Support for providing media stream audio track to speech recognition

Source/web/SpeechRecognitionClientProxy.cpp

Index: Source/web/SpeechRecognitionClientProxy.cpp
diff --git a/Source/web/SpeechRecognitionClientProxy.cpp b/Source/web/SpeechRecognitionClientProxy.cpp
index 479f15d5ed6cbc5bad804618516c0547c1a9c980..d8c11e0f324904cbccc1add8150c6d3d2ad2b928 100644
--- a/Source/web/SpeechRecognitionClientProxy.cpp
+++ b/Source/web/SpeechRecognitionClientProxy.cpp
@@ -33,6 +33,7 @@
 #include "modules/speech/SpeechRecognitionResult.h"
 #include "modules/speech/SpeechRecognitionResultList.h"
 #include "platform/weborigin/SecurityOrigin.h"
+#include "public/platform/WebMediaStreamTrack.h"
 #include "public/web/WebSecurityOrigin.h"
 #include "public/web/WebSpeechGrammar.h"
 #include "public/web/WebSpeechRecognitionHandle.h"
@@ -52,14 +53,17 @@ PassOwnPtr<SpeechRecognitionClientProxy> SpeechRecognitionClientProxy::create(We
     return adoptPtr(new SpeechRecognitionClientProxy(recognizer));
 }
 
-void SpeechRecognitionClientProxy::start(SpeechRecognition* recognition, const SpeechGrammarList* grammarList, const String& lang, bool continuous, bool interimResults, unsigned long maxAlternatives)
+void SpeechRecognitionClientProxy::start(SpeechRecognition* recognition, const SpeechGrammarList* grammarList, const String& lang, bool continuous, bool interimResults, unsigned long maxAlternatives, MediaStreamTrack* audioTrack)
 {
     WebVector<WebSpeechGrammar> webSpeechGrammars(static_cast<size_t>(grammarList->length()));
     for (unsigned long i = 0; i < grammarList->length(); ++i)
         webSpeechGrammars[i] = grammarList->item(i);
 
     WebSpeechRecognitionParams params(webSpeechGrammars, lang, continuous, interimResults, maxAlternatives, WebSecurityOrigin(recognition->executionContext()->securityOrigin()));
-    m_recognizer->start(WebSpeechRecognitionHandle(recognition), params, this);
+    if (audioTrack)
+        m_recognizer->start(WebSpeechRecognitionHandle(recognition), params, audioTrack->component(), this);

[Mike West, 2014/09/17 10:20:27]

Could the new MediaStreamComponent argument instead be added as a property on
the WebSpeechRecognitionParams object? That would avoid the requirement to
overload '::start()' here.

[burnik, 2014/09/17 11:17:20]

Good point as well. Although that would require more changes to align with those
params. Also, notice that most of those params are actually related to the
remote API and have nothing to do with audio.

On 2014/09/17 10:20:27, Mike West wrote:

[Mike West, 2014/09/17 11:30:46]

Sure. But I think it reduces the complexity of the system by ensuring that
there's one point where you make changes to the parameters necessary to
initialize the recognizer, rather than breaking the flow up into multiple
methods.
Well, it's called WebSpeechRecognitionParams, which suggests that it contains
bits and pieces which are necessary to do speech recognition. Adding an audio
track to that object doesn't seem like it's much of a stretch.

I don't think this sort of thing is a requirement, but it would be nice to keep
the public API simpler. *shrug*

+    else
+        m_recognizer->start(WebSpeechRecognitionHandle(recognition), params, this);
 }
 
 void SpeechRecognitionClientProxy::stop(SpeechRecognition* recognition)