.d(TAG, "beginningOfSpeech");
    240             mListener.onBeginningOfSpeech();
    241         }
    242 
    243         /**
    244          * The service should call this method when sound has been received. The purpose of this
    245          * function is to allow giving feedback to the user regarding the captured audio.
    246          *
    247          * @param buffer a buffer containing a sequence of big-endian 16-bit integers representing a
    248          *        single channel audio stream. The sample rate is implementation dependent.
    249          */
    250         public void bufferReceived(byte[] buffer) throws RemoteException {
    251             mListener.onBufferReceived(buffer);
    252         }
    253 
    254         /**
    255          * The service should call this method after the user stops speaking.
    256          */
    257         public void endOfSpeech() throws RemoteException {
    258             mListener.onEndOfSpeech();
    259         }
    260 
    261         /**
    262          * The service should call this method when a network or recognition error occurred.
    263          *
    264          * @param error code is defined in {@link SpeechRecognizer}
    265          */
    266         public void error(int error) throws RemoteException {
    267             Message.obtain(mHandler, MSG_RESET).sendToTarget();
    268             mListener.onError(error);
    269         }
    270 
    271         /**
    272          * The service should call this method when partial recognition results are available. This
    273          * method can be called at any time between {@link #beginningOfSpeech()} and
    274          * {@link #results(Bundle)} when partial results are ready. This method may be called zero,
    275          * one or multiple times for each call to {@link SpeechRecognizer#startListening(Intent)},
    276          * depending on the speech recognition service implementation.
    277          *
    278          * @param partialResults the returned results. To retrieve the results in
    279          *        ArrayList&lt;String&gt; format use {@link Bundle#getStringArrayList(String)} with
    280          *        {@link SpeechRecognizer#RESULTS_RECOGNITION} as a parameter
    281          */
    282         public void partialResults(Bundle partialResults) throws RemoteException {
    283             mListener.onPartialResults(partialResults);
    284         }
    285 
    286         /**
    287          * The service should call this method when the endpointer is ready for the user to start
    288          * speaking.
    289          *
    290          * @param params parameters set by the recognition service. Reserved for future use.
    291          */
    292         public void readyForSpeech(Bundle params) throws RemoteException {
    293             mListener.onReadyForSpeech(params);
    294         }
    295 
    296         /**
    297          * The service should call this method when recognition results are ready.
    298          *
    299          * @param results the recognition results. To retrieve the results in {@code
    300          *        ArrayList&lt;String&gt;} format use {@link Bundle#getStringArrayList(String)} with
    301          *        {@link SpeechRecognizer#RESULTS_RECOGNITION} as a parameter
    302          */
    303         public void results(Bundle results) throws RemoteException {
    304             Message.obtain(mHandler, MSG_RESET).sendToTarget();
    305             mListener.onResults(results);
    306         }
    307 
    308         /**
    309          * The service should call this method when the sound level in the audio stream has changed.
    310          * There is no guarantee that this method will be called.
    311          *
    312          * @param rmsdB the new RMS dB value
    313          */
    314         public void rmsChanged(float rmsdB) throws RemoteException {
    315             mListener.onRmsChanged(rmsdB);
    316         }
    317     }
    318 
    319     /** Binder of the recognition service */
    320     private static final class RecognitionServiceBinder extends IRecognitionService.Stub {
    321         private final WeakReference<RecognitionService> mServiceRef;
    322 
    323         public RecognitionServiceBinder(RecognitionService service) {
    324             mServiceRef = new WeakReference<RecognitionService>(service);
    325         }
    326 
    327         @Override
    328         public void startListening(Intent recognizerIntent, IRecognitionListener listener) {
    329             if (DBG) Log.d(TAG, "startListening called by:" + listener.asBinder());
    330             final RecognitionService service = mServiceRef.get();
    331             if (service != null && service.checkPermissions(listener)) {
    332                 service.mHandler.sendMessage(Message.obtain(service.mHandler,
    333                         MSG_START_LISTENING, service.new StartListeningArgs(
    334                                 recognizerIntent, listener)));
    335             }
    336         }
    337 
    338         @Override
    339         public void stopListening(IRecognitionListener listener) {
    340             if (DBG) Log.d(TAG, "stopListening called by:" + listener.asBinder());
    341             final RecognitionService service = mServiceRef.get();
    342             if (service != null && service.checkPermissions(listener)) {
    343                 service.mHandler.sendMessage(Message.obtain(service.mHandler,
    344                         MSG_STOP_LISTENING, listener));
    345             }
    346         }
    347 
    348         @Override
    349         public void cancel(IRecognitionListener listener) {
    350             if (DBG) Log.d(TAG, "cancel called by:" + listener.asBinder());
    351             final RecognitionService service = mServiceRef.get();
    352             if (service != null && service.checkPermissions(listener)) {
    353                 service.mHandler.sendMessage(Message.obtain(service.mHandler,
    354                         MSG_CANCEL, listener));
    355             }
    356         }
    357 
    358         public void clearReference() {
    359             mServiceRef.clear();
    360         }
    361     }
    362 }
    363