Switch speech recognition to using audio service input streams.

Use audio service IPC instead of AudioInputController for creating input stream.

This change is part of the effort of moving audio out of browser process [1]. By moving audio
into a separate process, speech recognition can no longer use AudioInputController directly,
and needs to use IPC to create and use input streams provided by audio service.

[1] Chrome audio process design doc: http://doc/12vz7UZBQIp1195V-xAYvHIoQ1r536BCmMKcPN7Si-OI

Bug: 834706
Cq-Include-Trybots: luci.chromium.try:android_optional_gpu_tests_rel;luci.chromium.try:linux_optional_gpu_tests_rel;luci.chromium.try:mac_optional_gpu_tests_rel;luci.chromium.try:win_optional_gpu_tests_rel
Change-Id: I1dc98ad670e1e9880320ae4bf054ae0d94f5b48f
Reviewed-on: https://chromium-review.googlesource.com/1049685
Commit-Queue: Marina Ciocea <marinaciocea@chromium.org>
Reviewed-by: Nasko Oskov <nasko@chromium.org>
Reviewed-by: Olga Sharonova <olka@chromium.org>
Cr-Commit-Position: refs/heads/master@{#557701}

content/browser/browser_main_loop.cc

@@ -1333,8 +1333,7 @@ int BrowserMainLoop::BrowserThreadsStarted() {
     TRACE_EVENT0("startup",
       "BrowserMainLoop::BrowserThreadsStarted:InitSpeechRecognition");
     speech_recognition_manager_.reset(new SpeechRecognitionManagerImpl(
-        audio_system_.get(), audio_manager_.get(),
-        media_stream_manager_.get()));
+        audio_system_.get(), media_stream_manager_.get()));
   }
 
   {

content/browser/speech/speech_recognition_manager_impl.cc

@@ -218,10 +218,8 @@ SpeechRecognitionManagerImpl* SpeechRecognitionManagerImpl::GetInstance() {
 
 SpeechRecognitionManagerImpl::SpeechRecognitionManagerImpl(
     media::AudioSystem* audio_system,
-    media::AudioManager* audio_manager,
     MediaStreamManager* media_stream_manager)
     : audio_system_(audio_system),
-      audio_manager_(audio_manager),
       media_stream_manager_(media_stream_manager),
       primary_session_id_(kSessionIDInvalid),
       last_session_id_(kSessionIDInvalid),
@@ -287,7 +285,7 @@ int SpeechRecognitionManagerImpl::CreateSession(
   google_remote_engine->SetConfig(remote_engine_config);
 
   session->recognizer = new SpeechRecognizerImpl(
-      this, audio_system_, audio_manager_, session_id, config.continuous,
+      this, audio_system_, session_id, config.continuous,
       config.interim_results, google_remote_engine);
 #else
   session->recognizer = new SpeechRecognizerImplAndroid(this, session_id);

content/browser/speech/speech_recognition_manager_impl.h

@@ -20,7 +20,6 @@
 
 namespace media {
 class AudioSystem;
-class AudioManager;
 }
 
 namespace content {
@@ -99,7 +98,6 @@ class CONTENT_EXPORT SpeechRecognitionManagerImpl
   friend class base::DeleteHelper<content::SpeechRecognitionManagerImpl>;
 
   SpeechRecognitionManagerImpl(media::AudioSystem* audio_system,
-                               media::AudioManager* audio_manager,
                                MediaStreamManager* media_stream_manager);
   ~SpeechRecognitionManagerImpl() override;
 
@@ -183,7 +181,6 @@ class CONTENT_EXPORT SpeechRecognitionManagerImpl
       frame_deletion_observer_;
 
   media::AudioSystem* audio_system_;
-  media::AudioManager* audio_manager_;
   MediaStreamManager* media_stream_manager_;
   base::flat_map<int, std::unique_ptr<Session>> sessions_;
   int primary_session_id_;

content/browser/speech/speech_recognizer_impl.h

@@ -16,8 +16,7 @@
 #include "content/browser/speech/speech_recognizer.h"
 #include "content/public/common/speech_recognition_error.h"
 #include "content/public/common/speech_recognition_result.h"
-#include "media/audio/audio_input_controller.h"
-#include "media/mojo/interfaces/audio_logging.mojom.h"
+#include "media/base/audio_capturer_source.h"
 #include "net/url_request/url_request_context_getter.h"
 
 namespace media {
@@ -34,8 +33,7 @@ class SpeechRecognitionEventListener;
 // SpeechRecognitionEngine.
 class CONTENT_EXPORT SpeechRecognizerImpl
     : public SpeechRecognizer,
-      public media::AudioInputController::EventHandler,
-      public media::AudioInputController::SyncWriter,
+      public media::AudioCapturerSource::CaptureCallback,
       public SpeechRecognitionEngine::Delegate {
  public:
   static const int kAudioSampleRate;
@@ -44,22 +42,24 @@ class CONTENT_EXPORT SpeechRecognizerImpl
   static const int kNoSpeechTimeoutMs;
   static const int kEndpointerEstimationTimeMs;
 
-  static void SetAudioEnvironmentForTesting(media::AudioSystem* audio_system,
-                                            media::AudioManager* audio_manager);
+  static void SetAudioEnvironmentForTesting(
+      media::AudioSystem* audio_system,
+      media::AudioCapturerSource* capturer_source);
 
   SpeechRecognizerImpl(SpeechRecognitionEventListener* listener,
                        media::AudioSystem* audio_system,
-                       media::AudioManager* audio_manager,
                        int session_id,
                        bool continuous,
                        bool provisional_results,
                        SpeechRecognitionEngine* engine);
 
+  // SpeechRecognizer methods.
   void StartRecognition(const std::string& device_id) override;
   void AbortRecognition() override;
   void StopAudioCapture() override;
   bool IsActive() const override;
   bool IsCapturingAudio() const override;
+
   const SpeechRecognitionEngine& recognition_engine() const;
 
  private:
@@ -138,23 +138,16 @@ class CONTENT_EXPORT SpeechRecognizerImpl
   // OnAudioLevelsChange event accordingly.
   void UpdateSignalAndNoiseLevels(const float& rms, bool clip_detected);
 
-  void CloseAudioControllerAsynchronously();
-
-  // Callback called on IO thread by audio_controller->Close().
-  void OnAudioClosed(media::AudioInputController*);
+  void CloseAudioCapturerSource();
 
-  // AudioInputController::EventHandler methods.
-  void OnCreated(bool initially_muted) override {}
-  void OnError(media::AudioInputController::ErrorCode error_code) override;
-  void OnLog(base::StringPiece) override {}
-  void OnMuted(bool is_muted) override {}
-
-  // AudioInputController::SyncWriter methods.
-  void Write(const media::AudioBus* data,
-             double volume,
-             bool key_pressed,
-             base::TimeTicks capture_time) override;
-  void Close() override;
+  // media::AudioCapturerSource::CaptureCallback methods.
+  void OnCaptureStarted() final {}
+  void Capture(const media::AudioBus* audio_bus,
+               int audio_delay_milliseconds,
+               double volume,
+               bool key_pressed) final;
+  void OnCaptureError(const std::string& message) final;
+  void OnCaptureMuted(bool is_muted) final {}
 
   // SpeechRecognitionEngineDelegate methods.
   void OnSpeechRecognitionEngineResults(
@@ -164,17 +157,17 @@ class CONTENT_EXPORT SpeechRecognizerImpl
       const SpeechRecognitionError& error) override;
 
   media::AudioSystem* GetAudioSystem();
-  media::AudioManager* GetAudioManager();
+  void CreateAudioCapturerSource();
+  media::AudioCapturerSource* GetAudioCapturerSource();
 
-  // Substitutes the real audio system in browser tests.
+  // Substitute the real audio system and capturer source in browser tests.
   static media::AudioSystem* audio_system_for_tests_;
-  static media::AudioManager* audio_manager_for_tests_;
+  static media::AudioCapturerSource* audio_capturer_source_for_tests_;
+
   media::AudioSystem* audio_system_;
-  media::AudioManager* audio_manager_;
   std::unique_ptr<SpeechRecognitionEngine> recognition_engine_;
   Endpointer endpointer_;
-  scoped_refptr<media::AudioInputController> audio_controller_;
-  media::mojom::AudioLogPtr audio_log_;
+  scoped_refptr<media::AudioCapturerSource> audio_capturer_source_;
   int num_samples_recorded_;
   float audio_level_;
   bool is_dispatching_event_;