Implement Google network speech synthesis.

See bug for context. Implements a component extension
that provides speech synthesis using Google's speech
synthesis API.

Adds a "remote" flag to the TTS and TTS Engine APIs
so that it's possible for clients to distinguish
between local and remote speech engines.

Adds a new private extension API to expose Google's
API key, needed to make the request.

BUG=308250

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=232242

Review URL: https://codereview.chromium.org/27034009

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@232412 0039d316-1c4b-4281-b951-d872f2087c98

25 files changed, 507 insertions, 46 deletions

diff --git a/chrome/browser/browser_resources.grd b/chrome/browser/browser_resources.grd
index db93304..687a4d0 100644
--- a/chrome/browser/browser_resources.grd
+++ b/chrome/browser/browser_resources.grd
@@ -149,6 +149,7 @@
       <include name="IDR_INSTANT_JS" file="resources\instant\instant.js" flattenhtml="true" type="BINDATA" />
       <include name="IDR_NET_EXPORT_HTML" file="resources\net_export\net_export.html" flattenhtml="true" type="BINDATA" />
       <include name="IDR_NET_EXPORT_JS" file="resources\net_export\net_export.js" flattenhtml="true" type="BINDATA" />
+      <include name="IDR_NETWORK_SPEECH_SYNTHESIS_MANIFEST" file="resources\network_speech_synthesis\manifest.json" type="BINDATA" />
       <if expr="not is_android">
         <include name="IDR_PERFORMANCE_MONITOR_CHART_CSS" file="resources\performance_monitor\chart.css" flattenhtml="true" type="BINDATA" />
         <include name="IDR_PERFORMANCE_MONITOR_CHART_JS" file="resources\performance_monitor\chart.js" type="BINDATA" />
diff --git a/chrome/browser/extensions/api/system_private/system_private_api.cc b/chrome/browser/extensions/api/system_private/system_private_api.cc
index 20e3c37..0ef4103 100644
--- a/chrome/browser/extensions/api/system_private/system_private_api.cc
+++ b/chrome/browser/extensions/api/system_private/system_private_api.cc
@@ -11,6 +11,7 @@
 #include "chrome/browser/profiles/profile.h"
 #include "chrome/common/extensions/api/system_private.h"
 #include "chrome/common/pref_names.h"
+#include "google_apis/google_api_keys.h"
 
 #if defined(OS_CHROMEOS)
 #include "chromeos/dbus/dbus_thread_manager.h"
@@ -132,6 +133,11 @@ bool SystemPrivateGetUpdateStatusFunction::RunImpl() {
   return true;
 }
 
+bool SystemPrivateGetApiKeyFunction::RunImpl() {
+  SetResult(new base::StringValue(google_apis::GetAPIKey()));
+  return true;
+}
+
 void DispatchVolumeChangedEvent(double volume, bool is_volume_muted) {
   base::DictionaryValue* dict = new base::DictionaryValue();
   dict->SetDouble(kVolumeKey, volume);
diff --git a/chrome/browser/extensions/api/system_private/system_private_api.h b/chrome/browser/extensions/api/system_private/system_private_api.h
index 33f91ba..ddb820b 100644
--- a/chrome/browser/extensions/api/system_private/system_private_api.h
+++ b/chrome/browser/extensions/api/system_private/system_private_api.h
@@ -39,6 +39,18 @@ class SystemPrivateGetUpdateStatusFunction
   virtual bool RunImpl() OVERRIDE;
 };
 
+// API function which returns the Google API key.
+class SystemPrivateGetApiKeyFunction : public SyncExtensionFunction {
+ public:
+  DECLARE_EXTENSION_FUNCTION("systemPrivate.getApiKey", SYSTEMPRIVATE_GETAPIKEY)
+
+ protected:
+  virtual ~SystemPrivateGetApiKeyFunction() {}
+
+  // ExtensionFunction:
+  virtual bool RunImpl() OVERRIDE;
+};
+
 // Dispatches systemPrivate.onBrightnessChanged event for extensions.
 void DispatchBrightnessChangedEvent(int brightness, bool user_initiated);
 
diff --git a/chrome/browser/extensions/component_loader.cc b/chrome/browser/extensions/component_loader.cc
index 07ea15e..c023be6 100644
--- a/chrome/browser/extensions/component_loader.cc
+++ b/chrome/browser/extensions/component_loader.cc
@@ -316,6 +316,11 @@ void ComponentLoader::AddBookmarksExtensions() {
 #endif
 }
 
+void ComponentLoader::AddNetworkSpeechSynthesisExtension() {
+  Add(IDR_NETWORK_SPEECH_SYNTHESIS_MANIFEST,
+      base::FilePath(FILE_PATH_LITERAL("network_speech_synthesis")));
+}
+
 void ComponentLoader::AddWithName(int manifest_resource_id,
                                   const base::FilePath& root_directory,
                                   const std::string& name) {
@@ -536,6 +541,10 @@ void ComponentLoader::AddDefaultComponentExtensionsWithBackgroundPages(
         base::FilePath(FILE_PATH_LITERAL("google_now")));
   }
 #endif
+
+#if defined(GOOGLE_CHROME_BUILD)
+  AddNetworkSpeechSynthesisExtension();
+#endif  // defined(GOOGLE_CHROME_BUILD)
 }
 
 void ComponentLoader::UnloadComponent(ComponentExtensionInfo* component) {
diff --git a/chrome/browser/extensions/component_loader.h b/chrome/browser/extensions/component_loader.h
index 8b1a961..1dd2e8d 100644
--- a/chrome/browser/extensions/component_loader.h
+++ b/chrome/browser/extensions/component_loader.h
@@ -115,6 +115,7 @@ class ComponentLoader {
   void AddHangoutServicesExtension();
   void AddImageLoaderExtension();
   void AddBookmarksExtensions();
+  void AddNetworkSpeechSynthesisExtension();
 
   void AddWithName(int manifest_resource_id,
                    const base::FilePath& root_directory,
@@ -135,6 +136,8 @@ class ComponentLoader {
   typedef std::vector<ComponentExtensionInfo> RegisteredComponentExtensions;
   RegisteredComponentExtensions component_extensions_;
 
+  FRIEND_TEST_ALL_PREFIXES(TtsApiTest, NetworkSpeechEngine);
+
   DISALLOW_COPY_AND_ASSIGN(ComponentLoader);
 };
 
diff --git a/chrome/browser/extensions/extension_function_histogram_value.h b/chrome/browser/extensions/extension_function_histogram_value.h
index 14d357d..42e5ccd 100644
--- a/chrome/browser/extensions/extension_function_histogram_value.h
+++ b/chrome/browser/extensions/extension_function_histogram_value.h
@@ -682,6 +682,7 @@ enum HistogramValue {
   APP_CURRENTWINDOWINTERNAL_SETMINHEIGHT,
   APP_CURRENTWINDOWINTERNAL_SETMAXWIDTH,
   APP_CURRENTWINDOWINTERNAL_SETMAXHEIGHT,
+  SYSTEMPRIVATE_GETAPIKEY,
   ENUM_BOUNDARY // Last entry: Add new entries above.
 };
 
diff --git a/chrome/browser/resources/component_extension_resources.grd b/chrome/browser/resources/component_extension_resources.grd
index f1a9202..5107a7d 100644
--- a/chrome/browser/resources/component_extension_resources.grd
+++ b/chrome/browser/resources/component_extension_resources.grd
@@ -21,6 +21,8 @@
       </if>
     </structures>
     <includes>
+      <include name="IDR_NETWORK_SPEECH_SYNTHESIS_JS" file="network_speech_synthesis/tts_extension.js" type="BINDATA" />
+
       <include name="IDR_BOOKMARK_MANAGER_BOOKMARK_MANAGER_SEARCH" file="bookmark_manager/images/bookmark_manager_search.png" type="BINDATA" />
       <include name="IDR_BOOKMARK_MANAGER_BOOKMARK_MANAGER_SEARCH_RTL" file="bookmark_manager/images/bookmark_manager_search_rtl.png" type="BINDATA" />
       <include name="IDR_BOOKMARK_MANAGER_BOOKMARK_MAIN_JS" file="bookmark_manager/js/main.js" type="BINDATA" />
diff --git a/chrome/browser/resources/network_speech_synthesis/manifest.json b/chrome/browser/resources/network_speech_synthesis/manifest.json
new file mode 100644
index 0000000..9019ee6
--- /dev/null
+++ b/chrome/browser/resources/network_speech_synthesis/manifest.json
@@ -0,0 +1,90 @@
+{
+   "background": {
+      "scripts": [ "tts_extension.js" ],
+      "persistent": false
+   },
+   "description": "Component extension providing speech via the Google network text-to-speech service.",
+   "key": "MIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8AMIIBCgKCAQEA8GSbNUMGygqQTNDMFGIjZNcwXsHLzkNkHjWbuY37PbNdSDZ4VqlVjzbWqODSe+MjELdv5Keb51IdytnoGYXBMyqKmWpUrg+RnKvQ5ibWr4MW9pyIceOIdp9GrzC1WZGgTmZismYR3AjaIpufZ7xDdQQv+XrghPWCkdVqLN+qZDA1HU+DURznkMICiDDSH2sU0egm9UbWfS218bZqzKeQDiC3OnTPlaxcbJtKUuupIm5knjze3Wo9Ae9poTDMzKgchg0VlFCv3uqox+wlD8sjXBoyBCCK9HpImdVAF1a7jpdgiUHpPeV/26oYzM9/grltwNR3bzECQgSpyXp0eyoegwIDAQAB",
+   "manifest_version": 2,
+   "name": "Google Network Speech",
+   "permissions": [
+     "systemPrivate",
+     "ttsEngine",
+     "https://www.google.com/"
+   ],
+   "tts_engine": {
+      "voices": [
+         {
+           "event_types": [ "start", "end", "error" ],
+           "gender": "female",
+           "lang": "en-US",
+           "voice_name": "Google US English",
+           "remote": true
+         },
+         {
+           "event_types": [ "start", "end", "error" ],
+           "gender": "male",
+           "lang": "en-GB",
+           "voice_name": "Google UK English Male",
+           "remote": true
+         },
+         {
+           "event_types": [ "start", "end", "error" ],
+           "gender": "female",
+           "lang": "en-GB",
+           "voice_name": "Google UK English Female",
+           "remote": true
+         },
+         {
+           "event_types": [ "start", "end", "error" ],
+           "gender": "female",
+           "lang": "es-ES",
+           "voice_name": "Google Español",
+           "remote": true
+         },
+         {
+           "event_types": [ "start", "end", "error" ],
+           "gender": "female",
+           "lang": "fr-FR",
+           "voice_name": "Google Français",
+           "remote": true
+         },
+         {
+           "event_types": [ "start", "end", "error" ],
+           "gender": "female",
+           "lang": "it-IT",
+           "voice_name": "Google Italiano",
+           "remote": true
+         },
+         {
+           "event_types": [ "start", "end", "error" ],
+           "gender": "female",
+           "lang": "de-DE",
+           "voice_name": "Google Deutsch",
+           "remote": true
+         },
+         {
+           "event_types": [ "start", "end", "error" ],
+           "gender": "female",
+           "lang": "ja-JP",
+           "voice_name": "Google 日本人",
+           "remote": true
+         },
+         {
+           "event_types": [ "start", "end", "error" ],
+           "gender": "female",
+           "lang": "ko-KR",
+           "voice_name": "Google 한국의",
+           "remote": true
+         },
+         {
+           "event_types": [ "start", "end", "error" ],
+           "gender": "female",
+           "lang": "zh-CN",
+           "voice_name": "Google 中国的",
+           "remote": true
+         }
+       ]
+   },
+   "version": "1.0"
+}
diff --git a/chrome/browser/resources/network_speech_synthesis/tts_extension.js b/chrome/browser/resources/network_speech_synthesis/tts_extension.js
new file mode 100644
index 0000000..b2a3e71
--- /dev/null
+++ b/chrome/browser/resources/network_speech_synthesis/tts_extension.js
@@ -0,0 +1,251 @@
+// Copyright 2013 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+/**
+ * @fileoverview
+ * This is a component extension that implements a text-to-speech (TTS)
+ * engine powered by Google's speech synthesis API.
+ *
+ * This is an "event page", so it's not loaded when the API isn't being used,
+ * and doesn't waste resources. When a web page or web app makes a speech
+ * request and the parameters match one of the voices in this extension's
+ * manifest, it makes a request to Google's API using Chrome's private key
+ * and plays the resulting speech using HTML5 audio.
+ */
+
+/**
+ * The main class for this extension. Adds listeners to
+ * chrome.ttsEngine.onSpeak and chrome.ttsEngine.onStop and implements
+ * them using Google's speech synthesis API.
+ * @constructor
+ */
+function TtsExtension() {}
+
+TtsExtension.prototype = {
+  /**
+   * The url prefix of the speech server, including static query
+   * parameters that don't change.
+   * @type {string}
+   * @const
+   * @private
+   */
+  SPEECH_SERVER_URL_:
+      'https://www.google.com/speech-api/v2/synthesize?' +
+      'enc=mpeg&client=chromium',
+
+  /**
+   * A mapping from language and gender to voice name, hardcoded for now
+   * until the speech synthesis server capabilities response provides this.
+   * The key of this map is of the form '<lang>-<gender>'.
+   * @type {Object.<string, string>}
+   * @private
+   */
+  LANG_AND_GENDER_TO_VOICE_NAME_: {
+    'en-gb-male': 'rjs',
+    'en-gb-female': 'fis',
+  },
+
+  /**
+   * The arguments passed to the onSpeak event handler for the utterance
+   * that's currently being spoken. Should be null when no object is
+   * pending.
+   *
+   * @type {?{utterance: string, options: Object, callback: Function}}
+   * @private
+   */
+  currentUtterance_: null,
+
+  /**
+   * The HTML5 audio element we use for playing the sound served by the
+   * speech server.
+   * @type {HTMLAudioElement}
+   * @private
+   */
+  audioElement_: null,
+
+  /**
+   * A mapping from voice name to language and gender, derived from the
+   * manifest file.  This is used in case the speech synthesis request
+   * specifies a voice name but doesn't specify a language code or gender.
+   * @type {Object.<string, {lang: string, gender: string}>}
+   * @private
+   */
+  voiceNameToLangAndGender_: {},
+
+  /**
+   * This is the main function called to initialize this extension.
+   * Initializes data structures and adds event listeners.
+   */
+  init: function() {
+    // Get voices from manifest.
+    var voices = chrome.app.getDetails().tts_engine.voices;
+    for (var i = 0; i < voices.length; i++) {
+      this.voiceNameToLangAndGender_[voices[i].voice_name] = {
+        lang: voices[i].lang,
+        gender: voices[i].gender
+      };
+    }
+
+    // Initialize the audio element and event listeners on it.
+    this.audioElement_ = document.createElement('audio');
+    document.body.appendChild(this.audioElement_);
+    this.audioElement_.addEventListener(
+        'ended', this.onStop_.bind(this), false);
+    this.audioElement_.addEventListener(
+        'canplaythrough', this.onStart_.bind(this), false);
+
+    // Install event listeners for the ttsEngine API.
+    chrome.ttsEngine.onSpeak.addListener(this.onSpeak_.bind(this));
+    chrome.ttsEngine.onStop.addListener(this.onStop_.bind(this));
+    chrome.ttsEngine.onPause.addListener(this.onPause_.bind(this));
+    chrome.ttsEngine.onResume.addListener(this.onResume_.bind(this));
+  },
+
+  /**
+   * Handler for the chrome.ttsEngine.onSpeak interface.
+   * Gets Chrome's Google API key and then uses it to generate a request
+   * url for the requested speech utterance. Sets that url as the source
+   * of the HTML5 audio element.
+   * @param {string} utterance The text to be spoken.
+   * @param {Object} options Options to control the speech, as defined
+   *     in the Chrome ttsEngine extension API.
+   * @private
+   */
+  onSpeak_: function(utterance, options, callback) {
+    // Truncate the utterance if it's too long. Both Chrome's tts
+    // extension api and the web speech api specify 32k as the
+    // maximum limit for an utterance.
+    if (utterance.length > 32768)
+      utterance = utterance.substr(0, 32768);
+
+    try {
+      // First, stop any pending audio.
+      this.onStop_();
+
+      this.currentUtterance_ = {
+        utterance: utterance,
+        options: options,
+        callback: callback
+      };
+
+      var lang = options.lang;
+      var gender = options.gender;
+      if (options.voiceName) {
+        lang = this.voiceNameToLangAndGender_[options.voiceName].lang;
+        gender = this.voiceNameToLangAndGender_[options.voiceName].gender;
+      }
+
+      // Look up the specific voice name for this language and gender.
+      // If it's not in the map, it doesn't matter - the language will
+      // be used directly. This is only used for languages where more
+      // than one gender is actually available.
+      var key = lang.toLowerCase() + '-' + gender;
+      var voiceName = this.LANG_AND_GENDER_TO_VOICE_NAME_[key];
+
+      var url = this.SPEECH_SERVER_URL_;
+      chrome.systemPrivate.getApiKey((function(key) {
+        url += '&key=' + key;
+        url += '&text=' + escape(utterance);
+        url += '&lang=' + lang.toLowerCase();
+
+        if (voiceName)
+          url += '&name=' + voiceName;
+
+        if (options.rate) {
+          // Input rate is between 0.1 and 10.0 with a default of 1.0.
+          // Output speed is between 0.0 and 1.0 with a default of 0.5.
+          url += '&speed=' + (options.rate / 2.0);
+        }
+
+        if (options.pitch) {
+          // Input pitch is between 0.0 and 2.0 with a default of 1.0.
+          // Output pitch is between 0.0 and 1.0 with a default of 0.5.
+          url += '&pitch=' + (options.pitch / 2.0);
+        }
+
+        // This begins loading the audio but does not play it.
+        // When enough of the audio has loaded to begin playback,
+        // the 'canplaythrough' handler will call this.onStart_,
+        // which sends a start event to the ttsEngine callback and
+        // then begins playing audio.
+        this.audioElement_.src = url;
+      }).bind(this));
+    } catch (err) {
+      console.error(String(err));
+      callback({
+        'type': 'error',
+        'errorMessage': String(err)
+      });
+      this.currentUtterance_ = null;
+    }
+  },
+
+  /**
+   * Handler for the chrome.ttsEngine.onStop interface.
+   * Called either when the ttsEngine API requests us to stop, or when
+   * we reach the end of the audio stream. Pause the audio element to
+   * silence it, and send a callback to the ttsEngine API to let it know
+   * that we've completed. Note that the ttsEngine API manages callback
+   * messages and will automatically replace the 'end' event with a
+   * more specific callback like 'interrupted' when sending it to the
+   * TTS client.
+   * @private
+   */
+  onStop_: function() {
+    if (this.currentUtterance_) {
+      this.audioElement_.pause();
+      this.currentUtterance_.callback({
+        'type': 'end',
+        'charIndex': this.currentUtterance_.utterance.length
+      });
+    }
+    this.currentUtterance_ = null;
+  },
+
+  /**
+   * Handler for the canplaythrough event on the audio element.
+   * Called when the audio element has buffered enough audio to begin
+   * playback. Send the 'start' event to the ttsEngine callback and
+   * then begin playing the audio element.
+   * @private
+   */
+  onStart_: function() {
+    if (this.currentUtterance_) {
+      if (this.currentUtterance_.options.volume !== undefined) {
+        // Both APIs use the same range for volume, between 0.0 and 1.0.
+        this.audioElement_.volume = this.currentUtterance_.options.volume;
+      }
+      this.audioElement_.play();
+      this.currentUtterance_.callback({
+          'type': 'start',
+          'charIndex': 0
+      });
+    }
+  },
+
+  /**
+   * Handler for the chrome.ttsEngine.onPause interface.
+   * Pauses audio if we're in the middle of an utterance.
+   * @private
+   */
+  onPause_: function() {
+    if (this.currentUtterance_) {
+      this.audioElement_.pause();
+    }
+  },
+
+  /**
+   * Handler for the chrome.ttsEngine.onPause interface.
+   * Resumes audio if we're in the middle of an utterance.
+   * @private
+   */
+  onResume_: function() {
+    if (this.currentUtterance_) {
+      this.audioElement_.play();
+    }
+  }
+
+};
+
+(new TtsExtension()).init();
diff --git a/chrome/browser/speech/extension_api/tts_engine_extension_api.cc b/chrome/browser/speech/extension_api/tts_engine_extension_api.cc
index ba5e65b..b6908760 100644
--- a/chrome/browser/speech/extension_api/tts_engine_extension_api.cc
+++ b/chrome/browser/speech/extension_api/tts_engine_extension_api.cc
@@ -91,6 +91,7 @@ void GetExtensionVoices(Profile* profile, std::vector<VoiceData>* out_voices) {
       result_voice.native = false;
       result_voice.name = voice.voice_name;
       result_voice.lang = voice.lang;
+      result_voice.remote = voice.remote;
       result_voice.extension_id = extension->id();
       if (voice.gender == constants::kGenderMale)
         result_voice.gender = TTS_GENDER_MALE;
diff --git a/chrome/browser/speech/extension_api/tts_extension_api.cc b/chrome/browser/speech/extension_api/tts_extension_api.cc
index 6523b4e..14c5641 100644
--- a/chrome/browser/speech/extension_api/tts_extension_api.cc
+++ b/chrome/browser/speech/extension_api/tts_extension_api.cc
@@ -307,6 +307,7 @@ bool TtsGetVoicesFunction::RunImpl() {
     const VoiceData& voice = voices[i];
     DictionaryValue* result_voice = new DictionaryValue();
     result_voice->SetString(constants::kVoiceNameKey, voice.name);
+    result_voice->SetBoolean(constants::kRemoteKey, voice.remote);
     if (!voice.lang.empty())
       result_voice->SetString(constants::kLangKey, voice.lang);
     if (voice.gender == TTS_GENDER_MALE)
diff --git a/chrome/browser/speech/extension_api/tts_extension_api_constants.cc b/chrome/browser/speech/extension_api/tts_extension_api_constants.cc
index 7b8c45b..f1853dc 100644
--- a/chrome/browser/speech/extension_api/tts_extension_api_constants.cc
+++ b/chrome/browser/speech/extension_api/tts_extension_api_constants.cc
@@ -6,48 +6,49 @@
 
 namespace tts_extension_api_constants {
 
-const char kVoiceNameKey[] = "voiceName";
-const char kLangKey[] = "lang";
-const char kGenderKey[] = "gender";
-const char kRateKey[] = "rate";
-const char kPitchKey[] = "pitch";
-const char kVolumeKey[] = "volume";
+const char kCharIndexKey[] = "charIndex";
+const char kDesiredEventTypesKey[] = "desiredEventTypes";
 const char kEnqueueKey[] = "enqueue";
+const char kErrorMessageKey[] = "errorMessage";
 const char kEventTypeKey[] = "type";
 const char kEventTypesKey[] = "eventTypes";
-const char kCharIndexKey[] = "charIndex";
-const char kErrorMessageKey[] = "errorMessage";
-const char kRequiredEventTypesKey[] = "requiredEventTypes";
-const char kDesiredEventTypesKey[] = "desiredEventTypes";
 const char kExtensionIdKey[] = "extensionId";
-const char kSrcIdKey[] = "srcId";
+const char kGenderKey[] = "gender";
 const char kIsFinalEventKey[] = "isFinalEvent";
+const char kLangKey[] = "lang";
 const char kOnEventKey[] = "onEvent";
+const char kPitchKey[] = "pitch";
+const char kRateKey[] = "rate";
+const char kRemoteKey[] = "remote";
+const char kRequiredEventTypesKey[] = "requiredEventTypes";
+const char kSrcIdKey[] = "srcId";
+const char kVoiceNameKey[] = "voiceName";
+const char kVolumeKey[] = "volume";
 
 const char kGenderFemale[] = "female";
 const char kGenderMale[] = "male";
 
-const char kEventTypeStart[] = "start";
-const char kEventTypeEnd[] = "end";
-const char kEventTypeWord[] = "word";
-const char kEventTypeSentence[] = "sentence";
-const char kEventTypeMarker[] = "marker";
-const char kEventTypeInterrupted[] = "interrupted";
 const char kEventTypeCancelled[] = "cancelled";
+const char kEventTypeEnd[] = "end";
 const char kEventTypeError[] = "error";
+const char kEventTypeInterrupted[] = "interrupted";
+const char kEventTypeMarker[] = "marker";
 const char kEventTypePause[] = "pause";
 const char kEventTypeResume[] = "resume";
+const char kEventTypeSentence[] = "sentence";
+const char kEventTypeStart[] = "start";
+const char kEventTypeWord[] = "word";
 
-const char kErrorUndeclaredEventType[] =
-    "Cannot send an event type that is not declared in the extension manifest.";
-const char kErrorUtteranceTooLong[] = "Utterance length is too long.";
-const char kErrorInvalidLang[] = "Invalid lang.";
 const char kErrorInvalidGender[] = "Invalid gender.";
-const char kErrorInvalidRate[] = "Invalid rate.";
+const char kErrorInvalidLang[] = "Invalid lang.";
 const char kErrorInvalidPitch[] = "Invalid pitch.";
+const char kErrorInvalidRate[] = "Invalid rate.";
 const char kErrorInvalidVolume[] = "Invalid volume.";
 const char kErrorMissingPauseOrResume[] =
     "A TTS engine extension should either listen for both onPause and onResume "
     "events, or neither.";
+const char kErrorUndeclaredEventType[] =
+    "Cannot send an event type that is not declared in the extension manifest.";
+const char kErrorUtteranceTooLong[] = "Utterance length is too long.";
 
 }  // namespace tts_extension_api_constants.
diff --git a/chrome/browser/speech/extension_api/tts_extension_api_constants.h b/chrome/browser/speech/extension_api/tts_extension_api_constants.h
index 0809473..bec1749 100644
--- a/chrome/browser/speech/extension_api/tts_extension_api_constants.h
+++ b/chrome/browser/speech/extension_api/tts_extension_api_constants.h
@@ -11,46 +11,47 @@
 
 namespace tts_extension_api_constants {
 
-extern const char kVoiceNameKey[];
-extern const char kLangKey[];
-extern const char kGenderKey[];
-extern const char kRateKey[];
-extern const char kPitchKey[];
-extern const char kVolumeKey[];
+extern const char kCharIndexKey[];
+extern const char kDesiredEventTypesKey[];
 extern const char kEnqueueKey[];
+extern const char kErrorMessageKey[];
 extern const char kEventTypeKey[];
 extern const char kEventTypesKey[];
-extern const char kCharIndexKey[];
-extern const char kErrorMessageKey[];
-extern const char kRequiredEventTypesKey[];
-extern const char kDesiredEventTypesKey[];
 extern const char kExtensionIdKey[];
-extern const char kSrcIdKey[];
+extern const char kGenderKey[];
 extern const char kIsFinalEventKey[];
+extern const char kLangKey[];
 extern const char kOnEventKey[];
+extern const char kPitchKey[];
+extern const char kRateKey[];
+extern const char kRemoteKey[];
+extern const char kRequiredEventTypesKey[];
+extern const char kSrcIdKey[];
+extern const char kVoiceNameKey[];
+extern const char kVolumeKey[];
 
 extern const char kGenderFemale[];
 extern const char kGenderMale[];
 
-extern const char kEventTypeStart[];
-extern const char kEventTypeEnd[];
-extern const char kEventTypeWord[];
-extern const char kEventTypeSentence[];
-extern const char kEventTypeMarker[];
-extern const char kEventTypeInterrupted[];
 extern const char kEventTypeCancelled[];
+extern const char kEventTypeEnd[];
 extern const char kEventTypeError[];
+extern const char kEventTypeInterrupted[];
+extern const char kEventTypeMarker[];
 extern const char kEventTypePause[];
 extern const char kEventTypeResume[];
+extern const char kEventTypeSentence[];
+extern const char kEventTypeStart[];
+extern const char kEventTypeWord[];
 
-extern const char kErrorUndeclaredEventType[];
-extern const char kErrorUtteranceTooLong[];
-extern const char kErrorInvalidLang[];
 extern const char kErrorInvalidGender[];
-extern const char kErrorInvalidRate[];
+extern const char kErrorInvalidLang[];
 extern const char kErrorInvalidPitch[];
+extern const char kErrorInvalidRate[];
 extern const char kErrorInvalidVolume[];
 extern const char kErrorMissingPauseOrResume[];
+extern const char kErrorUndeclaredEventType[];
+extern const char kErrorUtteranceTooLong[];
 
 }  // namespace tts_extension_api_constants.
 #endif  // CHROME_BROWSER_SPEECH_EXTENSION_API_TTS_EXTENSION_API_CONSTANTS_H_
diff --git a/chrome/browser/speech/extension_api/tts_extension_apitest.cc b/chrome/browser/speech/extension_api/tts_extension_apitest.cc
index 63cc83c..b9e7a9a 100644
--- a/chrome/browser/speech/extension_api/tts_extension_apitest.cc
+++ b/chrome/browser/speech/extension_api/tts_extension_apitest.cc
@@ -6,7 +6,10 @@
 #include "base/command_line.h"
 #include "base/memory/weak_ptr.h"
 #include "base/message_loop/message_loop.h"
+#include "chrome/browser/extensions/component_loader.h"
 #include "chrome/browser/extensions/extension_apitest.h"
+#include "chrome/browser/extensions/extension_service.h"
+#include "chrome/browser/extensions/extension_system.h"
 #include "chrome/browser/speech/extension_api/tts_extension_api.h"
 #include "chrome/browser/speech/tts_controller.h"
 #include "chrome/browser/speech/tts_platform.h"
@@ -32,6 +35,8 @@ namespace {
 int g_saved_utterance_id;
 }
 
+namespace extensions {
+
 class MockTtsPlatformImpl : public TtsPlatformImpl {
  public:
   MockTtsPlatformImpl()
@@ -375,7 +380,16 @@ IN_PROC_BROWSER_TEST_F(TtsApiTest, LangMatching) {
   ASSERT_TRUE(RunExtensionTest("tts_engine/lang_matching")) << message_;
 }
 
+IN_PROC_BROWSER_TEST_F(TtsApiTest, NetworkSpeechEngine) {
+  ExtensionService* service = extensions::ExtensionSystem::Get(
+      profile())->extension_service();
+  service->component_loader()->AddNetworkSpeechSynthesisExtension();
+  ASSERT_TRUE(RunExtensionTest("tts_engine/network_speech_engine")) << message_;
+}
+
 // http://crbug.com/122474
 IN_PROC_BROWSER_TEST_F(TtsApiTest, EngineApi) {
   ASSERT_TRUE(RunExtensionTest("tts_engine/engine_api")) << message_;
 }
+
+}  // namespace extensions
diff --git a/chrome/browser/speech/tts_controller.cc b/chrome/browser/speech/tts_controller.cc
index a0d376f..34f72a4 100644
--- a/chrome/browser/speech/tts_controller.cc
+++ b/chrome/browser/speech/tts_controller.cc
@@ -57,6 +57,7 @@ UtteranceContinuousParameters::UtteranceContinuousParameters()
 
 VoiceData::VoiceData()
     : gender(TTS_GENDER_NONE),
+      remote(false),
       native(false) {}
 
 VoiceData::~VoiceData() {}
@@ -419,4 +420,3 @@ void TtsController::RemoveVoicesChangedDelegate(
     VoicesChangedDelegate* delegate) {
   voices_changed_delegates_.erase(delegate);
 }
-
diff --git a/chrome/browser/speech/tts_controller.h b/chrome/browser/speech/tts_controller.h
index 18cbcd9..6081518 100644
--- a/chrome/browser/speech/tts_controller.h
+++ b/chrome/browser/speech/tts_controller.h
@@ -66,6 +66,10 @@ struct VoiceData {
   std::string extension_id;
   std::set<TtsEventType> events;
 
+  // If true, the synthesis engine is a remote network resource.
+  // It may be higher latency and may incur bandwidth costs.
+  bool remote;
+
   // If true, this is implemented by this platform's subclass of
   // TtsPlatformImpl. If false, this is implemented by an extension.
   bool native;
diff --git a/chrome/browser/speech/tts_message_filter.cc b/chrome/browser/speech/tts_message_filter.cc
index 714c51f..3c397cd 100644
--- a/chrome/browser/speech/tts_message_filter.cc
+++ b/chrome/browser/speech/tts_message_filter.cc
@@ -67,7 +67,7 @@ void TtsMessageFilter::OnInitializeVoiceList() {
     out_voice.voice_uri = voices[i].name;
     out_voice.name = voices[i].name;
     out_voice.lang = voices[i].lang;
-    out_voice.local_service = true;
+    out_voice.local_service = !voices[i].remote;
     out_voice.is_default = (i == 0);
   }
   Send(new TtsMsg_SetVoiceList(out_voices));
diff --git a/chrome/common/extensions/api/speech/tts_engine_manifest_handler.cc b/chrome/common/extensions/api/speech/tts_engine_manifest_handler.cc
index bed4937..e18c594 100644
--- a/chrome/common/extensions/api/speech/tts_engine_manifest_handler.cc
+++ b/chrome/common/extensions/api/speech/tts_engine_manifest_handler.cc
@@ -29,7 +29,8 @@ struct TtsVoices : public Extension::ManifestData {
 
 }  // namespace
 
-TtsVoice::TtsVoice() {}
+TtsVoice::TtsVoice() : remote(false) {}
+
 TtsVoice::~TtsVoice() {}
 
 // static
@@ -95,6 +96,13 @@ bool TtsEngineManifestHandler::Parse(Extension* extension, string16* error) {
         return false;
       }
     }
+    if (one_tts_voice->HasKey(keys::kTtsVoicesRemote)) {
+      if (!one_tts_voice->GetBoolean(
+              keys::kTtsVoicesRemote, &voice_data.remote)) {
+        *error = ASCIIToUTF16(errors::kInvalidTtsVoicesRemote);
+        return false;
+      }
+    }
     if (one_tts_voice->HasKey(keys::kTtsVoicesEventTypes)) {
       const base::ListValue* event_types_list;
       if (!one_tts_voice->GetList(
diff --git a/chrome/common/extensions/api/speech/tts_engine_manifest_handler.h b/chrome/common/extensions/api/speech/tts_engine_manifest_handler.h
index 71d8e05..13b194b 100644
--- a/chrome/common/extensions/api/speech/tts_engine_manifest_handler.h
+++ b/chrome/common/extensions/api/speech/tts_engine_manifest_handler.h
@@ -21,6 +21,7 @@ struct TtsVoice {
   std::string voice_name;
   std::string lang;
   std::string gender;
+  bool remote;
   std::set<std::string> event_types;
 
   static const std::vector<TtsVoice>* GetTtsVoices(const Extension* extension);
diff --git a/chrome/common/extensions/api/system_private.json b/chrome/common/extensions/api/system_private.json
index 7d95450..ea9ae38 100644
--- a/chrome/common/extensions/api/system_private.json
+++ b/chrome/common/extensions/api/system_private.json
@@ -86,6 +86,24 @@
             ]
           }
         ]
+      },
+      {
+        "name": "getApiKey",
+        "type": "function",
+        "description": "Gets Chrome's API key to use for requests to Google services.",
+        "parameters": [
+          {
+            "type": "function",
+            "name": "callback",
+            "parameters": [
+              {
+                "name": "key",
+                "type": "string",
+                "description": "The API key."
+              }
+            ]
+          }
+        ]
       }
     ],
     "events": [
diff --git a/chrome/common/extensions/api/tts.json b/chrome/common/extensions/api/tts.json
index ae8771e..995968ca 100644
--- a/chrome/common/extensions/api/tts.json
+++ b/chrome/common/extensions/api/tts.json
@@ -62,6 +62,11 @@
             "description": "This voice's gender.",
             "enum": ["male", "female"]
           },
+          "remote": {
+            "type": "boolean",
+            "optional": true,
+            "description": "If true, the synthesis engine is a remote network resource. It may be higher latency and may incur bandwidth costs."
+          },
           "extensionId": {
             "type": "string",
             "optional": true,
diff --git a/chrome/test/data/extensions/api_test/tts_engine/network_speech_engine/manifest.json b/chrome/test/data/extensions/api_test/tts_engine/network_speech_engine/manifest.json
new file mode 100644
index 0000000..14fec9e
--- /dev/null
+++ b/chrome/test/data/extensions/api_test/tts_engine/network_speech_engine/manifest.json
@@ -0,0 +1,10 @@
+{
+  "name": "chrome.tts",
+  "version": "0.1",
+  "manifest_version": 2,
+  "description": "browser test for chrome.tts API",
+  "background": {
+    "scripts": [ "test.js" ]
+  },
+  "permissions": ["tts"]
+}
diff --git a/chrome/test/data/extensions/api_test/tts_engine/network_speech_engine/test.js b/chrome/test/data/extensions/api_test/tts_engine/network_speech_engine/test.js
new file mode 100644
index 0000000..72b7e34
--- /dev/null
+++ b/chrome/test/data/extensions/api_test/tts_engine/network_speech_engine/test.js
@@ -0,0 +1,17 @@
+// Copyright 2013 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+// browser_tests.exe --gtest_filter="TtsApiTest.*"
+
+chrome.test.runTests([
+  function testNetworkSpeechVoices() {
+    chrome.tts.getVoices(function(voices) {
+      chrome.test.assertTrue(voices.length >= 6);
+      for (var i = 0; i < voices.length; i++) {
+        chrome.test.assertEq(true, voices[i].remote);
+      }
+      chrome.test.succeed();
+    });
+  }
+]);
diff --git a/extensions/common/manifest_constants.cc b/extensions/common/manifest_constants.cc
index 962ff47..b15db19 100644
--- a/extensions/common/manifest_constants.cc
+++ b/extensions/common/manifest_constants.cc
@@ -146,6 +146,7 @@ const char kTtsVoicesEventTypeWord[] = "word";
 const char kTtsVoicesEventTypes[] = "event_types";
 const char kTtsVoicesGender[] = "gender";
 const char kTtsVoicesLang[] = "lang";
+const char kTtsVoicesRemote[] = "remote";
 const char kTtsVoicesVoiceName[] = "voice_name";
 const char kType[] = "type";
 const char kUpdateURL[] = "update_url";
@@ -573,6 +574,8 @@ const char kInvalidTtsVoicesGender[] =
     "Invalid value for 'tts_engine.voices[*].gender'.";
 const char kInvalidTtsVoicesLang[] =
     "Invalid value for 'tts_engine.voices[*].lang'.";
+const char kInvalidTtsVoicesRemote[] =
+    "Invalid value for 'tts_engine.voices[*].remote'.";
 const char kInvalidTtsVoicesVoiceName[] =
     "Invalid value for 'tts_engine.voices[*].voice_name'.";
 const char kInvalidUpdateURL[] =
diff --git a/extensions/common/manifest_constants.h b/extensions/common/manifest_constants.h
index 9b3fe9b..f645477 100644
--- a/extensions/common/manifest_constants.h
+++ b/extensions/common/manifest_constants.h
@@ -154,6 +154,7 @@ extern const char kTtsVoicesEventTypeWord[];
 extern const char kTtsVoicesEventTypes[];
 extern const char kTtsVoicesGender[];
 extern const char kTtsVoicesLang[];
+extern const char kTtsVoicesRemote[];
 extern const char kTtsVoicesVoiceName[];
 extern const char kType[];
 extern const char kUpdateURL[];
@@ -398,6 +399,7 @@ extern const char kInvalidTtsVoices[];
 extern const char kInvalidTtsVoicesEventTypes[];
 extern const char kInvalidTtsVoicesGender[];
 extern const char kInvalidTtsVoicesLang[];
+extern const char kInvalidTtsVoicesRemote[];
 extern const char kInvalidTtsVoicesVoiceName[];
 extern const char kInvalidUpdateURL[];
 extern const char kInvalidURLPatternError[];