Injectable audio encoders: WebRtcVoiceEngine and company

webrtc/audio/audio_send_stream_unittest.cc

 /*
  *  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.
  */
 
 #include <string>
 #include <vector>
 
 #include "webrtc/audio/audio_send_stream.h"
 #include "webrtc/audio/audio_state.h"
 #include "webrtc/audio/conversion.h"
 #include "webrtc/base/task_queue.h"
 #include "webrtc/logging/rtc_event_log/mock/mock_rtc_event_log.h"
+#include "webrtc/modules/audio_coding/codecs/builtin_audio_encoder_factory.h"
 #include "webrtc/modules/audio_mixer/audio_mixer_impl.h"
 #include "webrtc/modules/audio_processing/include/mock_audio_processing.h"
 #include "webrtc/modules/congestion_controller/include/congestion_controller.h"
 #include "webrtc/modules/congestion_controller/include/mock/mock_congestion_controller.h"
 #include "webrtc/modules/pacing/paced_sender.h"
 #include "webrtc/modules/remote_bitrate_estimator/include/mock/mock_remote_bitrate_estimator.h"
 #include "webrtc/modules/rtp_rtcp/mocks/mock_rtcp_rtt_stats.h"
 #include "webrtc/test/gtest.h"
 #include "webrtc/test/mock_voe_channel_proxy.h"
 #include "webrtc/test/mock_voice_engine.h"
@@ skipping 19 matching lines @@
 const float kResidualEchoLikelihood = -1.0f;
 const unsigned int kSpeechInputLevel = 96;
 const CallStatistics kCallStats = {
     1345,  1678,  1901, 1234,  112, 13456, 17890, 1567, -1890, -1123};
 const ReportBlock kReportBlock = {456, 780, 123, 567, 890, 132, 143, 13354};
 const int kTelephoneEventPayloadType = 123;
 const int kTelephoneEventPayloadFrequency = 65432;
 const int kTelephoneEventCode = 45;
 const int kTelephoneEventDuration = 6789;
 const CodecInst kIsacCodec = {103, "isac", 16000, 320, 1, 32000};
+const int kIsacPayloadType = 103;
+const SdpAudioFormat kIsacFormat = {"isac", 16000, 1};
 
 class MockLimitObserver : public BitrateAllocator::LimitObserver {
  public:
   MOCK_METHOD2(OnAllocationLimitsChanged,
                void(uint32_t min_send_bitrate_bps,
                     uint32_t max_padding_bitrate_bps));
 };
 
 struct ConfigHelper {
   explicit ConfigHelper(bool audio_bwe_enabled)
@@ skipping 37 matching lines @@
     stream_config_.rtp.extensions.push_back(
         RtpExtension(RtpExtension::kAudioLevelUri, kAudioLevelId));
     if (audio_bwe_enabled) {
       stream_config_.rtp.extensions.push_back(
           RtpExtension(RtpExtension::kTransportSequenceNumberUri,
                        kTransportSequenceNumberId));
       stream_config_.send_codec_spec.transport_cc_enabled = true;
     }
     // Use ISAC as default codec so as to prevent unnecessary |voice_engine_|
     // calls from the default ctor behavior.
-    stream_config_.send_codec_spec.codec_inst = kIsacCodec;
+    stream_config_.send_codec_spec.payload_type = kIsacPayloadType;
+    stream_config_.send_codec_spec.format.format = kIsacFormat;
     stream_config_.min_bitrate_bps = 10000;
     stream_config_.max_bitrate_bps = 65000;
   }
 
   AudioSendStream::Config& config() { return stream_config_; }
   rtc::scoped_refptr<AudioState> audio_state() { return audio_state_; }
   MockVoEChannelProxy* channel_proxy() { return channel_proxy_; }
   PacketRouter* packet_router() { return &packet_router_; }
   CongestionController* congestion_controller() {
     return &congestion_controller_;
@@ skipping 42 matching lines @@
     EXPECT_CALL(*channel_proxy_, SetRtcpRttStats(testing::IsNull()))
         .Times(1);  // Destructor resets the rtt stats.
   }
 
   void SetupMockForSetupSendCodec() {
     EXPECT_CALL(*channel_proxy_, SetVADStatus(false))
         .WillOnce(Return(true));
     EXPECT_CALL(*channel_proxy_, SetCodecFECStatus(false))
         .WillOnce(Return(true));
     EXPECT_CALL(*channel_proxy_, DisableAudioNetworkAdaptor());
-    // Let |GetSendCodec| return false for the first time to indicate that no
-    // send codec has been set.
-    EXPECT_CALL(*channel_proxy_, GetSendCodec(_)).WillOnce(Return(false));
-    EXPECT_CALL(*channel_proxy_, SetSendCodec(_)).WillOnce(Return(true));
+    EXPECT_CALL(*channel_proxy_, SetSendFormat(_, _, _)).WillOnce(Return(true));
+    // These should no longer be called.
+    EXPECT_CALL(*channel_proxy_, GetSendCodec(_))
+        .Times(0);

[kwiberg-webrtc, 2017/02/21 23:35:03]

Won't clang-format remove this line break?

+    EXPECT_CALL(*channel_proxy_, SetSendCodec(_)).Times(0);
   }
   RtcpRttStats* rtcp_rtt_stats() { return &rtcp_rtt_stats_; }
 
   void SetupMockForSendTelephoneEvent() {
     EXPECT_TRUE(channel_proxy_);
     EXPECT_CALL(*channel_proxy_,
         SetSendTelephoneEventPayloadType(kTelephoneEventPayloadType,
                                          kTelephoneEventPayloadFrequency))
             .WillOnce(Return(true));
     EXPECT_CALL(*channel_proxy_,
@@ skipping 64 matching lines @@
 
 TEST(AudioSendStreamTest, ConfigToString) {
   AudioSendStream::Config config(nullptr);
   config.rtp.ssrc = kSsrc;
   config.rtp.c_name = kCName;
   config.voe_channel_id = kChannelId;
   config.min_bitrate_bps = 12000;
   config.max_bitrate_bps = 34000;
   config.send_codec_spec.nack_enabled = true;
   config.send_codec_spec.transport_cc_enabled = false;
-  config.send_codec_spec.enable_codec_fec = true;
-  config.send_codec_spec.enable_opus_dtx = false;
-  config.send_codec_spec.opus_max_playback_rate = 32000;
   config.send_codec_spec.cng_payload_type = 42;
   config.send_codec_spec.cng_plfreq = 56;
-  config.send_codec_spec.min_ptime_ms = 20;
-  config.send_codec_spec.max_ptime_ms = 60;
-  config.send_codec_spec.codec_inst = kIsacCodec;
+  config.send_codec_spec.payload_type = kIsacPayloadType;
+  config.send_codec_spec.format.format = kIsacFormat;
+  // TODO(ossu): Maybe mock this?
+  config.encoder_factory = CreateBuiltinAudioEncoderFactory();
   config.rtp.extensions.push_back(
       RtpExtension(RtpExtension::kAudioLevelUri, kAudioLevelId));
   EXPECT_EQ(
       "{rtp: {ssrc: 1234, extensions: [{uri: "
       "urn:ietf:params:rtp-hdrext:ssrc-audio-level, id: 2}], nack: "
       "{rtp_history_ms: 0}, c_name: foo_name}, send_transport: nullptr, "
       "voe_channel_id: 1, min_bitrate_bps: 12000, max_bitrate_bps: 34000, "
       "send_codec_spec: {nack_enabled: true, transport_cc_enabled: false, "
-      "enable_codec_fec: true, enable_opus_dtx: false, opus_max_playback_rate: "
-      "32000, cng_payload_type: 42, cng_plfreq: 56, min_ptime: 20, max_ptime: "
-      "60, codec_inst: {pltype: 103, plname: \"isac\", plfreq: 16000, pacsize: "
-      "320, channels: 1, rate: 32000}}}",
+      "cng_payload_type: 42, cng_plfreq: 56, payload_type: 103, "
+      "format: {name: isac, clockrate_hz: 16000, num_channels: 1, "
+      "parameters: {}}}}",
       config.ToString());
 }
 
 TEST(AudioSendStreamTest, ConstructDestruct) {
   ConfigHelper helper(false);
   internal::AudioSendStream send_stream(
       helper.config(), helper.audio_state(), helper.worker_queue(),
       helper.packet_router(), helper.congestion_controller(),
       helper.bitrate_allocator(), helper.event_log(), helper.rtcp_rtt_stats());
 }
@@ skipping 82 matching lines @@
       static_cast<VoiceEngineObserver*>(internal_audio_state);
   voe_observer->CallbackOnError(-1, VE_TYPING_NOISE_WARNING);
   EXPECT_TRUE(send_stream.GetStats().typing_noise_detected);
   voe_observer->CallbackOnError(-1, VE_TYPING_NOISE_OFF_WARNING);
   EXPECT_FALSE(send_stream.GetStats().typing_noise_detected);
 }
 
 TEST(AudioSendStreamTest, SendCodecAppliesConfigParams) {
   ConfigHelper helper(false);
   auto stream_config = helper.config();
-  const CodecInst kOpusCodec = {111, "opus", 48000, 960, 2, 64000};
-  stream_config.send_codec_spec.codec_inst = kOpusCodec;
-  stream_config.send_codec_spec.enable_codec_fec = true;
-  stream_config.send_codec_spec.enable_opus_dtx = true;
-  stream_config.send_codec_spec.opus_max_playback_rate = 12345;
+  const webrtc::SdpAudioFormat kOpusFormat = {"opus", 48000, 2};

[kwiberg-webrtc, 2017/02/21 23:35:03]

Too bad we can't make this constexpr. Remind me to talk about a nice small
vector optimization I learned about when we're both at the office at the same
time...

+  stream_config.send_codec_spec.format.format = kOpusFormat;
   stream_config.send_codec_spec.cng_plfreq = 16000;
   stream_config.send_codec_spec.cng_payload_type = 105;
-  stream_config.send_codec_spec.min_ptime_ms = 10;
-  stream_config.send_codec_spec.max_ptime_ms = 60;
   stream_config.audio_network_adaptor_config =
       rtc::Optional<std::string>("abced");
-  EXPECT_CALL(*helper.channel_proxy(), SetCodecFECStatus(true))
-      .WillOnce(Return(true));
-  EXPECT_CALL(
-      *helper.channel_proxy(),
-      SetOpusDtx(stream_config.send_codec_spec.enable_opus_dtx))
-      .WillOnce(Return(true));
-  EXPECT_CALL(
-      *helper.channel_proxy(),
-      SetOpusMaxPlaybackRate(
-          stream_config.send_codec_spec.opus_max_playback_rate))
-      .WillOnce(Return(true));
   EXPECT_CALL(*helper.channel_proxy(),
               SetSendCNPayloadType(
                   stream_config.send_codec_spec.cng_payload_type,
                   webrtc::kFreq16000Hz))
       .WillOnce(Return(true));
   EXPECT_CALL(
       *helper.channel_proxy(),
-      SetReceiverFrameLengthRange(stream_config.send_codec_spec.min_ptime_ms,
-                                  stream_config.send_codec_spec.max_ptime_ms));
-  EXPECT_CALL(
-      *helper.channel_proxy(),
       EnableAudioNetworkAdaptor(*stream_config.audio_network_adaptor_config));
   internal::AudioSendStream send_stream(
       stream_config, helper.audio_state(), helper.worker_queue(),
       helper.packet_router(), helper.congestion_controller(),
       helper.bitrate_allocator(), helper.event_log(), helper.rtcp_rtt_stats());
 }
 
 // VAD is applied when codec is mono and the CNG frequency matches the codec
 // sample rate.
 TEST(AudioSendStreamTest, SendCodecCanApplyVad) {
   ConfigHelper helper(false);
   auto stream_config = helper.config();
-  const CodecInst kG722Codec = {9, "g722", 8000, 160, 1, 16000};
-  stream_config.send_codec_spec.codec_inst = kG722Codec;
+  stream_config.send_codec_spec.payload_type = 9;
+  stream_config.send_codec_spec.format.format = {"g722", 8000, 1};
+  stream_config.send_codec_spec.format.info = {16000, 1, 64000};
   stream_config.send_codec_spec.cng_plfreq = 8000;
   stream_config.send_codec_spec.cng_payload_type = 105;
   EXPECT_CALL(*helper.channel_proxy(), SetVADStatus(true))
       .WillOnce(Return(true));
   internal::AudioSendStream send_stream(
       stream_config, helper.audio_state(), helper.worker_queue(),
       helper.packet_router(), helper.congestion_controller(),
       helper.bitrate_allocator(), helper.event_log(), helper.rtcp_rtt_stats());
 }
 
@@ skipping 14 matching lines @@
   internal::AudioSendStream send_stream(
       helper.config(), helper.audio_state(), helper.worker_queue(),
       helper.packet_router(), helper.congestion_controller(),
       helper.bitrate_allocator(), helper.event_log(), helper.rtcp_rtt_stats());
   EXPECT_CALL(*helper.channel_proxy(), SetBitrate(_, 5000));
   send_stream.OnBitrateUpdated(50000, 0.0, 50, 5000);
 }
 
 }  // namespace test
 }  // namespace webrtc