Re-Land: Implement AudioReceiveStream::GetStats().

R=tommi@webrtc.org
BUG=webrtc:4690

Committed: https://chromium.googlesource.com/external/webrtc/+/a457752f4afc496ed7f4d6b584b08d8635f18cc0

Review URL: https://codereview.webrtc.org/1390753002 .

Cr-Commit-Position: refs/heads/master@{#10369}
diff --git a/webrtc/audio/BUILD.gn b/webrtc/audio/BUILD.gn
index c6f4b6b..d5061db 100644
--- a/webrtc/audio/BUILD.gn
+++ b/webrtc/audio/BUILD.gn
@@ -14,6 +14,8 @@
     "audio_receive_stream.h",
     "audio_send_stream.cc",
     "audio_send_stream.h",
+    "conversion.h",
+    "scoped_voe_interface.h",
   ]
 
   configs += [ "..:common_config" ]
diff --git a/webrtc/audio/audio_receive_stream.cc b/webrtc/audio/audio_receive_stream.cc
index c725e37..0fd96d0 100644
--- a/webrtc/audio/audio_receive_stream.cc
+++ b/webrtc/audio/audio_receive_stream.cc
@@ -12,10 +12,17 @@
 
 #include <string>
 
+#include "webrtc/audio/conversion.h"
 #include "webrtc/base/checks.h"
 #include "webrtc/base/logging.h"
 #include "webrtc/modules/remote_bitrate_estimator/include/remote_bitrate_estimator.h"
 #include "webrtc/system_wrappers/interface/tick_util.h"
+#include "webrtc/voice_engine/include/voe_base.h"
+#include "webrtc/voice_engine/include/voe_codec.h"
+#include "webrtc/voice_engine/include/voe_neteq_stats.h"
+#include "webrtc/voice_engine/include/voe_rtp_rtcp.h"
+#include "webrtc/voice_engine/include/voe_video_sync.h"
+#include "webrtc/voice_engine/include/voe_volume_control.h"
 
 namespace webrtc {
 std::string AudioReceiveStream::Config::Rtp::ToString() const {
@@ -24,8 +31,9 @@
   ss << ", extensions: [";
   for (size_t i = 0; i < extensions.size(); ++i) {
     ss << extensions[i].ToString();
-    if (i != extensions.size() - 1)
+    if (i != extensions.size() - 1) {
       ss << ", ";
+    }
   }
   ss << ']';
   ss << '}';
@@ -36,8 +44,9 @@
   std::stringstream ss;
   ss << "{rtp: " << rtp.ToString();
   ss << ", voe_channel_id: " << voe_channel_id;
-  if (!sync_group.empty())
+  if (!sync_group.empty()) {
     ss << ", sync_group: " << sync_group;
+  }
   ss << '}';
   return ss.str();
 }
@@ -45,13 +54,18 @@
 namespace internal {
 AudioReceiveStream::AudioReceiveStream(
       RemoteBitrateEstimator* remote_bitrate_estimator,
-      const webrtc::AudioReceiveStream::Config& config)
+      const webrtc::AudioReceiveStream::Config& config,
+      VoiceEngine* voice_engine)
     : remote_bitrate_estimator_(remote_bitrate_estimator),
       config_(config),
+      voice_engine_(voice_engine),
+      voe_base_(voice_engine),
       rtp_header_parser_(RtpHeaderParser::Create()) {
+  RTC_DCHECK(thread_checker_.CalledOnValidThread());
   LOG(LS_INFO) << "AudioReceiveStream: " << config_.ToString();
   RTC_DCHECK(config.voe_channel_id != -1);
   RTC_DCHECK(remote_bitrate_estimator_ != nullptr);
+  RTC_DCHECK(voice_engine_ != nullptr);
   RTC_DCHECK(rtp_header_parser_ != nullptr);
   for (const auto& ext : config.rtp.extensions) {
     // One-byte-extension local identifiers are in the range 1-14 inclusive.
@@ -73,33 +87,117 @@
 }
 
 AudioReceiveStream::~AudioReceiveStream() {
+  RTC_DCHECK(thread_checker_.CalledOnValidThread());
   LOG(LS_INFO) << "~AudioReceiveStream: " << config_.ToString();
 }
 
 webrtc::AudioReceiveStream::Stats AudioReceiveStream::GetStats() const {
-  return webrtc::AudioReceiveStream::Stats();
+  RTC_DCHECK(thread_checker_.CalledOnValidThread());
+  webrtc::AudioReceiveStream::Stats stats;
+  stats.remote_ssrc = config_.rtp.remote_ssrc;
+  ScopedVoEInterface<VoECodec> codec(voice_engine_);
+  ScopedVoEInterface<VoENetEqStats> neteq(voice_engine_);
+  ScopedVoEInterface<VoERTP_RTCP> rtp(voice_engine_);
+  ScopedVoEInterface<VoEVideoSync> sync(voice_engine_);
+  ScopedVoEInterface<VoEVolumeControl> volume(voice_engine_);
+  unsigned int ssrc = 0;
+  webrtc::CallStatistics cs = {0};
+  webrtc::CodecInst ci = {0};
+  // Only collect stats if we have seen some traffic with the SSRC.
+  if (rtp->GetRemoteSSRC(config_.voe_channel_id, ssrc) == -1 ||
+      rtp->GetRTCPStatistics(config_.voe_channel_id, cs) == -1 ||
+      codec->GetRecCodec(config_.voe_channel_id, ci) == -1) {
+    return stats;
+  }
+
+  stats.bytes_rcvd = cs.bytesReceived;
+  stats.packets_rcvd = cs.packetsReceived;
+  stats.packets_lost = cs.cumulativeLost;
+  stats.fraction_lost = static_cast<float>(cs.fractionLost) / (1 << 8);
+  if (ci.pltype != -1) {
+    stats.codec_name = ci.plname;
+  }
+
+  stats.ext_seqnum = cs.extendedMax;
+  if (ci.plfreq / 1000 > 0) {
+    stats.jitter_ms = cs.jitterSamples / (ci.plfreq / 1000);
+  }
+  {
+    int jitter_buffer_delay_ms = 0;
+    int playout_buffer_delay_ms = 0;
+    sync->GetDelayEstimate(config_.voe_channel_id, &jitter_buffer_delay_ms,
+                           &playout_buffer_delay_ms);
+    stats.delay_estimate_ms =
+        jitter_buffer_delay_ms + playout_buffer_delay_ms;
+  }
+  {
+    unsigned int level = 0;
+    if (volume->GetSpeechOutputLevelFullRange(config_.voe_channel_id, level)
+        != -1) {
+      stats.audio_level = static_cast<int32_t>(level);
+    }
+  }
+
+  webrtc::NetworkStatistics ns = {0};
+  if (neteq->GetNetworkStatistics(config_.voe_channel_id, ns) != -1) {
+    // Get jitter buffer and total delay (alg + jitter + playout) stats.
+    stats.jitter_buffer_ms = ns.currentBufferSize;
+    stats.jitter_buffer_preferred_ms = ns.preferredBufferSize;
+    stats.expand_rate = Q14ToFloat(ns.currentExpandRate);
+    stats.speech_expand_rate = Q14ToFloat(ns.currentSpeechExpandRate);
+    stats.secondary_decoded_rate = Q14ToFloat(ns.currentSecondaryDecodedRate);
+    stats.accelerate_rate = Q14ToFloat(ns.currentAccelerateRate);
+    stats.preemptive_expand_rate = Q14ToFloat(ns.currentPreemptiveRate);
+  }
+
+  webrtc::AudioDecodingCallStats ds;
+  if (neteq->GetDecodingCallStatistics(config_.voe_channel_id, &ds) != -1) {
+    stats.decoding_calls_to_silence_generator =
+        ds.calls_to_silence_generator;
+    stats.decoding_calls_to_neteq = ds.calls_to_neteq;
+    stats.decoding_normal = ds.decoded_normal;
+    stats.decoding_plc = ds.decoded_plc;
+    stats.decoding_cng = ds.decoded_cng;
+    stats.decoding_plc_cng = ds.decoded_plc_cng;
+  }
+
+  stats.capture_start_ntp_time_ms = cs.capture_start_ntp_time_ms_;
+
+  return stats;
 }
 
 const webrtc::AudioReceiveStream::Config& AudioReceiveStream::config() const {
+  RTC_DCHECK(thread_checker_.CalledOnValidThread());
   return config_;
 }
 
 void AudioReceiveStream::Start() {
+  RTC_DCHECK(thread_checker_.CalledOnValidThread());
 }
 
 void AudioReceiveStream::Stop() {
+  RTC_DCHECK(thread_checker_.CalledOnValidThread());
 }
 
 void AudioReceiveStream::SignalNetworkState(NetworkState state) {
+  RTC_DCHECK(thread_checker_.CalledOnValidThread());
 }
 
 bool AudioReceiveStream::DeliverRtcp(const uint8_t* packet, size_t length) {
+  // TODO(solenberg): Tests call this function on a network thread, libjingle
+  // calls on the worker thread. We should move towards always using a network
+  // thread. Then this check can be enabled.
+  // RTC_DCHECK(!thread_checker_.CalledOnValidThread());
   return false;
 }
 
 bool AudioReceiveStream::DeliverRtp(const uint8_t* packet,
                                     size_t length,
                                     const PacketTime& packet_time) {
+  // TODO(solenberg): Tests call this function on a network thread, libjingle
+  // calls on the worker thread. We should move towards always using a network
+  // thread. Then this check can be enabled.
+  // RTC_DCHECK(!thread_checker_.CalledOnValidThread());
   RTPHeader header;
 
   if (!rtp_header_parser_->Parse(packet, length, &header)) {
diff --git a/webrtc/audio/audio_receive_stream.h b/webrtc/audio/audio_receive_stream.h
index 1e52724..5c77653 100644
--- a/webrtc/audio/audio_receive_stream.h
+++ b/webrtc/audio/audio_receive_stream.h
@@ -12,18 +12,23 @@
 #define WEBRTC_AUDIO_AUDIO_RECEIVE_STREAM_H_
 
 #include "webrtc/audio_receive_stream.h"
+#include "webrtc/audio/scoped_voe_interface.h"
+#include "webrtc/base/thread_checker.h"
 #include "webrtc/modules/rtp_rtcp/interface/rtp_header_parser.h"
+#include "webrtc/voice_engine/include/voe_base.h"
 
 namespace webrtc {
 
 class RemoteBitrateEstimator;
+class VoiceEngine;
 
 namespace internal {
 
 class AudioReceiveStream : public webrtc::AudioReceiveStream {
  public:
   AudioReceiveStream(RemoteBitrateEstimator* remote_bitrate_estimator,
-                     const webrtc::AudioReceiveStream::Config& config);
+                     const webrtc::AudioReceiveStream::Config& config,
+                     VoiceEngine* voice_engine);
   ~AudioReceiveStream() override;
 
   // webrtc::ReceiveStream implementation.
@@ -41,8 +46,12 @@
   const webrtc::AudioReceiveStream::Config& config() const;
 
  private:
+  rtc::ThreadChecker thread_checker_;
   RemoteBitrateEstimator* const remote_bitrate_estimator_;
   const webrtc::AudioReceiveStream::Config config_;
+  VoiceEngine* voice_engine_;
+  // We hold one interface pointer to the VoE to make sure it is kept alive.
+  ScopedVoEInterface<VoEBase> voe_base_;
   rtc::scoped_ptr<RtpHeaderParser> rtp_header_parser_;
 };
 }  // namespace internal
diff --git a/webrtc/audio/audio_receive_stream_unittest.cc b/webrtc/audio/audio_receive_stream_unittest.cc
index d6cce69..8809b35 100644
--- a/webrtc/audio/audio_receive_stream_unittest.cc
+++ b/webrtc/audio/audio_receive_stream_unittest.cc
@@ -11,10 +11,14 @@
 #include "testing/gtest/include/gtest/gtest.h"
 
 #include "webrtc/audio/audio_receive_stream.h"
+#include "webrtc/audio/conversion.h"
 #include "webrtc/modules/remote_bitrate_estimator/include/mock/mock_remote_bitrate_estimator.h"
 #include "webrtc/modules/rtp_rtcp/source/byte_io.h"
+#include "webrtc/test/fake_voice_engine.h"
 
-namespace webrtc {
+namespace {
+
+using webrtc::ByteWriter;
 
 const size_t kAbsoluteSendTimeLength = 4;
 
@@ -45,33 +49,93 @@
   ByteWriter<uint16_t>::WriteBigEndian(header + 2, 0x1234);  // Sequence number.
   ByteWriter<uint32_t>::WriteBigEndian(header + 4, 0x5678);  // Timestamp.
   ByteWriter<uint32_t>::WriteBigEndian(header + 8, 0x4321);  // SSRC.
-  int32_t rtp_header_length = kRtpHeaderSize;
+  int32_t rtp_header_length = webrtc::kRtpHeaderSize;
 
   BuildAbsoluteSendTimeExtension(header + rtp_header_length, extension_id,
                                  abs_send_time);
   rtp_header_length += kAbsoluteSendTimeLength;
   return rtp_header_length;
 }
+}  // namespace
+
+namespace webrtc {
+namespace test {
 
 TEST(AudioReceiveStreamTest, AudioPacketUpdatesBweWithTimestamp) {
-  MockRemoteBitrateEstimator rbe;
+  MockRemoteBitrateEstimator remote_bitrate_estimator;
+  FakeVoiceEngine voice_engine;
   AudioReceiveStream::Config config;
   config.combined_audio_video_bwe = true;
-  config.voe_channel_id = 1;
+  config.voe_channel_id = voice_engine.kReceiveChannelId;
   const int kAbsSendTimeId = 3;
   config.rtp.extensions.push_back(
       RtpExtension(RtpExtension::kAbsSendTime, kAbsSendTimeId));
-  internal::AudioReceiveStream recv_stream(&rbe, config);
+  internal::AudioReceiveStream recv_stream(&remote_bitrate_estimator, config,
+                                           &voice_engine);
   uint8_t rtp_packet[30];
   const int kAbsSendTimeValue = 1234;
   CreateRtpHeaderWithAbsSendTime(rtp_packet, kAbsSendTimeId, kAbsSendTimeValue);
   PacketTime packet_time(5678000, 0);
   const size_t kExpectedHeaderLength = 20;
-  EXPECT_CALL(rbe, IncomingPacket(packet_time.timestamp / 1000,
-                                  sizeof(rtp_packet) - kExpectedHeaderLength,
-                                  testing::_, false))
+  EXPECT_CALL(remote_bitrate_estimator,
+      IncomingPacket(packet_time.timestamp / 1000,
+          sizeof(rtp_packet) - kExpectedHeaderLength, testing::_, false))
       .Times(1);
   EXPECT_TRUE(
       recv_stream.DeliverRtp(rtp_packet, sizeof(rtp_packet), packet_time));
 }
+
+TEST(AudioReceiveStreamTest, GetStats) {
+  const uint32_t kSsrc1 = 667;
+
+  MockRemoteBitrateEstimator remote_bitrate_estimator;
+  FakeVoiceEngine voice_engine;
+  AudioReceiveStream::Config config;
+  config.rtp.remote_ssrc = kSsrc1;
+  config.voe_channel_id = voice_engine.kReceiveChannelId;
+  internal::AudioReceiveStream recv_stream(&remote_bitrate_estimator, config,
+                                           &voice_engine);
+
+  AudioReceiveStream::Stats stats = recv_stream.GetStats();
+  const CallStatistics& call_stats = voice_engine.GetRecvCallStats();
+  const CodecInst& codec_inst = voice_engine.GetRecvRecCodecInst();
+  const NetworkStatistics& net_stats = voice_engine.GetRecvNetworkStats();
+  const AudioDecodingCallStats& decode_stats =
+      voice_engine.GetRecvAudioDecodingCallStats();
+  EXPECT_EQ(kSsrc1, stats.remote_ssrc);
+  EXPECT_EQ(static_cast<int64_t>(call_stats.bytesReceived), stats.bytes_rcvd);
+  EXPECT_EQ(static_cast<uint32_t>(call_stats.packetsReceived),
+            stats.packets_rcvd);
+  EXPECT_EQ(call_stats.cumulativeLost, stats.packets_lost);
+  EXPECT_EQ(static_cast<float>(call_stats.fractionLost) / 256,
+            stats.fraction_lost);
+  EXPECT_EQ(std::string(codec_inst.plname), stats.codec_name);
+  EXPECT_EQ(call_stats.extendedMax, stats.ext_seqnum);
+  EXPECT_EQ(call_stats.jitterSamples / (codec_inst.plfreq / 1000),
+            stats.jitter_ms);
+  EXPECT_EQ(net_stats.currentBufferSize, stats.jitter_buffer_ms);
+  EXPECT_EQ(net_stats.preferredBufferSize, stats.jitter_buffer_preferred_ms);
+  EXPECT_EQ(static_cast<uint32_t>(voice_engine.kRecvJitterBufferDelay +
+      voice_engine.kRecvPlayoutBufferDelay), stats.delay_estimate_ms);
+  EXPECT_EQ(static_cast<int32_t>(voice_engine.kRecvSpeechOutputLevel),
+            stats.audio_level);
+  EXPECT_EQ(Q14ToFloat(net_stats.currentExpandRate), stats.expand_rate);
+  EXPECT_EQ(Q14ToFloat(net_stats.currentSpeechExpandRate),
+            stats.speech_expand_rate);
+  EXPECT_EQ(Q14ToFloat(net_stats.currentSecondaryDecodedRate),
+            stats.secondary_decoded_rate);
+  EXPECT_EQ(Q14ToFloat(net_stats.currentAccelerateRate), stats.accelerate_rate);
+  EXPECT_EQ(Q14ToFloat(net_stats.currentPreemptiveRate),
+            stats.preemptive_expand_rate);
+  EXPECT_EQ(decode_stats.calls_to_silence_generator,
+            stats.decoding_calls_to_silence_generator);
+  EXPECT_EQ(decode_stats.calls_to_neteq, stats.decoding_calls_to_neteq);
+  EXPECT_EQ(decode_stats.decoded_normal, stats.decoding_normal);
+  EXPECT_EQ(decode_stats.decoded_plc, stats.decoding_plc);
+  EXPECT_EQ(decode_stats.decoded_cng, stats.decoding_cng);
+  EXPECT_EQ(decode_stats.decoded_plc_cng, stats.decoding_plc_cng);
+  EXPECT_EQ(call_stats.capture_start_ntp_time_ms_,
+            stats.capture_start_ntp_time_ms);
+}
+}  // namespace test
 }  // namespace webrtc
diff --git a/webrtc/audio/conversion.h b/webrtc/audio/conversion.h
new file mode 100644
index 0000000..c1cf9b6
--- /dev/null
+++ b/webrtc/audio/conversion.h
@@ -0,0 +1,22 @@
+/*
+ *  Copyright (c) 2015 The WebRTC project authors. All Rights Reserved.
+ *
+ *  Use of this source code is governed by a BSD-style license
+ *  that can be found in the LICENSE file in the root of the source
+ *  tree. An additional intellectual property rights grant can be found
+ *  in the file PATENTS.  All contributing project authors may
+ *  be found in the AUTHORS file in the root of the source tree.
+ */
+
+#ifndef WEBRTC_AUDIO_CONVERSION_H_
+#define WEBRTC_AUDIO_CONVERSION_H_
+
+namespace webrtc {
+
+// Convert fixed point number with 14 bit fractional part, to floating point.
+inline float Q14ToFloat(uint16_t v) {
+  return static_cast<float>(v) / (1 << 14);
+}
+}  // namespace webrtc
+
+#endif  // WEBRTC_AUDIO_CONVERSION_H_
diff --git a/webrtc/audio/scoped_voe_interface.h b/webrtc/audio/scoped_voe_interface.h
new file mode 100644
index 0000000..1029337
--- /dev/null
+++ b/webrtc/audio/scoped_voe_interface.h
@@ -0,0 +1,45 @@
+/*
+ *  Copyright (c) 2015 The WebRTC project authors. All Rights Reserved.
+ *
+ *  Use of this source code is governed by a BSD-style license
+ *  that can be found in the LICENSE file in the root of the source
+ *  tree. An additional intellectual property rights grant can be found
+ *  in the file PATENTS.  All contributing project authors may
+ *  be found in the AUTHORS file in the root of the source tree.
+ */
+
+#ifndef WEBRTC_AUDIO_SCOPED_VOE_INTERFACE_H_
+#define WEBRTC_AUDIO_SCOPED_VOE_INTERFACE_H_
+
+#include "webrtc/base/checks.h"
+
+namespace webrtc {
+
+class VoiceEngine;
+
+namespace internal {
+
+// Utility template for obtaining and holding a reference to a VoiceEngine
+// interface and making sure it is released when this object goes out of scope.
+template<class T> class ScopedVoEInterface {
+ public:
+  explicit ScopedVoEInterface(webrtc::VoiceEngine* e)
+      : ptr_(T::GetInterface(e)) {
+    RTC_DCHECK(ptr_);
+  }
+  ~ScopedVoEInterface() {
+    if (ptr_) {
+      ptr_->Release();
+    }
+  }
+  T* operator->() {
+    RTC_DCHECK(ptr_);
+    return ptr_;
+  }
+ private:
+  T* ptr_;
+};
+}  // namespace internal
+}  // namespace webrtc
+
+#endif  // WEBRTC_AUDIO_SCOPED_VOE_INTERFACE_H_
diff --git a/webrtc/audio/webrtc_audio.gypi b/webrtc/audio/webrtc_audio.gypi
index 40ccff6..b9d45db 100644
--- a/webrtc/audio/webrtc_audio.gypi
+++ b/webrtc/audio/webrtc_audio.gypi
@@ -18,6 +18,8 @@
       'audio/audio_receive_stream.h',
       'audio/audio_send_stream.cc',
       'audio/audio_send_stream.h',
+      'audio/conversion.h',
+      'audio/scoped_voe_interface.h',
     ],
   },
 }
diff --git a/webrtc/audio_receive_stream.h b/webrtc/audio_receive_stream.h
index 70d6480..3e5a518 100644
--- a/webrtc/audio_receive_stream.h
+++ b/webrtc/audio_receive_stream.h
@@ -26,7 +26,32 @@
 
 class AudioReceiveStream : public ReceiveStream {
  public:
-  struct Stats {};
+  struct Stats {
+    uint32_t remote_ssrc = 0;
+    int64_t bytes_rcvd = 0;
+    uint32_t packets_rcvd = 0;
+    uint32_t packets_lost = 0;
+    float fraction_lost = 0.0f;
+    std::string codec_name;
+    uint32_t ext_seqnum = 0;
+    uint32_t jitter_ms = 0;
+    uint32_t jitter_buffer_ms = 0;
+    uint32_t jitter_buffer_preferred_ms = 0;
+    uint32_t delay_estimate_ms = 0;
+    int32_t audio_level = -1;
+    float expand_rate = 0.0f;
+    float speech_expand_rate = 0.0f;
+    float secondary_decoded_rate = 0.0f;
+    float accelerate_rate = 0.0f;
+    float preemptive_expand_rate = 0.0f;
+    int32_t decoding_calls_to_silence_generator = 0;
+    int32_t decoding_calls_to_neteq = 0;
+    int32_t decoding_normal = 0;
+    int32_t decoding_plc = 0;
+    int32_t decoding_cng = 0;
+    int32_t decoding_plc_cng = 0;
+    int64_t capture_start_ntp_time_ms = 0;
+  };
 
   struct Config {
     std::string ToString() const;
diff --git a/webrtc/call/bitrate_estimator_tests.cc b/webrtc/call/bitrate_estimator_tests.cc
index f7044ae..08e36c8 100644
--- a/webrtc/call/bitrate_estimator_tests.cc
+++ b/webrtc/call/bitrate_estimator_tests.cc
@@ -25,6 +25,7 @@
 #include "webrtc/test/encoder_settings.h"
 #include "webrtc/test/fake_decoder.h"
 #include "webrtc/test/fake_encoder.h"
+#include "webrtc/test/fake_voice_engine.h"
 #include "webrtc/test/frame_generator_capturer.h"
 
 namespace webrtc {
@@ -130,8 +131,10 @@
   }
 
   virtual void SetUp() {
-    receiver_call_.reset(Call::Create(Call::Config()));
-    sender_call_.reset(Call::Create(Call::Config()));
+    Call::Config config;
+    config.voice_engine = &fake_voice_engine_;
+    receiver_call_.reset(Call::Create(config));
+    sender_call_.reset(Call::Create(config));
 
     send_transport_.SetReceiver(receiver_call_->Receiver());
     receive_transport_.SetReceiver(sender_call_->Receiver());
@@ -265,6 +268,7 @@
     test::FakeDecoder fake_decoder_;
   };
 
+  test::FakeVoiceEngine fake_voice_engine_;
   TraceObserver receiver_trace_;
   test::DirectTransport send_transport_;
   test::DirectTransport receive_transport_;
diff --git a/webrtc/call/call.cc b/webrtc/call/call.cc
index b142453..3969bc6 100644
--- a/webrtc/call/call.cc
+++ b/webrtc/call/call.cc
@@ -126,7 +126,8 @@
 
   VideoSendStream::RtpStateMap suspended_video_send_ssrcs_;
 
-  RtcEventLog* event_log_;
+  RtcEventLog* event_log_ = nullptr;
+  VoECodec* voe_codec_ = nullptr;
 
   RTC_DISALLOW_COPY_AND_ASSIGN(Call);
 };
@@ -147,8 +148,7 @@
       config_(config),
       network_enabled_(true),
       receive_crit_(RWLockWrapper::CreateRWLock()),
-      send_crit_(RWLockWrapper::CreateRWLock()),
-      event_log_(nullptr) {
+      send_crit_(RWLockWrapper::CreateRWLock()) {
   RTC_DCHECK(configuration_thread_checker_.CalledOnValidThread());
   RTC_DCHECK_GE(config.bitrate_config.min_bitrate_bps, 0);
   RTC_DCHECK_GE(config.bitrate_config.start_bitrate_bps,
@@ -158,11 +158,11 @@
                   config.bitrate_config.start_bitrate_bps);
   }
   if (config.voice_engine) {
-    VoECodec* voe_codec = VoECodec::GetInterface(config.voice_engine);
-    if (voe_codec) {
-      event_log_ = voe_codec->GetEventLog();
-      voe_codec->Release();
-    }
+    // Keep a reference to VoECodec, so we're sure the VoiceEngine lives for the
+    // duration of the call.
+    voe_codec_ = VoECodec::GetInterface(config.voice_engine);
+    if (voe_codec_)
+      event_log_ = voe_codec_->GetEventLog();
   }
 
   Trace::CreateTrace();
@@ -187,6 +187,9 @@
   module_process_thread_->DeRegisterModule(call_stats_.get());
   module_process_thread_->Stop();
   Trace::ReturnTrace();
+
+  if (voe_codec_)
+    voe_codec_->Release();
 }
 
 PacketReceiver* Call::Receiver() {
@@ -237,7 +240,8 @@
   TRACE_EVENT0("webrtc", "Call::CreateAudioReceiveStream");
   RTC_DCHECK(configuration_thread_checker_.CalledOnValidThread());
   AudioReceiveStream* receive_stream = new AudioReceiveStream(
-      congestion_controller_->GetRemoteBitrateEstimator(false), config);
+      congestion_controller_->GetRemoteBitrateEstimator(false), config,
+      config_.voice_engine);
   {
     WriteLockScoped write_lock(*receive_crit_);
     RTC_DCHECK(audio_receive_ssrcs_.find(config.rtp.remote_ssrc) ==
diff --git a/webrtc/call/call_perf_tests.cc b/webrtc/call/call_perf_tests.cc
index a7714b1..cab3914 100644
--- a/webrtc/call/call_perf_tests.cc
+++ b/webrtc/call/call_perf_tests.cc
@@ -321,6 +321,8 @@
 
   receiver_call_->DestroyAudioReceiveStream(audio_receive_stream);
 
+  DestroyCalls();
+
   VoiceEngine::Delete(voice_engine);
 }
 
diff --git a/webrtc/call/call_unittest.cc b/webrtc/call/call_unittest.cc
index 9adecc3..9819b53 100644
--- a/webrtc/call/call_unittest.cc
+++ b/webrtc/call/call_unittest.cc
@@ -13,19 +13,21 @@
 #include "testing/gtest/include/gtest/gtest.h"
 
 #include "webrtc/call.h"
+#include "webrtc/test/fake_voice_engine.h"
 
 namespace {
 
 struct CallHelper {
-  CallHelper() {
+  CallHelper() : voice_engine_(new webrtc::test::FakeVoiceEngine()) {
     webrtc::Call::Config config;
-    // TODO(solenberg): Fill in with VoiceEngine* etc.
+    config.voice_engine = voice_engine_.get();
     call_.reset(webrtc::Call::Create(config));
   }
 
   webrtc::Call* operator->() { return call_.get(); }
 
  private:
+  rtc::scoped_ptr<webrtc::test::FakeVoiceEngine> voice_engine_;
   rtc::scoped_ptr<webrtc::Call> call_;
 };
 }  // namespace
diff --git a/webrtc/test/call_test.cc b/webrtc/test/call_test.cc
index 0986df5..f2b5f91 100644
--- a/webrtc/test/call_test.cc
+++ b/webrtc/test/call_test.cc
@@ -89,6 +89,11 @@
   receiver_call_.reset(Call::Create(config));
 }
 
+void CallTest::DestroyCalls() {
+  sender_call_.reset(nullptr);
+  receiver_call_.reset(nullptr);
+}
+
 void CallTest::CreateSendConfig(size_t num_streams,
                                 Transport* send_transport) {
   assert(num_streams <= kNumSsrcs);
diff --git a/webrtc/test/call_test.h b/webrtc/test/call_test.h
index 2b9dcee..4a645b4 100644
--- a/webrtc/test/call_test.h
+++ b/webrtc/test/call_test.h
@@ -52,6 +52,7 @@
                    const Call::Config& receiver_config);
   void CreateSenderCall(const Call::Config& config);
   void CreateReceiverCall(const Call::Config& config);
+  void DestroyCalls();
 
   void CreateSendConfig(size_t num_streams, Transport* send_transport);
   void CreateMatchingReceiveConfigs(Transport* rtcp_send_transport);
diff --git a/webrtc/test/fake_voice_engine.h b/webrtc/test/fake_voice_engine.h
new file mode 100644
index 0000000..72f6b27
--- /dev/null
+++ b/webrtc/test/fake_voice_engine.h
@@ -0,0 +1,421 @@
+/*
+ *  Copyright (c) 2015 The WebRTC project authors. All Rights Reserved.
+ *
+ *  Use of this source code is governed by a BSD-style license
+ *  that can be found in the LICENSE file in the root of the source
+ *  tree. An additional intellectual property rights grant can be found
+ *  in the file PATENTS.  All contributing project authors may
+ *  be found in the AUTHORS file in the root of the source tree.
+ */
+
+#ifndef WEBRTC_AUDIO_FAKE_VOICE_ENGINE_H_
+#define WEBRTC_AUDIO_FAKE_VOICE_ENGINE_H_
+
+#include <vector>
+
+#include "testing/gtest/include/gtest/gtest.h"
+
+#include "webrtc/voice_engine/voice_engine_impl.h"
+
+namespace webrtc {
+namespace test {
+
+// NOTE: This class inherits from VoiceEngineImpl so that its clients will be
+// able to get the various interfaces as usual, via T::GetInterface().
+class FakeVoiceEngine final : public VoiceEngineImpl {
+ public:
+  const int kSendChannelId = 1;
+  const int kReceiveChannelId = 2;
+
+  const int kRecvJitterBufferDelay = -7;
+  const int kRecvPlayoutBufferDelay = 302;
+  const unsigned int kRecvSpeechOutputLevel = 99;
+
+  FakeVoiceEngine() : VoiceEngineImpl(new Config(), true) {
+    // Increase ref count so this object isn't automatically deleted whenever
+    // interfaces are Release():d.
+    ++_ref_count;
+  }
+  ~FakeVoiceEngine() override {
+    // Decrease ref count before base class d-tor is called; otherwise it will
+    // trigger an assertion.
+    --_ref_count;
+  }
+
+  const CallStatistics& GetRecvCallStats() const {
+    static const CallStatistics kStats = {
+      345, 678, 901, 234, -1, 0, 0, 567, 890, 123
+    };
+    return kStats;
+  }
+
+  const CodecInst& GetRecvRecCodecInst() const {
+    static const CodecInst kStats = {
+      123, "codec_name", 96000, -1, -1, -1
+    };
+    return kStats;
+  }
+
+  const NetworkStatistics& GetRecvNetworkStats() const {
+    static const NetworkStatistics kStats = {
+      123, 456, false, 0, 0, 789, 12, 345, 678, 901, -1, -1, -1, -1, -1, 0
+    };
+    return kStats;
+  }
+
+  const AudioDecodingCallStats& GetRecvAudioDecodingCallStats() const {
+    static AudioDecodingCallStats stats;
+    if (stats.calls_to_silence_generator == 0) {
+      stats.calls_to_silence_generator = 234;
+      stats.calls_to_neteq = 567;
+      stats.decoded_normal = 890;
+      stats.decoded_plc = 123;
+      stats.decoded_cng = 456;
+      stats.decoded_plc_cng = 789;
+    }
+    return stats;
+  }
+
+  // VoEBase
+  int RegisterVoiceEngineObserver(VoiceEngineObserver& observer) override {
+    return -1;
+  }
+  int DeRegisterVoiceEngineObserver() override { return -1; }
+  int Init(AudioDeviceModule* external_adm = NULL,
+           AudioProcessing* audioproc = NULL) override { return -1; }
+  AudioProcessing* audio_processing() override { return nullptr; }
+  int Terminate() override { return -1; }
+  int CreateChannel() override { return -1; }
+  int CreateChannel(const Config& config) override { return -1; }
+  int DeleteChannel(int channel) override { return -1; }
+  int StartReceive(int channel) override { return -1; }
+  int StopReceive(int channel) override { return -1; }
+  int StartPlayout(int channel) override { return -1; }
+  int StopPlayout(int channel) override { return -1; }
+  int StartSend(int channel) override { return -1; }
+  int StopSend(int channel) override { return -1; }
+  int GetVersion(char version[1024]) override { return -1; }
+  int LastError() override { return -1; }
+  AudioTransport* audio_transport() { return nullptr; }
+  int AssociateSendChannel(int channel, int accociate_send_channel) override {
+    return -1;
+  }
+
+  // VoECodec
+  int NumOfCodecs() override { return -1; }
+  int GetCodec(int index, CodecInst& codec) override { return -1; }
+  int SetSendCodec(int channel, const CodecInst& codec) override { return -1; }
+  int GetSendCodec(int channel, CodecInst& codec) override { return -1; }
+  int SetBitRate(int channel, int bitrate_bps) override { return -1; }
+  int GetRecCodec(int channel, CodecInst& codec) override {
+    EXPECT_EQ(channel, kReceiveChannelId);
+    codec = GetRecvRecCodecInst();
+    return 0;
+  }
+  int SetRecPayloadType(int channel, const CodecInst& codec) override {
+    return -1;
+  }
+  int GetRecPayloadType(int channel, CodecInst& codec) override { return -1; }
+  int SetSendCNPayloadType(int channel, int type,
+      PayloadFrequencies frequency = kFreq16000Hz) override { return -1; }
+  int SetVADStatus(int channel,
+                   bool enable,
+                   VadModes mode = kVadConventional,
+                   bool disableDTX = false) override { return -1; }
+  int GetVADStatus(int channel,
+                   bool& enabled,
+                   VadModes& mode,
+                   bool& disabledDTX) override { return -1; }
+  int SetOpusMaxPlaybackRate(int channel, int frequency_hz) override {
+    return -1;
+  }
+  int SetOpusDtx(int channel, bool enable_dtx) override { return -1; }
+  RtcEventLog* GetEventLog() override { return nullptr; }
+
+  // VoEDtmf
+  int SendTelephoneEvent(int channel,
+                         int eventCode,
+                         bool outOfBand = true,
+                         int lengthMs = 160,
+                         int attenuationDb = 10) override { return -1; }
+  int SetSendTelephoneEventPayloadType(int channel,
+                                       unsigned char type) override {
+    return -1;
+  }
+  int GetSendTelephoneEventPayloadType(int channel,
+                                       unsigned char& type) override {
+    return -1;
+  }
+  int SetDtmfFeedbackStatus(bool enable,
+                            bool directFeedback = false) override { return -1; }
+  int GetDtmfFeedbackStatus(bool& enabled, bool& directFeedback) override {
+    return -1;
+  }
+  int PlayDtmfTone(int eventCode,
+                   int lengthMs = 200,
+                   int attenuationDb = 10) override { return -1; }
+
+  // VoEExternalMedia
+  int RegisterExternalMediaProcessing(
+      int channel,
+      ProcessingTypes type,
+      VoEMediaProcess& processObject) override { return -1; }
+  int DeRegisterExternalMediaProcessing(int channel,
+                                        ProcessingTypes type) override {
+    return -1;
+  }
+  int GetAudioFrame(int channel,
+                    int desired_sample_rate_hz,
+                    AudioFrame* frame) override { return -1; }
+  int SetExternalMixing(int channel, bool enable) override { return -1; }
+
+  // VoEFile
+  int StartPlayingFileLocally(
+      int channel,
+      const char fileNameUTF8[1024],
+      bool loop = false,
+      FileFormats format = kFileFormatPcm16kHzFile,
+      float volumeScaling = 1.0,
+      int startPointMs = 0,
+      int stopPointMs = 0) override { return -1; }
+  int StartPlayingFileLocally(
+      int channel,
+      InStream* stream,
+      FileFormats format = kFileFormatPcm16kHzFile,
+      float volumeScaling = 1.0,
+      int startPointMs = 0,
+      int stopPointMs = 0) override { return -1; }
+  int StopPlayingFileLocally(int channel) override { return -1; }
+  int IsPlayingFileLocally(int channel) override { return -1; }
+  int StartPlayingFileAsMicrophone(
+      int channel,
+      const char fileNameUTF8[1024],
+      bool loop = false,
+      bool mixWithMicrophone = false,
+      FileFormats format = kFileFormatPcm16kHzFile,
+      float volumeScaling = 1.0) override { return -1; }
+  int StartPlayingFileAsMicrophone(
+      int channel,
+      InStream* stream,
+      bool mixWithMicrophone = false,
+      FileFormats format = kFileFormatPcm16kHzFile,
+      float volumeScaling = 1.0) override { return -1; }
+  int StopPlayingFileAsMicrophone(int channel) override { return -1; }
+  int IsPlayingFileAsMicrophone(int channel) override { return -1; }
+  int StartRecordingPlayout(int channel,
+                            const char* fileNameUTF8,
+                            CodecInst* compression = NULL,
+                            int maxSizeBytes = -1) override { return -1; }
+  int StopRecordingPlayout(int channel) override { return -1; }
+  int StartRecordingPlayout(int channel,
+                            OutStream* stream,
+                            CodecInst* compression = NULL) override {
+    return -1;
+  }
+  int StartRecordingMicrophone(const char* fileNameUTF8,
+                               CodecInst* compression = NULL,
+                               int maxSizeBytes = -1) override { return -1; }
+  int StartRecordingMicrophone(OutStream* stream,
+                                       CodecInst* compression = NULL) override {
+    return -1;
+  }
+  int StopRecordingMicrophone() override { return -1; }
+
+  // VoEHardware
+  int GetNumOfRecordingDevices(int& devices) override { return -1; }
+
+  // Gets the number of audio devices available for playout.
+  int GetNumOfPlayoutDevices(int& devices) override { return -1; }
+
+  // Gets the name of a specific recording device given by an |index|.
+  // On Windows Vista/7, it also retrieves an additional unique ID
+  // (GUID) for the recording device.
+  int GetRecordingDeviceName(int index,
+                             char strNameUTF8[128],
+                             char strGuidUTF8[128]) override { return -1; }
+
+  // Gets the name of a specific playout device given by an |index|.
+  // On Windows Vista/7, it also retrieves an additional unique ID
+  // (GUID) for the playout device.
+  int GetPlayoutDeviceName(int index,
+                           char strNameUTF8[128],
+                           char strGuidUTF8[128]) override { return -1; }
+
+  // Sets the audio device used for recording.
+  int SetRecordingDevice(
+      int index,
+      StereoChannel recordingChannel = kStereoBoth) override { return -1; }
+
+  // Sets the audio device used for playout.
+  int SetPlayoutDevice(int index) override { return -1; }
+
+  // Sets the type of audio device layer to use.
+  int SetAudioDeviceLayer(AudioLayers audioLayer) override { return -1; }
+
+  // Gets the currently used (active) audio device layer.
+  int GetAudioDeviceLayer(AudioLayers& audioLayer) override { return -1; }
+
+  // Native sample rate controls (samples/sec)
+  int SetRecordingSampleRate(unsigned int samples_per_sec) override {
+    return -1;
+  }
+  int RecordingSampleRate(unsigned int* samples_per_sec) const override {
+    return -1;
+  }
+  int SetPlayoutSampleRate(unsigned int samples_per_sec) override {
+    return -1;
+  }
+  int PlayoutSampleRate(unsigned int* samples_per_sec) const override {
+    return -1;
+  }
+
+  // Queries and controls platform audio effects on Android devices.
+  bool BuiltInAECIsAvailable() const override { return false; }
+  int EnableBuiltInAEC(bool enable) override { return -1; }
+  bool BuiltInAGCIsAvailable() const override { return false; }
+  int EnableBuiltInAGC(bool enable) override { return -1; }
+  bool BuiltInNSIsAvailable() const override { return false; }
+  int EnableBuiltInNS(bool enable) override { return -1; }
+
+  // VoENetwork
+  int RegisterExternalTransport(int channel, Transport& transport) override {
+    return -1;
+  }
+  int DeRegisterExternalTransport(int channel) override { return -1; }
+  int ReceivedRTPPacket(int channel,
+                        const void* data,
+                        size_t length) override { return -1; }
+  int ReceivedRTPPacket(int channel,
+                        const void* data,
+                        size_t length,
+                        const PacketTime& packet_time) override { return -1; }
+  int ReceivedRTCPPacket(int channel,
+                         const void* data,
+                         size_t length) { return -1; }
+
+  // VoENetEqStats
+  int GetNetworkStatistics(int channel, NetworkStatistics& stats) override {
+    EXPECT_EQ(channel, kReceiveChannelId);
+    stats = GetRecvNetworkStats();
+    return 0;
+  }
+  int GetDecodingCallStatistics(int channel,
+                                AudioDecodingCallStats* stats) const override {
+    EXPECT_EQ(channel, kReceiveChannelId);
+    EXPECT_NE(nullptr, stats);
+    *stats = GetRecvAudioDecodingCallStats();
+    return 0;
+  }
+
+  // VoERTP_RTCP
+  int SetLocalSSRC(int channel, unsigned int ssrc) override { return -1; }
+  int GetLocalSSRC(int channel, unsigned int& ssrc) override { return -1; }
+  int GetRemoteSSRC(int channel, unsigned int& ssrc) override {
+    EXPECT_EQ(channel, kReceiveChannelId);
+    ssrc = 0;
+    return 0;
+  }
+  int SetSendAudioLevelIndicationStatus(int channel,
+                                        bool enable,
+                                        unsigned char id = 1) override {
+    return -1;
+  }
+  int SetSendAbsoluteSenderTimeStatus(int channel,
+                                      bool enable,
+                                      unsigned char id) override { return -1; }
+  int SetReceiveAbsoluteSenderTimeStatus(int channel,
+                                         bool enable,
+                                         unsigned char id) override {
+    return -1;
+  }
+  int SetRTCPStatus(int channel, bool enable) override { return -1; }
+  int GetRTCPStatus(int channel, bool& enabled) override { return -1; }
+  int SetRTCP_CNAME(int channel, const char cName[256]) override { return -1; }
+  int GetRTCP_CNAME(int channel, char cName[256]) { return -1; }
+  int GetRemoteRTCP_CNAME(int channel, char cName[256]) override { return -1; }
+  int GetRemoteRTCPData(int channel,
+                        unsigned int& NTPHigh,
+                        unsigned int& NTPLow,
+                        unsigned int& timestamp,
+                        unsigned int& playoutTimestamp,
+                        unsigned int* jitter = NULL,
+                        unsigned short* fractionLost = NULL) override {
+    return -1;
+  }
+  int GetRTPStatistics(int channel,
+                       unsigned int& averageJitterMs,
+                       unsigned int& maxJitterMs,
+                       unsigned int& discardedPackets) override { return -1; }
+  int GetRTCPStatistics(int channel, CallStatistics& stats) override {
+    EXPECT_EQ(channel, kReceiveChannelId);
+    stats = GetRecvCallStats();
+    return 0;
+  }
+  int GetRemoteRTCPReportBlocks(
+      int channel,
+      std::vector<ReportBlock>* receive_blocks) override { return -1; }
+  int SetNACKStatus(int channel, bool enable, int maxNoPackets) override {
+    return -1;
+  }
+
+  // VoEVideoSync
+  int GetPlayoutBufferSize(int& buffer_ms) override { return -1; }
+  int SetMinimumPlayoutDelay(int channel, int delay_ms) override { return -1; }
+  int SetInitialPlayoutDelay(int channel, int delay_ms) override { return -1; }
+  int GetDelayEstimate(int channel,
+                       int* jitter_buffer_delay_ms,
+                       int* playout_buffer_delay_ms) override {
+    EXPECT_EQ(channel, kReceiveChannelId);
+    *jitter_buffer_delay_ms = kRecvJitterBufferDelay;
+    *playout_buffer_delay_ms = kRecvPlayoutBufferDelay;
+    return 0;
+  }
+  int GetLeastRequiredDelayMs(int channel) const override { return -1; }
+  int SetInitTimestamp(int channel, unsigned int timestamp) override {
+    return -1;
+  }
+  int SetInitSequenceNumber(int channel, short sequenceNumber) override {
+    return -1;
+  }
+  int GetPlayoutTimestamp(int channel, unsigned int& timestamp) override {
+    return -1;
+  }
+  int GetRtpRtcp(int channel,
+                 RtpRtcp** rtpRtcpModule,
+                 RtpReceiver** rtp_receiver) override { return -1; }
+
+  // VoEVolumeControl
+  int SetSpeakerVolume(unsigned int volume) override { return -1; }
+  int GetSpeakerVolume(unsigned int& volume) override { return -1; }
+  int SetMicVolume(unsigned int volume) override { return -1; }
+  int GetMicVolume(unsigned int& volume) override { return -1; }
+  int SetInputMute(int channel, bool enable) override { return -1; }
+  int GetInputMute(int channel, bool& enabled) override { return -1; }
+  int GetSpeechInputLevel(unsigned int& level) override { return -1; }
+  int GetSpeechOutputLevel(int channel, unsigned int& level) override {
+    return -1;
+  }
+  int GetSpeechInputLevelFullRange(unsigned int& level) override { return -1; }
+  int GetSpeechOutputLevelFullRange(int channel,
+                                    unsigned int& level) override {
+    EXPECT_EQ(channel, kReceiveChannelId);
+    level = kRecvSpeechOutputLevel;
+    return 0;
+  }
+  int SetChannelOutputVolumeScaling(int channel, float scaling) override {
+    return -1;
+  }
+  int GetChannelOutputVolumeScaling(int channel, float& scaling) override {
+    return -1;
+  }
+  int SetOutputVolumePan(int channel, float left, float right) override {
+    return -1;
+  }
+  int GetOutputVolumePan(int channel, float& left, float& right) override {
+    return -1;
+  }
+};
+}  // namespace test
+}  // namespace webrtc
+
+#endif  // WEBRTC_AUDIO_FAKE_VOICE_ENGINE_H_
diff --git a/webrtc/test/webrtc_test_common.gyp b/webrtc/test/webrtc_test_common.gyp
index f8d3365..5076900 100644
--- a/webrtc/test/webrtc_test_common.gyp
+++ b/webrtc/test/webrtc_test_common.gyp
@@ -30,6 +30,7 @@
         'fake_encoder.h',
         'fake_network_pipe.cc',
         'fake_network_pipe.h',
+        'fake_voice_engine.h',
         'frame_generator_capturer.cc',
         'frame_generator_capturer.h',
         'layer_filtering_transport.cc',
diff --git a/webrtc/voice_engine/voice_engine_impl.h b/webrtc/voice_engine/voice_engine_impl.h
index 07f29c3..2103994 100644
--- a/webrtc/voice_engine/voice_engine_impl.h
+++ b/webrtc/voice_engine/voice_engine_impl.h
@@ -128,8 +128,11 @@
   // This implements the Release() method for all the inherited interfaces.
   int Release() override;
 
- private:
+ // This is *protected* so that FakeVoiceEngine can inherit from the class and
+ // manipulate the reference count. See: fake_voice_engine.h.
+ protected:
   Atomic32 _ref_count;
+ private:
   rtc::scoped_ptr<const Config> own_config_;
 };