Surface the IntelligibilityEnhancer on MediaConstraints

webrtc/media/engine/webrtcvoiceengine.cc

Index: webrtc/media/engine/webrtcvoiceengine.cc
diff --git a/webrtc/media/engine/webrtcvoiceengine.cc b/webrtc/media/engine/webrtcvoiceengine.cc
index ed67ce84ba00cc520872b1a17e0ea8bdbb579160..833e50651604918b46821ff394cbe81563f25f2d 100644
--- a/webrtc/media/engine/webrtcvoiceengine.cc
+++ b/webrtc/media/engine/webrtcvoiceengine.cc
@@ -577,6 +577,7 @@ WebRtcVoiceEngine::WebRtcVoiceEngine(webrtc::AudioDeviceModule* adm,
     options.extended_filter_aec = rtc::Optional<bool>(false);
     options.delay_agnostic_aec = rtc::Optional<bool>(false);
     options.experimental_ns = rtc::Optional<bool>(false);
+    options.intelligibility_enhancer = rtc::Optional<bool>(false);
     bool error = ApplyOptions(options);
     RTC_DCHECK(error);
   }
@@ -843,6 +844,16 @@ bool WebRtcVoiceEngine::ApplyOptions(const AudioOptions& options_in) {
         new webrtc::ExperimentalNs(*experimental_ns_));
   }
 
+  if (options.intelligibility_enhancer) {

[henrika_webrtc, 2016/05/11 08:50:36]

You have access to the ADM here. Why can't you use it to disable the HW NS.
Like so: adm()->EnableBuiltInNS()

[aluebs-webrtc, 2016/05/11 21:22:05]

Great point! I didn't realize there was a 2 way communication of these flags
with the Java layer. Done.

+    intelligibility_enhancer_ = options.intelligibility_enhancer;
+  }
+  if (intelligibility_enhancer_) {
+    LOG(LS_INFO) << "Intelligibility Enhancer is enabled? "
+                 << *intelligibility_enhancer_;
+    config.Set<webrtc::Intelligibility>(
+        new webrtc::Intelligibility(*intelligibility_enhancer_));

[the sun, 2016/05/11 11:23:41]

nit: the call to
adm()->EnableBuiltInNS(false)
should be here.

Also, note that on Android, the hardware effects settings are only applied when
adm()->InitRecord() is called, which will happen as soon as we know there is a
stream to be sent. I believe you should be fine though, since constraints are
passed when creating and audio source and applied when a send stream is added,
but you may want to keep an eye on the hardware NS and verify it isn't left
enabled for some reason.

[aluebs-webrtc, 2016/05/11 21:22:05]

Actually I think the best thing is to do it above together with the NS, so that
we don't disable the SWNS when IE is enabled.

+  }
+
   // We check audioproc for the benefit of tests, since FakeWebRtcVoiceEngine
   // returns NULL on audio_processing().
   webrtc::AudioProcessing* audioproc = voe_wrapper_->base()->audio_processing();