Home | History | Annotate | Download | only in tts
      1 /*
      2  * Copyright (C) 2011 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 package android.speech.tts;
     17 
     18 import android.annotation.NonNull;
     19 import android.app.Service;
     20 import android.content.Intent;
     21 import android.media.AudioAttributes;
     22 import android.media.AudioManager;
     23 import android.net.Uri;
     24 import android.os.Binder;
     25 import android.os.Bundle;
     26 import android.os.Handler;
     27 import android.os.HandlerThread;
     28 import android.os.IBinder;
     29 import android.os.Looper;
     30 import android.os.Message;
     31 import android.os.MessageQueue;
     32 import android.os.ParcelFileDescriptor;
     33 import android.os.RemoteCallbackList;
     34 import android.os.RemoteException;
     35 import android.provider.Settings;
     36 import android.speech.tts.TextToSpeech.Engine;
     37 import android.text.TextUtils;
     38 import android.util.Log;
     39 
     40 import java.io.FileOutputStream;
     41 import java.io.IOException;
     42 import java.util.ArrayList;
     43 import java.util.HashMap;
     44 import java.util.HashSet;
     45 import java.util.List;
     46 import java.util.Locale;
     47 import java.util.MissingResourceException;
     48 import java.util.Set;
     49 
     50 
     51 /**
     52  * Abstract base class for TTS engine implementations. The following methods
     53  * need to be implemented:
     54  * <ul>
     55  * <li>{@link #onIsLanguageAvailable}</li>
     56  * <li>{@link #onLoadLanguage}</li>
     57  * <li>{@link #onGetLanguage}</li>
     58  * <li>{@link #onSynthesizeText}</li>
     59  * <li>{@link #onStop}</li>
     60  * </ul>
     61  * The first three deal primarily with language management, and are used to
     62  * query the engine for it's support for a given language and indicate to it
     63  * that requests in a given language are imminent.
     64  *
     65  * {@link #onSynthesizeText} is central to the engine implementation. The
     66  * implementation should synthesize text as per the request parameters and
     67  * return synthesized data via the supplied callback. This class and its helpers
     68  * will then consume that data, which might mean queuing it for playback or writing
     69  * it to a file or similar. All calls to this method will be on a single thread,
     70  * which will be different from the main thread of the service. Synthesis must be
     71  * synchronous which means the engine must NOT hold on to the callback or call any
     72  * methods on it after the method returns.
     73  *
     74  * {@link #onStop} tells the engine that it should stop
     75  * all ongoing synthesis, if any. Any pending data from the current synthesis
     76  * will be discarded.
     77  *
     78  * {@link #onGetLanguage} is not required as of JELLYBEAN_MR2 (API 18) and later, it is only
     79  * called on earlier versions of Android.
     80  *
     81  * API Level 20 adds support for Voice objects. Voices are an abstraction that allow the TTS
     82  * service to expose multiple backends for a single locale. Each one of them can have a different
     83  * features set. In order to fully take advantage of voices, an engine should implement
     84  * the following methods:
     85  * <ul>
     86  * <li>{@link #onGetVoices()}</li>
     87  * <li>{@link #onIsValidVoiceName(String)}</li>
     88  * <li>{@link #onLoadVoice(String)}</li>
     89  * <li>{@link #onGetDefaultVoiceNameFor(String, String, String)}</li>
     90  * </ul>
     91  * The first three methods are siblings of the {@link #onGetLanguage},
     92  * {@link #onIsLanguageAvailable} and {@link #onLoadLanguage} methods. The last one,
     93  * {@link #onGetDefaultVoiceNameFor(String, String, String)} is a link between locale and voice
     94  * based methods. Since API level 21 {@link TextToSpeech#setLanguage} is implemented by
     95  * calling {@link TextToSpeech#setVoice} with the voice returned by
     96  * {@link #onGetDefaultVoiceNameFor(String, String, String)}.
     97  *
     98  * If the client uses a voice instead of a locale, {@link SynthesisRequest} will contain the
     99  * requested voice name.
    100  *
    101  * The default implementations of Voice-related methods implement them using the
    102  * pre-existing locale-based implementation.
    103  */
    104 public abstract class TextToSpeechService extends Service {
    105 
    106     private static final boolean DBG = false;
    107     private static final String TAG = "TextToSpeechService";
    108 
    109     private static final String SYNTH_THREAD_NAME = "SynthThread";
    110 
    111     private SynthHandler mSynthHandler;
    112     // A thread and it's associated handler for playing back any audio
    113     // associated with this TTS engine. Will handle all requests except synthesis
    114     // to file requests, which occur on the synthesis thread.
    115     @NonNull private AudioPlaybackHandler mAudioPlaybackHandler;
    116     private TtsEngines mEngineHelper;
    117 
    118     private CallbackMap mCallbacks;
    119     private String mPackageName;
    120 
    121     private final Object mVoicesInfoLock = new Object();
    122 
    123     @Override
    124     public void onCreate() {
    125         if (DBG) Log.d(TAG, "onCreate()");
    126         super.onCreate();
    127 
    128         SynthThread synthThread = new SynthThread();
    129         synthThread.start();
    130         mSynthHandler = new SynthHandler(synthThread.getLooper());
    131 
    132         mAudioPlaybackHandler = new AudioPlaybackHandler();
    133         mAudioPlaybackHandler.start();
    134 
    135         mEngineHelper = new TtsEngines(this);
    136 
    137         mCallbacks = new CallbackMap();
    138 
    139         mPackageName = getApplicationInfo().packageName;
    140 
    141         String[] defaultLocale = getSettingsLocale();
    142 
    143         // Load default language
    144         onLoadLanguage(defaultLocale[0], defaultLocale[1], defaultLocale[2]);
    145     }
    146 
    147     @Override
    148     public void onDestroy() {
    149         if (DBG) Log.d(TAG, "onDestroy()");
    150 
    151         // Tell the synthesizer to stop
    152         mSynthHandler.quit();
    153         // Tell the audio playback thread to stop.
    154         mAudioPlaybackHandler.quit();
    155         // Unregister all callbacks.
    156         mCallbacks.kill();
    157 
    158         super.onDestroy();
    159     }
    160 
    161     /**
    162      * Checks whether the engine supports a given language.
    163      *
    164      * Can be called on multiple threads.
    165      *
    166      * Its return values HAVE to be consistent with onLoadLanguage.
    167      *
    168      * @param lang ISO-3 language code.
    169      * @param country ISO-3 country code. May be empty or null.
    170      * @param variant Language variant. May be empty or null.
    171      * @return Code indicating the support status for the locale.
    172      *         One of {@link TextToSpeech#LANG_AVAILABLE},
    173      *         {@link TextToSpeech#LANG_COUNTRY_AVAILABLE},
    174      *         {@link TextToSpeech#LANG_COUNTRY_VAR_AVAILABLE},
    175      *         {@link TextToSpeech#LANG_MISSING_DATA}
    176      *         {@link TextToSpeech#LANG_NOT_SUPPORTED}.
    177      */
    178     protected abstract int onIsLanguageAvailable(String lang, String country, String variant);
    179 
    180     /**
    181      * Returns the language, country and variant currently being used by the TTS engine.
    182      *
    183      * This method will be called only on Android 4.2 and before (API <= 17). In later versions
    184      * this method is not called by the Android TTS framework.
    185      *
    186      * Can be called on multiple threads.
    187      *
    188      * @return A 3-element array, containing language (ISO 3-letter code),
    189      *         country (ISO 3-letter code) and variant used by the engine.
    190      *         The country and variant may be {@code ""}. If country is empty, then variant must
    191      *         be empty too.
    192      * @see Locale#getISO3Language()
    193      * @see Locale#getISO3Country()
    194      * @see Locale#getVariant()
    195      */
    196     protected abstract String[] onGetLanguage();
    197 
    198     /**
    199      * Notifies the engine that it should load a speech synthesis language. There is no guarantee
    200      * that this method is always called before the language is used for synthesis. It is merely
    201      * a hint to the engine that it will probably get some synthesis requests for this language
    202      * at some point in the future.
    203      *
    204      * Can be called on multiple threads.
    205      * In <= Android 4.2 (<= API 17) can be called on main and service binder threads.
    206      * In > Android 4.2 (> API 17) can be called on main and synthesis threads.
    207      *
    208      * @param lang ISO-3 language code.
    209      * @param country ISO-3 country code. May be empty or null.
    210      * @param variant Language variant. May be empty or null.
    211      * @return Code indicating the support status for the locale.
    212      *         One of {@link TextToSpeech#LANG_AVAILABLE},
    213      *         {@link TextToSpeech#LANG_COUNTRY_AVAILABLE},
    214      *         {@link TextToSpeech#LANG_COUNTRY_VAR_AVAILABLE},
    215      *         {@link TextToSpeech#LANG_MISSING_DATA}
    216      *         {@link TextToSpeech#LANG_NOT_SUPPORTED}.
    217      */
    218     protected abstract int onLoadLanguage(String lang, String country, String variant);
    219 
    220     /**
    221      * Notifies the service that it should stop any in-progress speech synthesis.
    222      * This method can be called even if no speech synthesis is currently in progress.
    223      *
    224      * Can be called on multiple threads, but not on the synthesis thread.
    225      */
    226     protected abstract void onStop();
    227 
    228     /**
    229      * Tells the service to synthesize speech from the given text. This method should block until
    230      * the synthesis is finished. Called on the synthesis thread.
    231      *
    232      * @param request The synthesis request.
    233      * @param callback The callback that the engine must use to make data available for playback or
    234      *     for writing to a file.
    235      */
    236     protected abstract void onSynthesizeText(SynthesisRequest request, SynthesisCallback callback);
    237 
    238     /**
    239      * Queries the service for a set of features supported for a given language.
    240      *
    241      * Can be called on multiple threads.
    242      *
    243      * @param lang ISO-3 language code.
    244      * @param country ISO-3 country code. May be empty or null.
    245      * @param variant Language variant. May be empty or null.
    246      * @return A list of features supported for the given language.
    247      */
    248     protected Set<String> onGetFeaturesForLanguage(String lang, String country, String variant) {
    249         return new HashSet<String>();
    250     }
    251 
    252     private int getExpectedLanguageAvailableStatus(Locale locale) {
    253         int expectedStatus = TextToSpeech.LANG_COUNTRY_VAR_AVAILABLE;
    254         if (locale.getVariant().isEmpty()) {
    255             if (locale.getCountry().isEmpty()) {
    256                 expectedStatus = TextToSpeech.LANG_AVAILABLE;
    257             } else {
    258                 expectedStatus = TextToSpeech.LANG_COUNTRY_AVAILABLE;
    259             }
    260         }
    261         return expectedStatus;
    262     }
    263 
    264     /**
    265      * Queries the service for a set of supported voices.
    266      *
    267      * Can be called on multiple threads.
    268      *
    269      * The default implementation tries to enumerate all available locales, pass them to
    270      * {@link #onIsLanguageAvailable(String, String, String)} and create Voice instances (using
    271      * the locale's BCP-47 language tag as the voice name) for the ones that are supported.
    272      * Note, that this implementation is suitable only for engines that don't have multiple voices
    273      * for a single locale. Also, this implementation won't work with Locales not listed in the
    274      * set returned by the {@link Locale#getAvailableLocales()} method.
    275      *
    276      * @return A list of voices supported.
    277      */
    278     public List<Voice> onGetVoices() {
    279         // Enumerate all locales and check if they are available
    280         ArrayList<Voice> voices = new ArrayList<Voice>();
    281         for (Locale locale : Locale.getAvailableLocales()) {
    282             int expectedStatus = getExpectedLanguageAvailableStatus(locale);
    283             try {
    284                 int localeStatus = onIsLanguageAvailable(locale.getISO3Language(),
    285                         locale.getISO3Country(), locale.getVariant());
    286                 if (localeStatus != expectedStatus) {
    287                     continue;
    288                 }
    289             } catch (MissingResourceException e) {
    290                 // Ignore locale without iso 3 codes
    291                 continue;
    292             }
    293             Set<String> features = onGetFeaturesForLanguage(locale.getISO3Language(),
    294                     locale.getISO3Country(), locale.getVariant());
    295             String voiceName = onGetDefaultVoiceNameFor(locale.getISO3Language(),
    296                     locale.getISO3Country(), locale.getVariant());
    297             voices.add(new Voice(voiceName, locale, Voice.QUALITY_NORMAL,
    298                     Voice.LATENCY_NORMAL, false, features));
    299         }
    300         return voices;
    301     }
    302 
    303     /**
    304      * Return a name of the default voice for a given locale.
    305      *
    306      * This method provides a mapping between locales and available voices. This method is
    307      * used in {@link TextToSpeech#setLanguage}, which calls this method and then calls
    308      * {@link TextToSpeech#setVoice} with the voice returned by this method.
    309      *
    310      * Also, it's used by {@link TextToSpeech#getDefaultVoice()} to find a default voice for
    311      * the default locale.
    312      *
    313      * @param lang ISO-3 language code.
    314      * @param country ISO-3 country code. May be empty or null.
    315      * @param variant Language variant. May be empty or null.
    316 
    317      * @return A name of the default voice for a given locale.
    318      */
    319     public String onGetDefaultVoiceNameFor(String lang, String country, String variant) {
    320         int localeStatus = onIsLanguageAvailable(lang, country, variant);
    321         Locale iso3Locale = null;
    322         switch (localeStatus) {
    323             case TextToSpeech.LANG_AVAILABLE:
    324                 iso3Locale = new Locale(lang);
    325                 break;
    326             case TextToSpeech.LANG_COUNTRY_AVAILABLE:
    327                 iso3Locale = new Locale(lang, country);
    328                 break;
    329             case TextToSpeech.LANG_COUNTRY_VAR_AVAILABLE:
    330                 iso3Locale = new Locale(lang, country, variant);
    331                 break;
    332             default:
    333                 return null;
    334         }
    335         Locale properLocale = TtsEngines.normalizeTTSLocale(iso3Locale);
    336         String voiceName = properLocale.toLanguageTag();
    337         if (onIsValidVoiceName(voiceName) == TextToSpeech.SUCCESS) {
    338             return voiceName;
    339         } else {
    340             return null;
    341         }
    342     }
    343 
    344     /**
    345      * Notifies the engine that it should load a speech synthesis voice. There is no guarantee
    346      * that this method is always called before the voice is used for synthesis. It is merely
    347      * a hint to the engine that it will probably get some synthesis requests for this voice
    348      * at some point in the future.
    349      *
    350      * Will be called only on synthesis thread.
    351      *
    352      * The default implementation creates a Locale from the voice name (by interpreting the name as
    353      * a BCP-47 tag for the locale), and passes it to
    354      * {@link #onLoadLanguage(String, String, String)}.
    355      *
    356      * @param voiceName Name of the voice.
    357      * @return {@link TextToSpeech#ERROR} or {@link TextToSpeech#SUCCESS}.
    358      */
    359     public int onLoadVoice(String voiceName) {
    360         Locale locale = Locale.forLanguageTag(voiceName);
    361         if (locale == null) {
    362             return TextToSpeech.ERROR;
    363         }
    364         int expectedStatus = getExpectedLanguageAvailableStatus(locale);
    365         try {
    366             int localeStatus = onIsLanguageAvailable(locale.getISO3Language(),
    367                     locale.getISO3Country(), locale.getVariant());
    368             if (localeStatus != expectedStatus) {
    369                 return TextToSpeech.ERROR;
    370             }
    371             onLoadLanguage(locale.getISO3Language(),
    372                     locale.getISO3Country(), locale.getVariant());
    373             return TextToSpeech.SUCCESS;
    374         } catch (MissingResourceException e) {
    375             return TextToSpeech.ERROR;
    376         }
    377     }
    378 
    379     /**
    380      * Checks whether the engine supports a voice with a given name.
    381      *
    382      * Can be called on multiple threads.
    383      *
    384      * The default implementation treats the voice name as a language tag, creating a Locale from
    385      * the voice name, and passes it to {@link #onIsLanguageAvailable(String, String, String)}.
    386      *
    387      * @param voiceName Name of the voice.
    388      * @return {@link TextToSpeech#ERROR} or {@link TextToSpeech#SUCCESS}.
    389      */
    390     public int onIsValidVoiceName(String voiceName) {
    391         Locale locale = Locale.forLanguageTag(voiceName);
    392         if (locale == null) {
    393             return TextToSpeech.ERROR;
    394         }
    395         int expectedStatus = getExpectedLanguageAvailableStatus(locale);
    396         try {
    397             int localeStatus = onIsLanguageAvailable(locale.getISO3Language(),
    398                     locale.getISO3Country(), locale.getVariant());
    399             if (localeStatus != expectedStatus) {
    400                 return TextToSpeech.ERROR;
    401             }
    402             return TextToSpeech.SUCCESS;
    403         } catch (MissingResourceException e) {
    404             return TextToSpeech.ERROR;
    405         }
    406     }
    407 
    408     private int getDefaultSpeechRate() {
    409         return getSecureSettingInt(Settings.Secure.TTS_DEFAULT_RATE, Engine.DEFAULT_RATE);
    410     }
    411 
    412     private int getDefaultPitch() {
    413         return getSecureSettingInt(Settings.Secure.TTS_DEFAULT_PITCH, Engine.DEFAULT_PITCH);
    414     }
    415 
    416     private String[] getSettingsLocale() {
    417         final Locale locale = mEngineHelper.getLocalePrefForEngine(mPackageName);
    418         return TtsEngines.toOldLocaleStringFormat(locale);
    419     }
    420 
    421     private int getSecureSettingInt(String name, int defaultValue) {
    422         return Settings.Secure.getInt(getContentResolver(), name, defaultValue);
    423     }
    424 
    425     /**
    426      * Synthesizer thread. This thread is used to run {@link SynthHandler}.
    427      */
    428     private class SynthThread extends HandlerThread implements MessageQueue.IdleHandler {
    429 
    430         private boolean mFirstIdle = true;
    431 
    432         public SynthThread() {
    433             super(SYNTH_THREAD_NAME, android.os.Process.THREAD_PRIORITY_DEFAULT);
    434         }
    435 
    436         @Override
    437         protected void onLooperPrepared() {
    438             getLooper().getQueue().addIdleHandler(this);
    439         }
    440 
    441         @Override
    442         public boolean queueIdle() {
    443             if (mFirstIdle) {
    444                 mFirstIdle = false;
    445             } else {
    446                 broadcastTtsQueueProcessingCompleted();
    447             }
    448             return true;
    449         }
    450 
    451         private void broadcastTtsQueueProcessingCompleted() {
    452             Intent i = new Intent(TextToSpeech.ACTION_TTS_QUEUE_PROCESSING_COMPLETED);
    453             if (DBG) Log.d(TAG, "Broadcasting: " + i);
    454             sendBroadcast(i);
    455         }
    456     }
    457 
    458     private class SynthHandler extends Handler {
    459         private SpeechItem mCurrentSpeechItem = null;
    460 
    461         // When a message with QUEUE_FLUSH arrives we add the caller identity to the List and when a
    462         // message with QUEUE_DESTROY arrives we increment mFlushAll. Then a message is added to the
    463         // handler queue that removes the caller identify from the list and decrements the mFlushAll
    464         // counter. This is so that when a message is processed and the caller identity is in the
    465         // list or mFlushAll is not zero, we know that the message should be flushed.
    466         // It's important that mFlushedObjects is a List and not a Set, and that mFlushAll is an
    467         // int and not a bool. This is because when multiple messages arrive with QUEUE_FLUSH or
    468         // QUEUE_DESTROY, we want to keep flushing messages until we arrive at the last QUEUE_FLUSH
    469         // or QUEUE_DESTROY message.
    470         private List<Object> mFlushedObjects = new ArrayList<>();
    471         private int mFlushAll = 0;
    472 
    473         public SynthHandler(Looper looper) {
    474             super(looper);
    475         }
    476 
    477         private void startFlushingSpeechItems(Object callerIdentity) {
    478             synchronized (mFlushedObjects) {
    479                 if (callerIdentity == null) {
    480                     mFlushAll += 1;
    481                 } else {
    482                     mFlushedObjects.add(callerIdentity);
    483                 }
    484             }
    485         }
    486         private void endFlushingSpeechItems(Object callerIdentity) {
    487             synchronized (mFlushedObjects) {
    488                 if (callerIdentity == null) {
    489                     mFlushAll -= 1;
    490                 } else {
    491                     mFlushedObjects.remove(callerIdentity);
    492                 }
    493             }
    494         }
    495         private boolean isFlushed(SpeechItem speechItem) {
    496             synchronized (mFlushedObjects) {
    497                 return mFlushAll > 0 || mFlushedObjects.contains(speechItem.getCallerIdentity());
    498             }
    499         }
    500 
    501         private synchronized SpeechItem getCurrentSpeechItem() {
    502             return mCurrentSpeechItem;
    503         }
    504 
    505         private synchronized SpeechItem setCurrentSpeechItem(SpeechItem speechItem) {
    506             SpeechItem old = mCurrentSpeechItem;
    507             mCurrentSpeechItem = speechItem;
    508             return old;
    509         }
    510 
    511         private synchronized SpeechItem maybeRemoveCurrentSpeechItem(Object callerIdentity) {
    512             if (mCurrentSpeechItem != null &&
    513                     (mCurrentSpeechItem.getCallerIdentity() == callerIdentity)) {
    514                 SpeechItem current = mCurrentSpeechItem;
    515                 mCurrentSpeechItem = null;
    516                 return current;
    517             }
    518 
    519             return null;
    520         }
    521 
    522         public boolean isSpeaking() {
    523             return getCurrentSpeechItem() != null;
    524         }
    525 
    526         public void quit() {
    527             // Don't process any more speech items
    528             getLooper().quit();
    529             // Stop the current speech item
    530             SpeechItem current = setCurrentSpeechItem(null);
    531             if (current != null) {
    532                 current.stop();
    533             }
    534             // The AudioPlaybackHandler will be destroyed by the caller.
    535         }
    536 
    537         /**
    538          * Adds a speech item to the queue.
    539          *
    540          * Called on a service binder thread.
    541          */
    542         public int enqueueSpeechItem(int queueMode, final SpeechItem speechItem) {
    543             UtteranceProgressDispatcher utterenceProgress = null;
    544             if (speechItem instanceof UtteranceProgressDispatcher) {
    545                 utterenceProgress = (UtteranceProgressDispatcher) speechItem;
    546             }
    547 
    548             if (!speechItem.isValid()) {
    549                 if (utterenceProgress != null) {
    550                     utterenceProgress.dispatchOnError(
    551                             TextToSpeech.ERROR_INVALID_REQUEST);
    552                 }
    553                 return TextToSpeech.ERROR;
    554             }
    555 
    556             if (queueMode == TextToSpeech.QUEUE_FLUSH) {
    557                 stopForApp(speechItem.getCallerIdentity());
    558             } else if (queueMode == TextToSpeech.QUEUE_DESTROY) {
    559                 stopAll();
    560             }
    561             Runnable runnable = new Runnable() {
    562                 @Override
    563                 public void run() {
    564                     if (isFlushed(speechItem)) {
    565                         speechItem.stop();
    566                     } else {
    567                         setCurrentSpeechItem(speechItem);
    568                         speechItem.play();
    569                         setCurrentSpeechItem(null);
    570                     }
    571                 }
    572             };
    573             Message msg = Message.obtain(this, runnable);
    574 
    575             // The obj is used to remove all callbacks from the given app in
    576             // stopForApp(String).
    577             //
    578             // Note that this string is interned, so the == comparison works.
    579             msg.obj = speechItem.getCallerIdentity();
    580 
    581             if (sendMessage(msg)) {
    582                 return TextToSpeech.SUCCESS;
    583             } else {
    584                 Log.w(TAG, "SynthThread has quit");
    585                 if (utterenceProgress != null) {
    586                     utterenceProgress.dispatchOnError(TextToSpeech.ERROR_SERVICE);
    587                 }
    588                 return TextToSpeech.ERROR;
    589             }
    590         }
    591 
    592         /**
    593          * Stops all speech output and removes any utterances still in the queue for
    594          * the calling app.
    595          *
    596          * Called on a service binder thread.
    597          */
    598         public int stopForApp(final Object callerIdentity) {
    599             if (callerIdentity == null) {
    600                 return TextToSpeech.ERROR;
    601             }
    602 
    603             // Flush pending messages from callerIdentity
    604             startFlushingSpeechItems(callerIdentity);
    605 
    606             // This stops writing data to the file / or publishing
    607             // items to the audio playback handler.
    608             //
    609             // Note that the current speech item must be removed only if it
    610             // belongs to the callingApp, else the item will be "orphaned" and
    611             // not stopped correctly if a stop request comes along for the item
    612             // from the app it belongs to.
    613             SpeechItem current = maybeRemoveCurrentSpeechItem(callerIdentity);
    614             if (current != null) {
    615                 current.stop();
    616             }
    617 
    618             // Remove any enqueued audio too.
    619             mAudioPlaybackHandler.stopForApp(callerIdentity);
    620 
    621             // Stop flushing pending messages
    622             Runnable runnable = new Runnable() {
    623                 @Override
    624                 public void run() {
    625                     endFlushingSpeechItems(callerIdentity);
    626                 }
    627             };
    628             sendMessage(Message.obtain(this, runnable));
    629             return TextToSpeech.SUCCESS;
    630         }
    631 
    632         public int stopAll() {
    633             // Order to flush pending messages
    634             startFlushingSpeechItems(null);
    635 
    636             // Stop the current speech item unconditionally .
    637             SpeechItem current = setCurrentSpeechItem(null);
    638             if (current != null) {
    639                 current.stop();
    640             }
    641             // Remove all pending playback as well.
    642             mAudioPlaybackHandler.stop();
    643 
    644             // Message to stop flushing pending messages
    645             Runnable runnable = new Runnable() {
    646                 @Override
    647                 public void run() {
    648                     endFlushingSpeechItems(null);
    649                 }
    650             };
    651             sendMessage(Message.obtain(this, runnable));
    652 
    653 
    654             return TextToSpeech.SUCCESS;
    655         }
    656     }
    657 
    658     interface UtteranceProgressDispatcher {
    659         void dispatchOnStop();
    660 
    661         void dispatchOnSuccess();
    662 
    663         void dispatchOnStart();
    664 
    665         void dispatchOnError(int errorCode);
    666 
    667         void dispatchOnBeginSynthesis(int sampleRateInHz, int audioFormat, int channelCount);
    668 
    669         void dispatchOnAudioAvailable(byte[] audio);
    670 
    671         public void dispatchOnRangeStart(int start, int end, int frame);
    672     }
    673 
    674     /** Set of parameters affecting audio output. */
    675     static class AudioOutputParams {
    676         /**
    677          * Audio session identifier. May be used to associate audio playback with one of the
    678          * {@link android.media.audiofx.AudioEffect} objects. If not specified by client,
    679          * it should be equal to {@link AudioManager#AUDIO_SESSION_ID_GENERATE}.
    680          */
    681         public final int mSessionId;
    682 
    683         /**
    684          * Volume, in the range [0.0f, 1.0f]. The default value is
    685          * {@link TextToSpeech.Engine#DEFAULT_VOLUME} (1.0f).
    686          */
    687         public final float mVolume;
    688 
    689         /**
    690          * Left/right position of the audio, in the range [-1.0f, 1.0f].
    691          * The default value is {@link TextToSpeech.Engine#DEFAULT_PAN} (0.0f).
    692          */
    693         public final float mPan;
    694 
    695 
    696         /**
    697          * Audio attributes, set by {@link TextToSpeech#setAudioAttributes}
    698          * or created from the value of {@link TextToSpeech.Engine#KEY_PARAM_STREAM}.
    699          */
    700         public final AudioAttributes mAudioAttributes;
    701 
    702         /** Create AudioOutputParams with default values */
    703         AudioOutputParams() {
    704             mSessionId = AudioManager.AUDIO_SESSION_ID_GENERATE;
    705             mVolume = Engine.DEFAULT_VOLUME;
    706             mPan = Engine.DEFAULT_PAN;
    707             mAudioAttributes = null;
    708         }
    709 
    710         AudioOutputParams(int sessionId, float volume, float pan,
    711                 AudioAttributes audioAttributes) {
    712             mSessionId = sessionId;
    713             mVolume = volume;
    714             mPan = pan;
    715             mAudioAttributes = audioAttributes;
    716         }
    717 
    718         /** Create AudioOutputParams from A {@link SynthesisRequest#getParams()} bundle */
    719         static AudioOutputParams createFromParamsBundle(Bundle paramsBundle, boolean isSpeech) {
    720             if (paramsBundle == null) {
    721                 return new AudioOutputParams();
    722             }
    723 
    724             AudioAttributes audioAttributes =
    725                     (AudioAttributes) paramsBundle.getParcelable(
    726                             Engine.KEY_PARAM_AUDIO_ATTRIBUTES);
    727             if (audioAttributes == null) {
    728                 int streamType = paramsBundle.getInt(
    729                         Engine.KEY_PARAM_STREAM, Engine.DEFAULT_STREAM);
    730                 audioAttributes = (new AudioAttributes.Builder())
    731                         .setLegacyStreamType(streamType)
    732                         .setContentType((isSpeech ?
    733                                 AudioAttributes.CONTENT_TYPE_SPEECH :
    734                                 AudioAttributes.CONTENT_TYPE_SONIFICATION))
    735                         .build();
    736             }
    737 
    738             return new AudioOutputParams(
    739                     paramsBundle.getInt(
    740                             Engine.KEY_PARAM_SESSION_ID,
    741                             AudioManager.AUDIO_SESSION_ID_GENERATE),
    742                     paramsBundle.getFloat(
    743                             Engine.KEY_PARAM_VOLUME,
    744                             Engine.DEFAULT_VOLUME),
    745                     paramsBundle.getFloat(
    746                             Engine.KEY_PARAM_PAN,
    747                             Engine.DEFAULT_PAN),
    748                     audioAttributes);
    749         }
    750     }
    751 
    752 
    753     /**
    754      * An item in the synth thread queue.
    755      */
    756     private abstract class SpeechItem {
    757         private final Object mCallerIdentity;
    758         private final int mCallerUid;
    759         private final int mCallerPid;
    760         private boolean mStarted = false;
    761         private boolean mStopped = false;
    762 
    763         public SpeechItem(Object caller, int callerUid, int callerPid) {
    764             mCallerIdentity = caller;
    765             mCallerUid = callerUid;
    766             mCallerPid = callerPid;
    767         }
    768 
    769         public Object getCallerIdentity() {
    770             return mCallerIdentity;
    771         }
    772 
    773         public int getCallerUid() {
    774             return mCallerUid;
    775         }
    776 
    777         public int getCallerPid() {
    778             return mCallerPid;
    779         }
    780 
    781         /**
    782          * Checker whether the item is valid. If this method returns false, the item should not
    783          * be played.
    784          */
    785         public abstract boolean isValid();
    786 
    787         /**
    788          * Plays the speech item. Blocks until playback is finished.
    789          * Must not be called more than once.
    790          *
    791          * Only called on the synthesis thread.
    792          */
    793         public void play() {
    794             synchronized (this) {
    795                 if (mStarted) {
    796                     throw new IllegalStateException("play() called twice");
    797                 }
    798                 mStarted = true;
    799             }
    800             playImpl();
    801         }
    802 
    803         protected abstract void playImpl();
    804 
    805         /**
    806          * Stops the speech item.
    807          * Must not be called more than once.
    808          *
    809          * Can be called on multiple threads,  but not on the synthesis thread.
    810          */
    811         public void stop() {
    812             synchronized (this) {
    813                 if (mStopped) {
    814                     throw new IllegalStateException("stop() called twice");
    815                 }
    816                 mStopped = true;
    817             }
    818             stopImpl();
    819         }
    820 
    821         protected abstract void stopImpl();
    822 
    823         protected synchronized boolean isStopped() {
    824              return mStopped;
    825         }
    826 
    827         protected synchronized boolean isStarted() {
    828             return mStarted;
    829        }
    830     }
    831 
    832     /**
    833      * An item in the synth thread queue that process utterance (and call back to client about
    834      * progress).
    835      */
    836     private abstract class UtteranceSpeechItem extends SpeechItem
    837         implements UtteranceProgressDispatcher  {
    838 
    839         public UtteranceSpeechItem(Object caller, int callerUid, int callerPid) {
    840             super(caller, callerUid, callerPid);
    841         }
    842 
    843         @Override
    844         public void dispatchOnSuccess() {
    845             final String utteranceId = getUtteranceId();
    846             if (utteranceId != null) {
    847                 mCallbacks.dispatchOnSuccess(getCallerIdentity(), utteranceId);
    848             }
    849         }
    850 
    851         @Override
    852         public void dispatchOnStop() {
    853             final String utteranceId = getUtteranceId();
    854             if (utteranceId != null) {
    855                 mCallbacks.dispatchOnStop(getCallerIdentity(), utteranceId, isStarted());
    856             }
    857         }
    858 
    859         @Override
    860         public void dispatchOnStart() {
    861             final String utteranceId = getUtteranceId();
    862             if (utteranceId != null) {
    863                 mCallbacks.dispatchOnStart(getCallerIdentity(), utteranceId);
    864             }
    865         }
    866 
    867         @Override
    868         public void dispatchOnError(int errorCode) {
    869             final String utteranceId = getUtteranceId();
    870             if (utteranceId != null) {
    871                 mCallbacks.dispatchOnError(getCallerIdentity(), utteranceId, errorCode);
    872             }
    873         }
    874 
    875         @Override
    876         public void dispatchOnBeginSynthesis(int sampleRateInHz, int audioFormat, int channelCount) {
    877             final String utteranceId = getUtteranceId();
    878             if (utteranceId != null) {
    879                 mCallbacks.dispatchOnBeginSynthesis(getCallerIdentity(), utteranceId, sampleRateInHz, audioFormat, channelCount);
    880             }
    881         }
    882 
    883         @Override
    884         public void dispatchOnAudioAvailable(byte[] audio) {
    885             final String utteranceId = getUtteranceId();
    886             if (utteranceId != null) {
    887                 mCallbacks.dispatchOnAudioAvailable(getCallerIdentity(), utteranceId, audio);
    888             }
    889         }
    890 
    891         @Override
    892         public void dispatchOnRangeStart(int start, int end, int frame) {
    893             final String utteranceId = getUtteranceId();
    894             if (utteranceId != null) {
    895                 mCallbacks.dispatchOnRangeStart(
    896                         getCallerIdentity(), utteranceId, start, end, frame);
    897             }
    898         }
    899 
    900         abstract public String getUtteranceId();
    901 
    902         String getStringParam(Bundle params, String key, String defaultValue) {
    903             return params == null ? defaultValue : params.getString(key, defaultValue);
    904         }
    905 
    906         int getIntParam(Bundle params, String key, int defaultValue) {
    907             return params == null ? defaultValue : params.getInt(key, defaultValue);
    908         }
    909 
    910         float getFloatParam(Bundle params, String key, float defaultValue) {
    911             return params == null ? defaultValue : params.getFloat(key, defaultValue);
    912         }
    913     }
    914 
    915     /**
    916      * Synthesis parameters are kept in a single Bundle passed as parameter. This class allow
    917      * subclasses to access them conveniently.
    918      */
    919     private abstract class UtteranceSpeechItemWithParams extends UtteranceSpeechItem {
    920         protected final Bundle mParams;
    921         protected final String mUtteranceId;
    922 
    923         UtteranceSpeechItemWithParams(
    924                 Object callerIdentity,
    925                 int callerUid,
    926                 int callerPid,
    927                 Bundle params,
    928                 String utteranceId) {
    929             super(callerIdentity, callerUid, callerPid);
    930             mParams = params;
    931             mUtteranceId = utteranceId;
    932         }
    933 
    934         boolean hasLanguage() {
    935             return !TextUtils.isEmpty(getStringParam(mParams, Engine.KEY_PARAM_LANGUAGE, null));
    936         }
    937 
    938         int getSpeechRate() {
    939             return getIntParam(mParams, Engine.KEY_PARAM_RATE, getDefaultSpeechRate());
    940         }
    941 
    942         int getPitch() {
    943             return getIntParam(mParams, Engine.KEY_PARAM_PITCH, getDefaultPitch());
    944         }
    945 
    946         @Override
    947         public String getUtteranceId() {
    948             return mUtteranceId;
    949         }
    950 
    951         AudioOutputParams getAudioParams() {
    952             return AudioOutputParams.createFromParamsBundle(mParams, true);
    953         }
    954     }
    955 
    956     class SynthesisSpeechItem extends UtteranceSpeechItemWithParams {
    957         // Never null.
    958         private final CharSequence mText;
    959         private final SynthesisRequest mSynthesisRequest;
    960         private final String[] mDefaultLocale;
    961         // Non null after synthesis has started, and all accesses
    962         // guarded by 'this'.
    963         private AbstractSynthesisCallback mSynthesisCallback;
    964         private final EventLogger mEventLogger;
    965         private final int mCallerUid;
    966 
    967         public SynthesisSpeechItem(
    968                 Object callerIdentity,
    969                 int callerUid,
    970                 int callerPid,
    971                 Bundle params,
    972                 String utteranceId,
    973                 CharSequence text) {
    974             super(callerIdentity, callerUid, callerPid, params, utteranceId);
    975             mText = text;
    976             mCallerUid = callerUid;
    977             mSynthesisRequest = new SynthesisRequest(mText, mParams);
    978             mDefaultLocale = getSettingsLocale();
    979             setRequestParams(mSynthesisRequest);
    980             mEventLogger = new EventLogger(mSynthesisRequest, callerUid, callerPid, mPackageName);
    981         }
    982 
    983         public CharSequence getText() {
    984             return mText;
    985         }
    986 
    987         @Override
    988         public boolean isValid() {
    989             if (mText == null) {
    990                 Log.e(TAG, "null synthesis text");
    991                 return false;
    992             }
    993             if (mText.length() >= TextToSpeech.getMaxSpeechInputLength()) {
    994                 Log.w(TAG, "Text too long: " + mText.length() + " chars");
    995                 return false;
    996             }
    997             return true;
    998         }
    999 
   1000         @Override
   1001         protected void playImpl() {
   1002             AbstractSynthesisCallback synthesisCallback;
   1003             mEventLogger.onRequestProcessingStart();
   1004             synchronized (this) {
   1005                 // stop() might have been called before we enter this
   1006                 // synchronized block.
   1007                 if (isStopped()) {
   1008                     return;
   1009                 }
   1010                 mSynthesisCallback = createSynthesisCallback();
   1011                 synthesisCallback = mSynthesisCallback;
   1012             }
   1013 
   1014             TextToSpeechService.this.onSynthesizeText(mSynthesisRequest, synthesisCallback);
   1015 
   1016             // Fix for case where client called .start() & .error(), but did not called .done()
   1017             if (synthesisCallback.hasStarted() && !synthesisCallback.hasFinished()) {
   1018                 synthesisCallback.done();
   1019             }
   1020         }
   1021 
   1022         protected AbstractSynthesisCallback createSynthesisCallback() {
   1023             return new PlaybackSynthesisCallback(getAudioParams(),
   1024                     mAudioPlaybackHandler, this, getCallerIdentity(), mEventLogger, false);
   1025         }
   1026 
   1027         private void setRequestParams(SynthesisRequest request) {
   1028             String voiceName = getVoiceName();
   1029             request.setLanguage(getLanguage(), getCountry(), getVariant());
   1030             if (!TextUtils.isEmpty(voiceName)) {
   1031                 request.setVoiceName(getVoiceName());
   1032             }
   1033             request.setSpeechRate(getSpeechRate());
   1034             request.setCallerUid(mCallerUid);
   1035             request.setPitch(getPitch());
   1036         }
   1037 
   1038         @Override
   1039         protected void stopImpl() {
   1040             AbstractSynthesisCallback synthesisCallback;
   1041             synchronized (this) {
   1042                 synthesisCallback = mSynthesisCallback;
   1043             }
   1044             if (synthesisCallback != null) {
   1045                 // If the synthesis callback is null, it implies that we haven't
   1046                 // entered the synchronized(this) block in playImpl which in
   1047                 // turn implies that synthesis would not have started.
   1048                 synthesisCallback.stop();
   1049                 TextToSpeechService.this.onStop();
   1050             } else {
   1051                 dispatchOnStop();
   1052             }
   1053         }
   1054 
   1055         private String getCountry() {
   1056             if (!hasLanguage()) return mDefaultLocale[1];
   1057             return getStringParam(mParams, Engine.KEY_PARAM_COUNTRY, "");
   1058         }
   1059 
   1060         private String getVariant() {
   1061             if (!hasLanguage()) return mDefaultLocale[2];
   1062             return getStringParam(mParams, Engine.KEY_PARAM_VARIANT, "");
   1063         }
   1064 
   1065         public String getLanguage() {
   1066             return getStringParam(mParams, Engine.KEY_PARAM_LANGUAGE, mDefaultLocale[0]);
   1067         }
   1068 
   1069         public String getVoiceName() {
   1070             return getStringParam(mParams, Engine.KEY_PARAM_VOICE_NAME, "");
   1071         }
   1072     }
   1073 
   1074     private class SynthesisToFileOutputStreamSpeechItem extends SynthesisSpeechItem {
   1075         private final FileOutputStream mFileOutputStream;
   1076 
   1077         public SynthesisToFileOutputStreamSpeechItem(
   1078                 Object callerIdentity,
   1079                 int callerUid,
   1080                 int callerPid,
   1081                 Bundle params,
   1082                 String utteranceId,
   1083                 CharSequence text,
   1084                 FileOutputStream fileOutputStream) {
   1085             super(callerIdentity, callerUid, callerPid, params, utteranceId, text);
   1086             mFileOutputStream = fileOutputStream;
   1087         }
   1088 
   1089         @Override
   1090         protected AbstractSynthesisCallback createSynthesisCallback() {
   1091             return new FileSynthesisCallback(mFileOutputStream.getChannel(), this, false);
   1092         }
   1093 
   1094         @Override
   1095         protected void playImpl() {
   1096             dispatchOnStart();
   1097             super.playImpl();
   1098             try {
   1099               mFileOutputStream.close();
   1100             } catch(IOException e) {
   1101               Log.w(TAG, "Failed to close output file", e);
   1102             }
   1103         }
   1104     }
   1105 
   1106     private class AudioSpeechItem extends UtteranceSpeechItemWithParams {
   1107         private final AudioPlaybackQueueItem mItem;
   1108 
   1109         public AudioSpeechItem(
   1110                 Object callerIdentity,
   1111                 int callerUid,
   1112                 int callerPid,
   1113                 Bundle params,
   1114                 String utteranceId,
   1115                 Uri uri) {
   1116             super(callerIdentity, callerUid, callerPid, params, utteranceId);
   1117             mItem = new AudioPlaybackQueueItem(this, getCallerIdentity(),
   1118                     TextToSpeechService.this, uri, getAudioParams());
   1119         }
   1120 
   1121         @Override
   1122         public boolean isValid() {
   1123             return true;
   1124         }
   1125 
   1126         @Override
   1127         protected void playImpl() {
   1128             mAudioPlaybackHandler.enqueue(mItem);
   1129         }
   1130 
   1131         @Override
   1132         protected void stopImpl() {
   1133             // Do nothing.
   1134         }
   1135 
   1136         @Override
   1137         public String getUtteranceId() {
   1138             return getStringParam(mParams, Engine.KEY_PARAM_UTTERANCE_ID, null);
   1139         }
   1140 
   1141         @Override
   1142         AudioOutputParams getAudioParams() {
   1143             return AudioOutputParams.createFromParamsBundle(mParams, false);
   1144         }
   1145     }
   1146 
   1147     private class SilenceSpeechItem extends UtteranceSpeechItem {
   1148         private final long mDuration;
   1149         private final String mUtteranceId;
   1150 
   1151         public SilenceSpeechItem(Object callerIdentity, int callerUid, int callerPid,
   1152                 String utteranceId, long duration) {
   1153             super(callerIdentity, callerUid, callerPid);
   1154             mUtteranceId = utteranceId;
   1155             mDuration = duration;
   1156         }
   1157 
   1158         @Override
   1159         public boolean isValid() {
   1160             return true;
   1161         }
   1162 
   1163         @Override
   1164         protected void playImpl() {
   1165             mAudioPlaybackHandler.enqueue(new SilencePlaybackQueueItem(
   1166                     this, getCallerIdentity(), mDuration));
   1167         }
   1168 
   1169         @Override
   1170         protected void stopImpl() {
   1171 
   1172         }
   1173 
   1174         @Override
   1175         public String getUtteranceId() {
   1176             return mUtteranceId;
   1177         }
   1178     }
   1179 
   1180     /**
   1181      * Call {@link TextToSpeechService#onLoadLanguage} on synth thread.
   1182      */
   1183     private class LoadLanguageItem extends SpeechItem {
   1184         private final String mLanguage;
   1185         private final String mCountry;
   1186         private final String mVariant;
   1187 
   1188         public LoadLanguageItem(Object callerIdentity, int callerUid, int callerPid,
   1189                 String language, String country, String variant) {
   1190             super(callerIdentity, callerUid, callerPid);
   1191             mLanguage = language;
   1192             mCountry = country;
   1193             mVariant = variant;
   1194         }
   1195 
   1196         @Override
   1197         public boolean isValid() {
   1198             return true;
   1199         }
   1200 
   1201         @Override
   1202         protected void playImpl() {
   1203             TextToSpeechService.this.onLoadLanguage(mLanguage, mCountry, mVariant);
   1204         }
   1205 
   1206         @Override
   1207         protected void stopImpl() {
   1208             // No-op
   1209         }
   1210     }
   1211 
   1212     /**
   1213      * Call {@link TextToSpeechService#onLoadLanguage} on synth thread.
   1214      */
   1215     private class LoadVoiceItem extends SpeechItem {
   1216         private final String mVoiceName;
   1217 
   1218         public LoadVoiceItem(Object callerIdentity, int callerUid, int callerPid,
   1219                 String voiceName) {
   1220             super(callerIdentity, callerUid, callerPid);
   1221             mVoiceName = voiceName;
   1222         }
   1223 
   1224         @Override
   1225         public boolean isValid() {
   1226             return true;
   1227         }
   1228 
   1229         @Override
   1230         protected void playImpl() {
   1231             TextToSpeechService.this.onLoadVoice(mVoiceName);
   1232         }
   1233 
   1234         @Override
   1235         protected void stopImpl() {
   1236             // No-op
   1237         }
   1238     }
   1239 
   1240 
   1241     @Override
   1242     public IBinder onBind(Intent intent) {
   1243         if (TextToSpeech.Engine.INTENT_ACTION_TTS_SERVICE.equals(intent.getAction())) {
   1244             return mBinder;
   1245         }
   1246         return null;
   1247     }
   1248 
   1249     /**
   1250      * Binder returned from {@code #onBind(Intent)}. The methods in this class can be called called
   1251      * from several different threads.
   1252      */
   1253     // NOTE: All calls that are passed in a calling app are interned so that
   1254     // they can be used as message objects (which are tested for equality using ==).
   1255     private final ITextToSpeechService.Stub mBinder =
   1256             new ITextToSpeechService.Stub() {
   1257                 @Override
   1258                 public int speak(
   1259                         IBinder caller,
   1260                         CharSequence text,
   1261                         int queueMode,
   1262                         Bundle params,
   1263                         String utteranceId) {
   1264                     if (!checkNonNull(caller, text, params)) {
   1265                         return TextToSpeech.ERROR;
   1266                     }
   1267 
   1268                     SpeechItem item =
   1269                             new SynthesisSpeechItem(
   1270                                     caller,
   1271                                     Binder.getCallingUid(),
   1272                                     Binder.getCallingPid(),
   1273                                     params,
   1274                                     utteranceId,
   1275                                     text);
   1276                     return mSynthHandler.enqueueSpeechItem(queueMode, item);
   1277                 }
   1278 
   1279                 @Override
   1280                 public int synthesizeToFileDescriptor(
   1281                         IBinder caller,
   1282                         CharSequence text,
   1283                         ParcelFileDescriptor fileDescriptor,
   1284                         Bundle params,
   1285                         String utteranceId) {
   1286                     if (!checkNonNull(caller, text, fileDescriptor, params)) {
   1287                         return TextToSpeech.ERROR;
   1288                     }
   1289 
   1290                     // In test env, ParcelFileDescriptor instance may be EXACTLY the same
   1291                     // one that is used by client. And it will be closed by a client, thus
   1292                     // preventing us from writing anything to it.
   1293                     final ParcelFileDescriptor sameFileDescriptor =
   1294                             ParcelFileDescriptor.adoptFd(fileDescriptor.detachFd());
   1295 
   1296                     SpeechItem item =
   1297                             new SynthesisToFileOutputStreamSpeechItem(
   1298                                     caller,
   1299                                     Binder.getCallingUid(),
   1300                                     Binder.getCallingPid(),
   1301                                     params,
   1302                                     utteranceId,
   1303                                     text,
   1304                                     new ParcelFileDescriptor.AutoCloseOutputStream(
   1305                                             sameFileDescriptor));
   1306                     return mSynthHandler.enqueueSpeechItem(TextToSpeech.QUEUE_ADD, item);
   1307                 }
   1308 
   1309                 @Override
   1310                 public int playAudio(
   1311                         IBinder caller,
   1312                         Uri audioUri,
   1313                         int queueMode,
   1314                         Bundle params,
   1315                         String utteranceId) {
   1316                     if (!checkNonNull(caller, audioUri, params)) {
   1317                         return TextToSpeech.ERROR;
   1318                     }
   1319 
   1320                     SpeechItem item =
   1321                             new AudioSpeechItem(
   1322                                     caller,
   1323                                     Binder.getCallingUid(),
   1324                                     Binder.getCallingPid(),
   1325                                     params,
   1326                                     utteranceId,
   1327                                     audioUri);
   1328                     return mSynthHandler.enqueueSpeechItem(queueMode, item);
   1329                 }
   1330 
   1331                 @Override
   1332                 public int playSilence(
   1333                         IBinder caller, long duration, int queueMode, String utteranceId) {
   1334                     if (!checkNonNull(caller)) {
   1335                         return TextToSpeech.ERROR;
   1336                     }
   1337 
   1338                     SpeechItem item =
   1339                             new SilenceSpeechItem(
   1340                                     caller,
   1341                                     Binder.getCallingUid(),
   1342                                     Binder.getCallingPid(),
   1343                                     utteranceId,
   1344                                     duration);
   1345                     return mSynthHandler.enqueueSpeechItem(queueMode, item);
   1346                 }
   1347 
   1348                 @Override
   1349                 public boolean isSpeaking() {
   1350                     return mSynthHandler.isSpeaking() || mAudioPlaybackHandler.isSpeaking();
   1351                 }
   1352 
   1353                 @Override
   1354                 public int stop(IBinder caller) {
   1355                     if (!checkNonNull(caller)) {
   1356                         return TextToSpeech.ERROR;
   1357                     }
   1358 
   1359                     return mSynthHandler.stopForApp(caller);
   1360                 }
   1361 
   1362                 @Override
   1363                 public String[] getLanguage() {
   1364                     return onGetLanguage();
   1365                 }
   1366 
   1367                 @Override
   1368                 public String[] getClientDefaultLanguage() {
   1369                     return getSettingsLocale();
   1370                 }
   1371 
   1372                 /*
   1373                  * If defaults are enforced, then no language is "available" except
   1374                  * perhaps the default language selected by the user.
   1375                  */
   1376                 @Override
   1377                 public int isLanguageAvailable(String lang, String country, String variant) {
   1378                     if (!checkNonNull(lang)) {
   1379                         return TextToSpeech.ERROR;
   1380                     }
   1381 
   1382                     return onIsLanguageAvailable(lang, country, variant);
   1383                 }
   1384 
   1385                 @Override
   1386                 public String[] getFeaturesForLanguage(
   1387                         String lang, String country, String variant) {
   1388                     Set<String> features = onGetFeaturesForLanguage(lang, country, variant);
   1389                     String[] featuresArray = null;
   1390                     if (features != null) {
   1391                         featuresArray = new String[features.size()];
   1392                         features.toArray(featuresArray);
   1393                     } else {
   1394                         featuresArray = new String[0];
   1395                     }
   1396                     return featuresArray;
   1397                 }
   1398 
   1399                 /*
   1400                  * There is no point loading a non default language if defaults
   1401                  * are enforced.
   1402                  */
   1403                 @Override
   1404                 public int loadLanguage(
   1405                         IBinder caller, String lang, String country, String variant) {
   1406                     if (!checkNonNull(lang)) {
   1407                         return TextToSpeech.ERROR;
   1408                     }
   1409                     int retVal = onIsLanguageAvailable(lang, country, variant);
   1410 
   1411                     if (retVal == TextToSpeech.LANG_AVAILABLE
   1412                             || retVal == TextToSpeech.LANG_COUNTRY_AVAILABLE
   1413                             || retVal == TextToSpeech.LANG_COUNTRY_VAR_AVAILABLE) {
   1414 
   1415                         SpeechItem item =
   1416                                 new LoadLanguageItem(
   1417                                         caller,
   1418                                         Binder.getCallingUid(),
   1419                                         Binder.getCallingPid(),
   1420                                         lang,
   1421                                         country,
   1422                                         variant);
   1423 
   1424                         if (mSynthHandler.enqueueSpeechItem(TextToSpeech.QUEUE_ADD, item)
   1425                                 != TextToSpeech.SUCCESS) {
   1426                             return TextToSpeech.ERROR;
   1427                         }
   1428                     }
   1429                     return retVal;
   1430                 }
   1431 
   1432                 @Override
   1433                 public List<Voice> getVoices() {
   1434                     return onGetVoices();
   1435                 }
   1436 
   1437                 @Override
   1438                 public int loadVoice(IBinder caller, String voiceName) {
   1439                     if (!checkNonNull(voiceName)) {
   1440                         return TextToSpeech.ERROR;
   1441                     }
   1442                     int retVal = onIsValidVoiceName(voiceName);
   1443 
   1444                     if (retVal == TextToSpeech.SUCCESS) {
   1445                         SpeechItem item =
   1446                                 new LoadVoiceItem(
   1447                                         caller,
   1448                                         Binder.getCallingUid(),
   1449                                         Binder.getCallingPid(),
   1450                                         voiceName);
   1451                         if (mSynthHandler.enqueueSpeechItem(TextToSpeech.QUEUE_ADD, item)
   1452                                 != TextToSpeech.SUCCESS) {
   1453                             return TextToSpeech.ERROR;
   1454                         }
   1455                     }
   1456                     return retVal;
   1457                 }
   1458 
   1459                 public String getDefaultVoiceNameFor(String lang, String country, String variant) {
   1460                     if (!checkNonNull(lang)) {
   1461                         return null;
   1462                     }
   1463                     int retVal = onIsLanguageAvailable(lang, country, variant);
   1464 
   1465                     if (retVal == TextToSpeech.LANG_AVAILABLE
   1466                             || retVal == TextToSpeech.LANG_COUNTRY_AVAILABLE
   1467                             || retVal == TextToSpeech.LANG_COUNTRY_VAR_AVAILABLE) {
   1468                         return onGetDefaultVoiceNameFor(lang, country, variant);
   1469                     } else {
   1470                         return null;
   1471                     }
   1472                 }
   1473 
   1474                 @Override
   1475                 public void setCallback(IBinder caller, ITextToSpeechCallback cb) {
   1476                     // Note that passing in a null callback is a valid use case.
   1477                     if (!checkNonNull(caller)) {
   1478                         return;
   1479                     }
   1480 
   1481                     mCallbacks.setCallback(caller, cb);
   1482                 }
   1483 
   1484                 private String intern(String in) {
   1485                     // The input parameter will be non null.
   1486                     return in.intern();
   1487                 }
   1488 
   1489                 private boolean checkNonNull(Object... args) {
   1490                     for (Object o : args) {
   1491                         if (o == null) return false;
   1492                     }
   1493                     return true;
   1494                 }
   1495             };
   1496 
   1497     private class CallbackMap extends RemoteCallbackList<ITextToSpeechCallback> {
   1498         private final HashMap<IBinder, ITextToSpeechCallback> mCallerToCallback
   1499                 = new HashMap<IBinder, ITextToSpeechCallback>();
   1500 
   1501         public void setCallback(IBinder caller, ITextToSpeechCallback cb) {
   1502             synchronized (mCallerToCallback) {
   1503                 ITextToSpeechCallback old;
   1504                 if (cb != null) {
   1505                     register(cb, caller);
   1506                     old = mCallerToCallback.put(caller, cb);
   1507                 } else {
   1508                     old = mCallerToCallback.remove(caller);
   1509                 }
   1510                 if (old != null && old != cb) {
   1511                     unregister(old);
   1512                 }
   1513             }
   1514         }
   1515 
   1516         public void dispatchOnStop(Object callerIdentity, String utteranceId, boolean started) {
   1517             ITextToSpeechCallback cb = getCallbackFor(callerIdentity);
   1518             if (cb == null) return;
   1519             try {
   1520                 cb.onStop(utteranceId, started);
   1521             } catch (RemoteException e) {
   1522                 Log.e(TAG, "Callback onStop failed: " + e);
   1523             }
   1524         }
   1525 
   1526         public void dispatchOnSuccess(Object callerIdentity, String utteranceId) {
   1527             ITextToSpeechCallback cb = getCallbackFor(callerIdentity);
   1528             if (cb == null) return;
   1529             try {
   1530                 cb.onSuccess(utteranceId);
   1531             } catch (RemoteException e) {
   1532                 Log.e(TAG, "Callback onDone failed: " + e);
   1533             }
   1534         }
   1535 
   1536         public void dispatchOnStart(Object callerIdentity, String utteranceId) {
   1537             ITextToSpeechCallback cb = getCallbackFor(callerIdentity);
   1538             if (cb == null) return;
   1539             try {
   1540                 cb.onStart(utteranceId);
   1541             } catch (RemoteException e) {
   1542                 Log.e(TAG, "Callback onStart failed: " + e);
   1543             }
   1544         }
   1545 
   1546         public void dispatchOnError(Object callerIdentity, String utteranceId,
   1547                 int errorCode) {
   1548             ITextToSpeechCallback cb = getCallbackFor(callerIdentity);
   1549             if (cb == null) return;
   1550             try {
   1551                 cb.onError(utteranceId, errorCode);
   1552             } catch (RemoteException e) {
   1553                 Log.e(TAG, "Callback onError failed: " + e);
   1554             }
   1555         }
   1556 
   1557         public void dispatchOnBeginSynthesis(Object callerIdentity, String utteranceId, int sampleRateInHz, int audioFormat, int channelCount) {
   1558             ITextToSpeechCallback cb = getCallbackFor(callerIdentity);
   1559             if (cb == null) return;
   1560             try {
   1561                 cb.onBeginSynthesis(utteranceId, sampleRateInHz, audioFormat, channelCount);
   1562             } catch (RemoteException e) {
   1563                 Log.e(TAG, "Callback dispatchOnBeginSynthesis(String, int, int, int) failed: " + e);
   1564             }
   1565         }
   1566 
   1567         public void dispatchOnAudioAvailable(Object callerIdentity, String utteranceId, byte[] buffer) {
   1568             ITextToSpeechCallback cb = getCallbackFor(callerIdentity);
   1569             if (cb == null) return;
   1570             try {
   1571                 cb.onAudioAvailable(utteranceId, buffer);
   1572             } catch (RemoteException e) {
   1573                 Log.e(TAG, "Callback dispatchOnAudioAvailable(String, byte[]) failed: " + e);
   1574             }
   1575         }
   1576 
   1577         public void dispatchOnRangeStart(
   1578                 Object callerIdentity, String utteranceId, int start, int end, int frame) {
   1579             ITextToSpeechCallback cb = getCallbackFor(callerIdentity);
   1580             if (cb == null) return;
   1581             try {
   1582                 cb.onRangeStart(utteranceId, start, end, frame);
   1583             } catch (RemoteException e) {
   1584                 Log.e(TAG, "Callback dispatchOnRangeStart(String, int, int, int) failed: " + e);
   1585             }
   1586         }
   1587 
   1588         @Override
   1589         public void onCallbackDied(ITextToSpeechCallback callback, Object cookie) {
   1590             IBinder caller = (IBinder) cookie;
   1591             synchronized (mCallerToCallback) {
   1592                 mCallerToCallback.remove(caller);
   1593             }
   1594             //mSynthHandler.stopForApp(caller);
   1595         }
   1596 
   1597         @Override
   1598         public void kill() {
   1599             synchronized (mCallerToCallback) {
   1600                 mCallerToCallback.clear();
   1601                 super.kill();
   1602             }
   1603         }
   1604 
   1605         private ITextToSpeechCallback getCallbackFor(Object caller) {
   1606             ITextToSpeechCallback cb;
   1607             IBinder asBinder = (IBinder) caller;
   1608             synchronized (mCallerToCallback) {
   1609                 cb = mCallerToCallback.get(asBinder);
   1610             }
   1611 
   1612             return cb;
   1613         }
   1614     }
   1615 }
   1616