2 files changed, 40 insertions, 7 deletions

diff --git a/dom/media/webspeech/recognition/SpeechRecognition.cpp b/dom/media/webspeech/recognition/SpeechRecognition.cpp
index 7239a88237..44ad5fdd61 100644
--- a/dom/media/webspeech/recognition/SpeechRecognition.cpp
+++ b/dom/media/webspeech/recognition/SpeechRecognition.cpp
@@ -136,7 +136,7 @@ CreateSpeechRecognitionService(nsPIDOMWindowInner* aWindow,
 NS_IMPL_CYCLE_COLLECTION_WEAK_PTR_INHERITED(SpeechRecognition,
                                             DOMEventTargetHelper, mStream,
                                             mTrack, mRecognitionService,
-                                            mSpeechGrammarList)
+                                            mSpeechGrammarList, mListener)
 
 NS_INTERFACE_MAP_BEGIN_CYCLE_COLLECTION(SpeechRecognition)
   NS_INTERFACE_MAP_ENTRY(nsIObserver)
@@ -145,6 +145,16 @@ NS_INTERFACE_MAP_END_INHERITING(DOMEventTargetHelper)
 NS_IMPL_ADDREF_INHERITED(SpeechRecognition, DOMEventTargetHelper)
 NS_IMPL_RELEASE_INHERITED(SpeechRecognition, DOMEventTargetHelper)
 
+NS_IMPL_CYCLE_COLLECTION_INHERITED(SpeechRecognition::TrackListener,
+                                   DOMMediaStream::TrackListener,
+                                   mSpeechRecognition)
+NS_IMPL_ADDREF_INHERITED(SpeechRecognition::TrackListener,
+                         DOMMediaStream::TrackListener)
+NS_IMPL_RELEASE_INHERITED(SpeechRecognition::TrackListener,
+                          DOMMediaStream::TrackListener)
+NS_INTERFACE_MAP_BEGIN_CYCLE_COLLECTION(SpeechRecognition::TrackListener)
+NS_INTERFACE_MAP_END_INHERITING(DOMMediaStream::TrackListener)
+
 SpeechRecognition::SpeechRecognition(nsPIDOMWindowInner* aOwnerWindow)
     : DOMEventTargetHelper(aOwnerWindow),
       mEndpointer(kSAMPLE_RATE),
@@ -472,8 +482,9 @@ void SpeechRecognition::Reset() {
 
   ++mStreamGeneration;
   if (mStream) {
-    mStream->UnregisterTrackListener(this);
+    mStream->UnregisterTrackListener(mListener);
     mStream = nullptr;
+    mListener = nullptr;
   }
   mTrack = nullptr;
   mTrackIsOwned = false;
@@ -642,7 +653,8 @@ RefPtr<GenericNonExclusivePromise> SpeechRecognition::StopRecording() {
     if (mStream) {
       // Ensure we don't start recording because a track became available
       // before we get reset.
-      mStream->UnregisterTrackListener(this);
+      mStream->UnregisterTrackListener(mListener);
+      mListener = nullptr;
     }
     return GenericNonExclusivePromise::CreateAndResolve(true, __func__);
   }
@@ -801,10 +813,13 @@ void SpeechRecognition::Start(const Optional<NonNull<DOMMediaStream>>& aStream,
   MediaStreamConstraints constraints;
   constraints.mAudio.SetAsBoolean() = true;
 
+  MOZ_ASSERT(!mListener);
+  mListener = new TrackListener(this);
+
   if (aStream.WasPassed()) {
     mStream = &aStream.Value();
     mTrackIsOwned = false;
-    mStream->RegisterTrackListener(this);
+    mStream->RegisterTrackListener(mListener);
     nsTArray<RefPtr<AudioStreamTrack>> tracks;
     mStream->GetAudioTracks(tracks);
     for (const RefPtr<AudioStreamTrack>& track : tracks) {
@@ -839,7 +854,7 @@ void SpeechRecognition::Start(const Optional<NonNull<DOMMediaStream>>& aStream,
                 return;
               }
               mStream = std::move(aStream);
-              mStream->RegisterTrackListener(this);
+              mStream->RegisterTrackListener(mListener);
               for (const RefPtr<AudioStreamTrack>& track : tracks) {
                 if (!track->Ended()) {
                   NotifyTrackAdded(track);
diff --git a/dom/media/webspeech/recognition/SpeechRecognition.h b/dom/media/webspeech/recognition/SpeechRecognition.h
index 687f38041e..465cadc8cb 100644
--- a/dom/media/webspeech/recognition/SpeechRecognition.h
+++ b/dom/media/webspeech/recognition/SpeechRecognition.h
@@ -52,7 +52,6 @@ LogModule* GetSpeechRecognitionLog();
 
 class SpeechRecognition final : public DOMEventTargetHelper,
                                 public nsIObserver,
-                                public DOMMediaStream::TrackListener,
                                 public SupportsWeakPtr {
  public:
   explicit SpeechRecognition(nsPIDOMWindowInner* aOwnerWindow);
@@ -133,7 +132,24 @@ class SpeechRecognition final : public DOMEventTargetHelper,
     EVENT_COUNT
   };
 
-  void NotifyTrackAdded(const RefPtr<MediaStreamTrack>& aTrack) override;
+  void NotifyTrackAdded(const RefPtr<MediaStreamTrack>& aTrack);
+
+  class TrackListener final : public DOMMediaStream::TrackListener {
+   public:
+    NS_DECL_ISUPPORTS_INHERITED
+    NS_DECL_CYCLE_COLLECTION_CLASS_INHERITED(TrackListener,
+                                             DOMMediaStream::TrackListener)
+    explicit TrackListener(SpeechRecognition* aSpeechRecognition)
+        : mSpeechRecognition(aSpeechRecognition) {}
+    void NotifyTrackAdded(const RefPtr<MediaStreamTrack>& aTrack) override {
+      mSpeechRecognition->NotifyTrackAdded(aTrack);
+    }
+
+   private:
+    virtual ~TrackListener() = default;
+    RefPtr<SpeechRecognition> mSpeechRecognition;
+  };
+
   // aMessage should be valid UTF-8, but invalid UTF-8 byte sequences are
   // replaced with the REPLACEMENT CHARACTER on conversion to UTF-16.
   void DispatchError(EventType aErrorType,
@@ -266,6 +282,8 @@ class SpeechRecognition final : public DOMEventTargetHelper,
   // a conforming implementation.
   uint32_t mMaxAlternatives;
 
+  RefPtr<TrackListener> mListener;
+
   void ProcessTestEventRequest(nsISupports* aSubject,
                                const nsAString& aEventName);