Added a level indicator to new mixer.

webrtc/voice_engine/level_indicator.cc

Index: webrtc/voice_engine/level_indicator.cc
diff --git a/webrtc/voice_engine/level_indicator.cc b/webrtc/voice_engine/level_indicator.cc
index f44ea8e3f0e6cca496811a3248608cf6ef4bde26..92fc8efe4c37ceafc5e12f045829c467c8d730d9 100644
--- a/webrtc/voice_engine/level_indicator.cc
+++ b/webrtc/voice_engine/level_indicator.cc
@@ -28,6 +28,7 @@ AudioLevel::AudioLevel() :
     _count(0),
     _currentLevel(0),
     _currentLevelFullRange(0) {
+  WebRtcSpl_Init();

[mflodman, 2016/08/18 09:34:13]

This has already been called in VoEBaseImpl::Init, should this really be done
here?

[aleloi, 2016/08/18 09:40:33]

I wanted to use the level indicator as a separate dependency in the new audio
mixer without having to depend on the whole of VoE. But then the mixer crashed
in unit tests, because the SPL functions (which this component uses) were not
initialized.

My understanding is that if this component is going to be a separate build
target, it should initialize its dependencies by itself.

[mflodman, 2016/08/18 09:45:28]

But is it ok to do this twice in the case you run this in VoiceEngine? Should
this rather be done in the test itself?

This might be right, but I don't know how SPLib initialization is supposed to
work and I think it's better if someone in the audio team looks at this then me.

[minyue-webrtc, 2016/08/18 09:53:08]

It reads that WebRtcSpl_Init() should be called only once (see below). so we
should not go wrong with it. But true, it should be called on a high level. If
possible, do it in the test?

https://cs.chromium.org/chromium/src/third_party/webrtc/common_audio/signal_p...

[kwiberg-webrtc, 2016/08/19 11:24:44]

No, it's implemented so that it should be safe to call multiple times, even from
several different threads.

 }
 
 AudioLevel::~AudioLevel() {