Refactoring of chrome speech recognition architecture (CL1.3)

content/browser/speech/speech_recognizer_impl.h

 // Copyright (c) 2012 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.
 
 #ifndef CONTENT_BROWSER_SPEECH_SPEECH_RECOGNIZER_IMPL_H_
 #define CONTENT_BROWSER_SPEECH_SPEECH_RECOGNIZER_IMPL_H_
 
-#include <list>
-#include <utility>
-
-#include "base/compiler_specific.h"
+#include "base/basictypes.h"
+#include "base/memory/ref_counted.h"
 #include "base/memory/scoped_ptr.h"
-#include "content/browser/speech/audio_encoder.h"
 #include "content/browser/speech/endpointer/endpointer.h"
-#include "content/browser/speech/speech_recognition_request.h"
+#include "content/browser/speech/speech_recognition_engine.h"
 #include "content/public/browser/speech_recognizer.h"
 #include "content/public/common/speech_recognition_result.h"
 #include "media/audio/audio_input_controller.h"
+#include "net/url_request/url_request_context_getter.h"
 
-class AudioManager;
+namespace media {
+class  AudioInputController;
+}
 
 namespace content {
+struct SpeechRecognitionError;
 class SpeechRecognitionEventListener;
 }
 
 namespace speech {
+// TODO(primiano) Current CL: check CONTENT_EXPORT and NON_EXPORTED_BASE

[hans, 2012/03/16 11:12:56]

should this be addressed before this cl is landed?

[Primiano Tucci (use gerrit), 2012/03/16 15:03:42]

Done.

+// TODO(primiano) Next CL: Remove the Impl suffix and the exported
+// /content/public/browser/speech_recognizer.h interface since this class should
+// not be visible outside (currently we need it for speech input extension API).
 
-// Records audio, sends recorded audio to server and translates server response
-// to recognition result.
+// Handles speech recognition for a session (identified by |caller_id|), taking
+// care of audio capture, silence detection/endpointer and interaction with the
+// SpeechRecognitionEngine (that must be injected through set_delegate(...)).

[hans, 2012/03/16 11:12:56]

i'm confused by the last bit here.. "SpeechRecognitionEngine (that must be
injected through set_delegate)"

isn't the SpeechRecognitionEngine passed into the constructor?

[Primiano Tucci (use gerrit), 2012/03/16 15:03:42]

Uh, definitely right. Probably it was just a wrong copy/paste.

 class CONTENT_EXPORT SpeechRecognizerImpl
     : NON_EXPORTED_BASE(public content::SpeechRecognizer),
       public media::AudioInputController::EventHandler,
-      public SpeechRecognitionRequestDelegate {
+      public SpeechRecognitionEngineDelegate {
  public:
   static const int kAudioSampleRate;
-  static const int kAudioPacketIntervalMs;  // Duration of each audio packet.
   static const ChannelLayout kChannelLayout;
   static const int kNumBitsPerAudioSample;
-  static const int kNoSpeechTimeoutSec;
+  static const int kNoSpeechTimeoutMs;
   static const int kEndpointerEstimationTimeMs;
 
-  SpeechRecognizerImpl(content::SpeechRecognitionEventListener* listener,
-                       int caller_id,
-                       const std::string& language,
-                       const std::string& grammar,
-                       net::URLRequestContextGetter* context_getter,
-                       bool filter_profanities,
-                       const std::string& hardware_info,
-                       const std::string& origin_url);
-
+  SpeechRecognizerImpl(
+      content::SpeechRecognitionEventListener* listener,
+      int caller_id,
+      SpeechRecognitionEngine* engine);
   virtual ~SpeechRecognizerImpl();
 
   // content::SpeechRecognizer methods.
-  virtual bool StartRecognition() OVERRIDE;
+  virtual void StartRecognition() OVERRIDE;
   virtual void AbortRecognition() OVERRIDE;
   virtual void StopAudioCapture() OVERRIDE;
   virtual bool IsActive() const OVERRIDE;
   virtual bool IsCapturingAudio() const OVERRIDE;
+  const SpeechRecognitionEngine& recognition_engine() const;
+
+ private:
+  friend class SpeechRecognizerImplTest;
+
+  enum FSMState {
+    kIdle = 0,
+    kStartingRecognition,
+    kEstimatingEnvironment,
+    kWaitingForSpeech,
+    kRecognizingSpeech,
+    kWaitingFinalResult,
+    kMaxState = kWaitingFinalResult
+  };
+
+  enum FSMEvent {
+    kAbortRequest = 0,
+    kStartRequest,
+    kStopCaptureRequest,
+    kAudioData,
+    kRecognitionResult,
+    kRecognitionError,
+    kAudioError,
+    kMaxEvent = kAudioError
+  };
+
+  struct FSMEventArgs {
+    int audio_error_code;
+    AudioChunk* audio_data;
+    content::SpeechRecognitionResult speech_result;
+    content::SpeechRecognitionError error;
+    FSMEventArgs();
+  };
 
   // AudioInputController::EventHandler methods.
   virtual void OnCreated(media::AudioInputController* controller) OVERRIDE {}
   virtual void OnRecording(media::AudioInputController* controller) OVERRIDE {}
   virtual void OnError(media::AudioInputController* controller,
                        int error_code) OVERRIDE;
   virtual void OnData(media::AudioInputController* controller,
-                      const uint8* data,
-                      uint32 size) OVERRIDE;
+                      const uint8* data, uint32 size) OVERRIDE;
 
-  // SpeechRecognitionRequest::Delegate methods.
-  virtual void SetRecognitionResult(
+  // SpeechRecognitionEngineDelegate methods.
+  virtual void OnSpeechEngineResult(
       const content::SpeechRecognitionResult& result) OVERRIDE;
+  virtual void OnSpeechEngineError(
+      const content::SpeechRecognitionError& error) OVERRIDE;
 
- private:
-  friend class SpeechRecognizerImplTest;
-
-  void InformErrorAndAbortRecognition(
-      content::SpeechRecognitionErrorCode error);
-  void SendRecordedAudioToServer();
-
-  void HandleOnError(int error_code);  // Handles OnError in the IO thread.
-
-  // Handles OnData in the IO thread. Takes ownership of |raw_audio|.
-  void HandleOnData(AudioChunk* raw_audio);
-
-  // Helper method which closes the audio controller and blocks until done.
+  void DispatchEvent(FSMEvent event, FSMEventArgs);
+  void ProcessAudioPipeline();
+  FSMState ProcessEvent(FSMEvent event);
+  FSMState InitializeAndStartRecording();
+  FSMState StartSpeechRecognition();
+  FSMState EnvironmentEstimation();
+  FSMState DetectUserSpeechOrTimeout();
+  FSMState StopCaptureAndWaitForResult();
+  FSMState ProcessIntermediateRecognitionResult();
+  FSMState ProcessFinalRecognitionResult();
+  FSMState Abort();
+  FSMState Abort(const content::SpeechRecognitionError& error);
+  FSMState Abort(bool has_error, const content::SpeechRecognitionError& error);
+  FSMState DetectEndOfSpeech();
+  FSMState DoNothing() const;
+  int GetElapsedTimeMs() const;
+  void UpdateSignalAndNoiseLevels(const float& rms);
   void CloseAudioControllerSynchronously();
-
   void SetAudioManagerForTesting(AudioManager* audio_manager);
 
   content::SpeechRecognitionEventListener* listener_;
+  AudioManager* testing_audio_manager_;
+  scoped_ptr<SpeechRecognitionEngine> recognition_engine_;
+  Endpointer endpointer_;
+  scoped_refptr<media::AudioInputController> audio_controller_;
   int caller_id_;
-  std::string language_;
-  std::string grammar_;
-  bool filter_profanities_;
-  std::string hardware_info_;
-  std::string origin_url_;
-
-  scoped_ptr<SpeechRecognitionRequest> request_;
-  scoped_refptr<media::AudioInputController> audio_controller_;
-  scoped_refptr<net::URLRequestContextGetter> context_getter_;
-  AudioEncoder::Codec codec_;
-  scoped_ptr<AudioEncoder> encoder_;
-  Endpointer endpointer_;
   int num_samples_recorded_;
+  bool clipper_detected_clip_;
   float audio_level_;
-  AudioManager* audio_manager_;
+  float rms_;
+  int event_dispatch_nesting_level_;
+  FSMState state_;
+  FSMEvent event_;
+  FSMEventArgs* event_args_;
 
   DISALLOW_COPY_AND_ASSIGN(SpeechRecognizerImpl);
 };
 
 }  // namespace speech
 
 #endif  // CONTENT_BROWSER_SPEECH_SPEECH_RECOGNIZER_IMPL_H_