Added a level indicator to new mixer.

webrtc/modules/audio_mixer/test/audio_mixer_unittest.cc

Index: webrtc/modules/audio_mixer/test/audio_mixer_unittest.cc
diff --git a/webrtc/modules/audio_mixer/test/audio_mixer_unittest.cc b/webrtc/modules/audio_mixer/test/audio_mixer_unittest.cc
index 38f2b2ebff569981ea4b902f08748c9c316ab957..710fb3cca00da49755dfb9fe696e648576eecb73 100644
--- a/webrtc/modules/audio_mixer/test/audio_mixer_unittest.cc
+++ b/webrtc/modules/audio_mixer/test/audio_mixer_unittest.cc
@@ -370,6 +370,65 @@ TEST(AudioMixer, ParticipantNumberOfChannels) {
   }
 }
 
+// Test that the volume is reported as zero when the mixer input is

[minyue-webrtc, 2016/08/11 10:27:33]

is -> comprises only

[aleloi, 2016/08/11 12:15:20]

Done.

+// frames with no sound.

[minyue-webrtc, 2016/08/11 10:27:33]

no sound -> zero values

[aleloi, 2016/08/11 12:15:20]

Done.

+TEST(AudioMixer, ZeroVolume) {

[minyue-webrtc, 2016/08/11 10:27:33]

ZeroVolume -> LevelIsZeroWhenMixingZeros

[aleloi, 2016/08/11 12:15:20]

Done.

+  const int kId = 1;
+  std::unique_ptr<NewAudioConferenceMixer> mixer(
+      NewAudioConferenceMixer::Create(kId));
+  AudioFrame frame_for_mixing;
+
+  MockMixerAudioSource participant;
+  participant.fake_frame()->sample_rate_hz_ = 8000;

[minyue-webrtc, 2016/08/11 10:27:33]

define kSampleRate and use it to replace 8000

[aleloi, 2016/08/11 12:15:19]

Done.

[aleloi, 2016/08/11 12:15:19]

Done.

+  participant.fake_frame()->num_channels_ = 1;
+
+  // Frame duration 10ms.
+  participant.fake_frame()->samples_per_channel_ = 8000 / 100;
+
+  EXPECT_EQ(0, mixer->SetMixabilityStatus(&participant, true));
+  for (size_t i = 0; i < 10; i++) {
+    EXPECT_CALL(participant, GetAudioFrameWithMuted(_, 8000)).Times(Exactly(1));
+    mixer->Mix(8000, 1, &frame_for_mixing);
+
+    EXPECT_EQ(static_cast<uint32_t>(0), mixer->GetSpeechOutputLevel());
+    EXPECT_EQ(static_cast<uint32_t>(0), mixer->GetSpeechOutputLevelFullRange());
+  }
+}
+
+// Test that the volume is reported as full when the mixer input is

[minyue-webrtc, 2016/08/11 10:27:33]

comprises frames with max values of int16_t.

[minyue-webrtc, 2016/08/11 10:27:33]

maximum

[aleloi, 2016/08/11 12:15:20]

Done.

[aleloi, 2016/08/11 12:15:20]

Done.

+// full frames.
+TEST(AudioMixer, FullVolume) {

[minyue-webrtc, 2016/08/11 10:27:33]

curious, what if one participant is zero and another is max

[minyue-webrtc, 2016/08/11 10:27:33]

FullVolume -> LevelIsMaxWhenMixingMaxValues

[aleloi, 2016/08/11 12:15:20]

Thanks! I thought about how to name the tests, but I only looked at my other
tests for inspiration (which would also benefit from better names).

[aleloi, 2016/08/11 12:15:20]

It should be max, or one off from max due to rounding and limiting. Do you think
it's meaningful to test?

[aleloi, 2016/08/11 12:25:22]

Thinking on it again, I see that I was probably wrong. Since there are two
participants, the limiter kicks in.

We first reduce the volume by shifting down both frames, so the max frame is now
full of 0x1fff and the other is zero.

Then we tell the limiter to limit to -7dBFS, which is just below half max
volume. The frame should get a little more quiet. Then we multiply it by two, so
we get almost the full volume back. I will run a test to see if this is correct!

+  const int kId = 1;
+  std::unique_ptr<NewAudioConferenceMixer> mixer(
+      NewAudioConferenceMixer::Create(kId));
+  AudioFrame frame_for_mixing;
+
+  MockMixerAudioSource participant;
+  participant.fake_frame()->sample_rate_hz_ = 8000;
+  participant.fake_frame()->num_channels_ = 1;
+
+  // Frame duration 10ms.
+  participant.fake_frame()->samples_per_channel_ = 8000 / 100;
+
+  // Fill participant frame data with maximal sound.
+  std::fill(participant.fake_frame()->data_,
+            participant.fake_frame()->data_ + 8000 / 100,
+            std::numeric_limits<int16_t>::max());
+
+  EXPECT_EQ(0, mixer->SetMixabilityStatus(&participant, true));
+  for (size_t i = 0; i < 10; i++) {
+    EXPECT_CALL(participant, GetAudioFrameWithMuted(_, 8000)).Times(Exactly(1));
+    mixer->Mix(8000, 1, &frame_for_mixing);
+
+    // 9 is the highest possible audio level
+    EXPECT_EQ(static_cast<uint32_t>(9), mixer->GetSpeechOutputLevel());
+
+    // 0x7fff = 32767 is the highest full range audio level.
+    EXPECT_EQ(static_cast<uint32_t>(std::numeric_limits<int16_t>::max()),
+              mixer->GetSpeechOutputLevelFullRange());
+  }
+}
+
 TEST_F(BothMixersTest, CompareInitialFrameAudio) {
   EXPECT_CALL(participant_, GetAudioFrameWithMuted(_, _)).Times(Exactly(1));