| OLD | NEW |
| (Empty) |
| 1 /* | |
| 2 * Copyright (c) 2016 The WebRTC project authors. All Rights Reserved. | |
| 3 * | |
| 4 * Use of this source code is governed by a BSD-style license | |
| 5 * that can be found in the LICENSE file in the root of the source | |
| 6 * tree. An additional intellectual property rights grant can be found | |
| 7 * in the file PATENTS. All contributing project authors may | |
| 8 * be found in the AUTHORS file in the root of the source tree. | |
| 9 */ | |
| 10 | |
| 11 #include <string.h> | |
| 12 | |
| 13 #include <memory> | |
| 14 #include <utility> | |
| 15 | |
| 16 #include "webrtc/base/bind.h" | |
| 17 #include "webrtc/base/thread.h" | |
| 18 #include "webrtc/modules/audio_mixer/audio_mixer_impl.h" | |
| 19 #include "webrtc/modules/audio_mixer/audio_mixer.h" | |
| 20 #include "webrtc/test/gmock.h" | |
| 21 | |
| 22 using testing::_; | |
| 23 using testing::Exactly; | |
| 24 using testing::Invoke; | |
| 25 using testing::Return; | |
| 26 | |
| 27 namespace webrtc { | |
| 28 | |
| 29 namespace { | |
| 30 | |
| 31 constexpr int kDefaultSampleRateHz = 48000; | |
| 32 constexpr int kId = 1; | |
| 33 | |
| 34 // Utility function that resets the frame member variables with | |
| 35 // sensible defaults. | |
| 36 void ResetFrame(AudioFrame* frame) { | |
| 37 frame->id_ = kId; | |
| 38 frame->sample_rate_hz_ = kDefaultSampleRateHz; | |
| 39 frame->num_channels_ = 1; | |
| 40 | |
| 41 // Frame duration 10ms. | |
| 42 frame->samples_per_channel_ = kDefaultSampleRateHz / 100; | |
| 43 frame->vad_activity_ = AudioFrame::kVadActive; | |
| 44 frame->speech_type_ = AudioFrame::kNormalSpeech; | |
| 45 } | |
| 46 | |
| 47 AudioFrame frame_for_mixing; | |
| 48 | |
| 49 } // namespace | |
| 50 | |
| 51 class MockMixerAudioSource : public AudioMixer::Source { | |
| 52 public: | |
| 53 MockMixerAudioSource() | |
| 54 : fake_audio_frame_info_(AudioMixer::Source::AudioFrameInfo::kNormal) { | |
| 55 ON_CALL(*this, GetAudioFrameWithInfo(_)) | |
| 56 .WillByDefault( | |
| 57 Invoke(this, &MockMixerAudioSource::FakeAudioFrameWithInfo)); | |
| 58 } | |
| 59 | |
| 60 MOCK_METHOD1(GetAudioFrameWithInfo, AudioFrameWithInfo(int sample_rate_hz)); | |
| 61 | |
| 62 AudioFrame* fake_frame() { return &fake_frame_; } | |
| 63 AudioFrameInfo fake_info() { return fake_audio_frame_info_; } | |
| 64 void set_fake_info(const AudioFrameInfo audio_frame_info) { | |
| 65 fake_audio_frame_info_ = audio_frame_info; | |
| 66 } | |
| 67 | |
| 68 private: | |
| 69 AudioFrame fake_frame_, fake_output_frame_; | |
| 70 AudioFrameInfo fake_audio_frame_info_; | |
| 71 AudioFrameWithInfo FakeAudioFrameWithInfo(int sample_rate_hz) { | |
| 72 fake_output_frame_.CopyFrom(fake_frame_); | |
| 73 return { | |
| 74 &fake_output_frame_, // audio_frame_pointer | |
| 75 fake_info(), // audio_frame_info | |
| 76 }; | |
| 77 } | |
| 78 }; | |
| 79 | |
| 80 // Creates participants from |frames| and |frame_info| and adds them | |
| 81 // to the mixer. Compares mixed status with |expected_status| | |
| 82 void MixAndCompare( | |
| 83 const std::vector<AudioFrame>& frames, | |
| 84 const std::vector<AudioMixer::Source::AudioFrameInfo>& frame_info, | |
| 85 const std::vector<bool>& expected_status) { | |
| 86 int num_audio_sources = frames.size(); | |
| 87 RTC_DCHECK(frames.size() == frame_info.size()); | |
| 88 RTC_DCHECK(frame_info.size() == expected_status.size()); | |
| 89 | |
| 90 const std::unique_ptr<AudioMixerImpl> mixer(AudioMixerImpl::Create()); | |
| 91 std::vector<MockMixerAudioSource> participants(num_audio_sources); | |
| 92 | |
| 93 for (int i = 0; i < num_audio_sources; i++) { | |
| 94 participants[i].fake_frame()->CopyFrom(frames[i]); | |
| 95 participants[i].set_fake_info(frame_info[i]); | |
| 96 } | |
| 97 | |
| 98 for (int i = 0; i < num_audio_sources; i++) { | |
| 99 EXPECT_EQ(0, mixer->SetMixabilityStatus(&participants[i], true)); | |
| 100 EXPECT_CALL(participants[i], GetAudioFrameWithInfo(kDefaultSampleRateHz)) | |
| 101 .Times(Exactly(1)); | |
| 102 } | |
| 103 | |
| 104 mixer->Mix(kDefaultSampleRateHz, 1, &frame_for_mixing); | |
| 105 | |
| 106 for (int i = 0; i < num_audio_sources; i++) { | |
| 107 EXPECT_EQ(expected_status[i], | |
| 108 mixer->GetAudioSourceMixabilityStatusForTest(&participants[i])) | |
| 109 << "Mixed status of AudioSource #" << i << " wrong."; | |
| 110 } | |
| 111 } | |
| 112 | |
| 113 TEST(AudioMixer, LargestEnergyVadActiveMixed) { | |
| 114 constexpr int kAudioSources = | |
| 115 AudioMixer::kMaximumAmountOfMixedAudioSources + 3; | |
| 116 | |
| 117 const std::unique_ptr<AudioMixerImpl> mixer(AudioMixerImpl::Create()); | |
| 118 | |
| 119 MockMixerAudioSource participants[kAudioSources]; | |
| 120 | |
| 121 for (int i = 0; i < kAudioSources; ++i) { | |
| 122 ResetFrame(participants[i].fake_frame()); | |
| 123 | |
| 124 // We set the 80-th sample value since the first 80 samples may be | |
| 125 // modified by a ramped-in window. | |
| 126 participants[i].fake_frame()->data_[80] = i; | |
| 127 | |
| 128 EXPECT_EQ(0, mixer->SetMixabilityStatus(&participants[i], true)); | |
| 129 EXPECT_CALL(participants[i], GetAudioFrameWithInfo(_)).Times(Exactly(1)); | |
| 130 } | |
| 131 | |
| 132 // Last participant gives audio frame with passive VAD, although it has the | |
| 133 // largest energy. | |
| 134 participants[kAudioSources - 1].fake_frame()->vad_activity_ = | |
| 135 AudioFrame::kVadPassive; | |
| 136 | |
| 137 AudioFrame audio_frame; | |
| 138 mixer->Mix(kDefaultSampleRateHz, | |
| 139 1, // number of channels | |
| 140 &audio_frame); | |
| 141 | |
| 142 for (int i = 0; i < kAudioSources; ++i) { | |
| 143 bool is_mixed = | |
| 144 mixer->GetAudioSourceMixabilityStatusForTest(&participants[i]); | |
| 145 if (i == kAudioSources - 1 || | |
| 146 i < kAudioSources - 1 - AudioMixer::kMaximumAmountOfMixedAudioSources) { | |
| 147 EXPECT_FALSE(is_mixed) << "Mixing status of AudioSource #" << i | |
| 148 << " wrong."; | |
| 149 } else { | |
| 150 EXPECT_TRUE(is_mixed) << "Mixing status of AudioSource #" << i | |
| 151 << " wrong."; | |
| 152 } | |
| 153 } | |
| 154 } | |
| 155 | |
| 156 TEST(AudioMixer, FrameNotModifiedForSingleParticipant) { | |
| 157 const std::unique_ptr<AudioMixer> mixer(AudioMixer::Create()); | |
| 158 | |
| 159 MockMixerAudioSource participant; | |
| 160 | |
| 161 ResetFrame(participant.fake_frame()); | |
| 162 const int n_samples = participant.fake_frame()->samples_per_channel_; | |
| 163 | |
| 164 // Modify the frame so that it's not zero. | |
| 165 for (int j = 0; j < n_samples; j++) { | |
| 166 participant.fake_frame()->data_[j] = j; | |
| 167 } | |
| 168 | |
| 169 EXPECT_EQ(0, mixer->SetMixabilityStatus(&participant, true)); | |
| 170 EXPECT_CALL(participant, GetAudioFrameWithInfo(_)).Times(Exactly(2)); | |
| 171 | |
| 172 AudioFrame audio_frame; | |
| 173 // Two mix iteration to compare after the ramp-up step. | |
| 174 for (int i = 0; i < 2; i++) { | |
| 175 mixer->Mix(kDefaultSampleRateHz, | |
| 176 1, // number of channels | |
| 177 &audio_frame); | |
| 178 } | |
| 179 | |
| 180 EXPECT_EQ( | |
| 181 0, memcmp(participant.fake_frame()->data_, audio_frame.data_, n_samples)); | |
| 182 } | |
| 183 | |
| 184 TEST(AudioMixer, ParticipantSampleRate) { | |
| 185 const std::unique_ptr<AudioMixer> mixer(AudioMixer::Create()); | |
| 186 | |
| 187 MockMixerAudioSource participant; | |
| 188 ResetFrame(participant.fake_frame()); | |
| 189 | |
| 190 EXPECT_EQ(0, mixer->SetMixabilityStatus(&participant, true)); | |
| 191 for (auto frequency : {8000, 16000, 32000, 48000}) { | |
| 192 EXPECT_CALL(participant, GetAudioFrameWithInfo(frequency)) | |
| 193 .Times(Exactly(1)); | |
| 194 participant.fake_frame()->sample_rate_hz_ = frequency; | |
| 195 participant.fake_frame()->samples_per_channel_ = frequency / 100; | |
| 196 mixer->Mix(frequency, 1, &frame_for_mixing); | |
| 197 EXPECT_EQ(frequency, frame_for_mixing.sample_rate_hz_); | |
| 198 } | |
| 199 } | |
| 200 | |
| 201 TEST(AudioMixer, ParticipantNumberOfChannels) { | |
| 202 const std::unique_ptr<AudioMixer> mixer(AudioMixer::Create()); | |
| 203 | |
| 204 MockMixerAudioSource participant; | |
| 205 ResetFrame(participant.fake_frame()); | |
| 206 | |
| 207 EXPECT_EQ(0, mixer->SetMixabilityStatus(&participant, true)); | |
| 208 for (size_t number_of_channels : {1, 2}) { | |
| 209 EXPECT_CALL(participant, GetAudioFrameWithInfo(kDefaultSampleRateHz)) | |
| 210 .Times(Exactly(1)); | |
| 211 mixer->Mix(kDefaultSampleRateHz, number_of_channels, &frame_for_mixing); | |
| 212 EXPECT_EQ(number_of_channels, frame_for_mixing.num_channels_); | |
| 213 } | |
| 214 } | |
| 215 | |
| 216 // Maximal amount of participants are mixed one iteration, then | |
| 217 // another participant with higher energy is added. | |
| 218 TEST(AudioMixer, RampedOutSourcesShouldNotBeMarkedMixed) { | |
| 219 constexpr int kAudioSources = | |
| 220 AudioMixer::kMaximumAmountOfMixedAudioSources + 1; | |
| 221 | |
| 222 const std::unique_ptr<AudioMixerImpl> mixer(AudioMixerImpl::Create()); | |
| 223 MockMixerAudioSource participants[kAudioSources]; | |
| 224 | |
| 225 for (int i = 0; i < kAudioSources; i++) { | |
| 226 ResetFrame(participants[i].fake_frame()); | |
| 227 // Set the participant audio energy to increase with the index | |
| 228 // |i|. | |
| 229 participants[i].fake_frame()->data_[0] = 100 * i; | |
| 230 } | |
| 231 | |
| 232 // Add all participants but the loudest for mixing. | |
| 233 for (int i = 0; i < kAudioSources - 1; i++) { | |
| 234 EXPECT_EQ(0, mixer->SetMixabilityStatus(&participants[i], true)); | |
| 235 EXPECT_CALL(participants[i], GetAudioFrameWithInfo(kDefaultSampleRateHz)) | |
| 236 .Times(Exactly(1)); | |
| 237 } | |
| 238 | |
| 239 // First mixer iteration | |
| 240 mixer->Mix(kDefaultSampleRateHz, 1, &frame_for_mixing); | |
| 241 | |
| 242 // All participants but the loudest should have been mixed. | |
| 243 for (int i = 0; i < kAudioSources - 1; i++) { | |
| 244 EXPECT_TRUE(mixer->GetAudioSourceMixabilityStatusForTest(&participants[i])) | |
| 245 << "Mixed status of AudioSource #" << i << " wrong."; | |
| 246 } | |
| 247 | |
| 248 // Add new participant with higher energy. | |
| 249 EXPECT_EQ(0, | |
| 250 mixer->SetMixabilityStatus(&participants[kAudioSources - 1], true)); | |
| 251 for (int i = 0; i < kAudioSources; i++) { | |
| 252 EXPECT_CALL(participants[i], GetAudioFrameWithInfo(kDefaultSampleRateHz)) | |
| 253 .Times(Exactly(1)); | |
| 254 } | |
| 255 | |
| 256 mixer->Mix(kDefaultSampleRateHz, 1, &frame_for_mixing); | |
| 257 | |
| 258 // The most quiet participant should not have been mixed. | |
| 259 EXPECT_FALSE(mixer->GetAudioSourceMixabilityStatusForTest(&participants[0])) | |
| 260 << "Mixed status of AudioSource #0 wrong."; | |
| 261 | |
| 262 // The loudest participants should have been mixed. | |
| 263 for (int i = 1; i < kAudioSources; i++) { | |
| 264 EXPECT_EQ(true, | |
| 265 mixer->GetAudioSourceMixabilityStatusForTest(&participants[i])) | |
| 266 << "Mixed status of AudioSource #" << i << " wrong."; | |
| 267 } | |
| 268 } | |
| 269 | |
| 270 // This test checks that the initialization and participant addition | |
| 271 // can be done on a different thread. | |
| 272 TEST(AudioMixer, ConstructFromOtherThread) { | |
| 273 std::unique_ptr<rtc::Thread> init_thread = rtc::Thread::Create(); | |
| 274 std::unique_ptr<rtc::Thread> participant_thread = rtc::Thread::Create(); | |
| 275 init_thread->Start(); | |
| 276 std::unique_ptr<AudioMixer> mixer( | |
| 277 init_thread->Invoke<std::unique_ptr<AudioMixer>>( | |
| 278 RTC_FROM_HERE, &AudioMixer::Create)); | |
| 279 MockMixerAudioSource participant; | |
| 280 | |
| 281 ResetFrame(participant.fake_frame()); | |
| 282 | |
| 283 participant_thread->Start(); | |
| 284 EXPECT_EQ(0, participant_thread->Invoke<int>( | |
| 285 RTC_FROM_HERE, rtc::Bind(&AudioMixer::SetMixabilityStatus, | |
| 286 mixer.get(), &participant, true))); | |
| 287 | |
| 288 EXPECT_CALL(participant, GetAudioFrameWithInfo(kDefaultSampleRateHz)) | |
| 289 .Times(Exactly(1)); | |
| 290 | |
| 291 // Do one mixer iteration | |
| 292 mixer->Mix(kDefaultSampleRateHz, 1, &frame_for_mixing); | |
| 293 } | |
| 294 | |
| 295 TEST(AudioMixer, MutedShouldMixAfterUnmuted) { | |
| 296 constexpr int kAudioSources = | |
| 297 AudioMixer::kMaximumAmountOfMixedAudioSources + 1; | |
| 298 | |
| 299 std::vector<AudioFrame> frames(kAudioSources); | |
| 300 for (auto& frame : frames) { | |
| 301 ResetFrame(&frame); | |
| 302 } | |
| 303 | |
| 304 std::vector<AudioMixer::Source::AudioFrameInfo> frame_info( | |
| 305 kAudioSources, AudioMixer::Source::AudioFrameInfo::kNormal); | |
| 306 frame_info[0] = AudioMixer::Source::AudioFrameInfo::kMuted; | |
| 307 std::vector<bool> expected_status(kAudioSources, true); | |
| 308 expected_status[0] = false; | |
| 309 | |
| 310 MixAndCompare(frames, frame_info, expected_status); | |
| 311 } | |
| 312 | |
| 313 TEST(AudioMixer, PassiveShouldMixAfterNormal) { | |
| 314 constexpr int kAudioSources = | |
| 315 AudioMixer::kMaximumAmountOfMixedAudioSources + 1; | |
| 316 | |
| 317 std::vector<AudioFrame> frames(kAudioSources); | |
| 318 for (auto& frame : frames) { | |
| 319 ResetFrame(&frame); | |
| 320 } | |
| 321 | |
| 322 std::vector<AudioMixer::Source::AudioFrameInfo> frame_info( | |
| 323 kAudioSources, AudioMixer::Source::AudioFrameInfo::kNormal); | |
| 324 frames[0].vad_activity_ = AudioFrame::kVadPassive; | |
| 325 std::vector<bool> expected_status(kAudioSources, true); | |
| 326 expected_status[0] = false; | |
| 327 | |
| 328 MixAndCompare(frames, frame_info, expected_status); | |
| 329 } | |
| 330 | |
| 331 TEST(AudioMixer, ActiveShouldMixBeforeLoud) { | |
| 332 constexpr int kAudioSources = | |
| 333 AudioMixer::kMaximumAmountOfMixedAudioSources + 1; | |
| 334 | |
| 335 std::vector<AudioFrame> frames(kAudioSources); | |
| 336 for (auto& frame : frames) { | |
| 337 ResetFrame(&frame); | |
| 338 } | |
| 339 | |
| 340 std::vector<AudioMixer::Source::AudioFrameInfo> frame_info( | |
| 341 kAudioSources, AudioMixer::Source::AudioFrameInfo::kNormal); | |
| 342 frames[0].vad_activity_ = AudioFrame::kVadPassive; | |
| 343 std::fill(frames[0].data_, frames[0].data_ + kDefaultSampleRateHz / 100, | |
| 344 std::numeric_limits<int16_t>::max()); | |
| 345 std::vector<bool> expected_status(kAudioSources, true); | |
| 346 expected_status[0] = false; | |
| 347 | |
| 348 MixAndCompare(frames, frame_info, expected_status); | |
| 349 } | |
| 350 | |
| 351 TEST(AudioMixer, UnmutedShouldMixBeforeLoud) { | |
| 352 constexpr int kAudioSources = | |
| 353 AudioMixer::kMaximumAmountOfMixedAudioSources + 1; | |
| 354 | |
| 355 std::vector<AudioFrame> frames(kAudioSources); | |
| 356 for (auto& frame : frames) { | |
| 357 ResetFrame(&frame); | |
| 358 } | |
| 359 | |
| 360 std::vector<AudioMixer::Source::AudioFrameInfo> frame_info( | |
| 361 kAudioSources, AudioMixer::Source::AudioFrameInfo::kNormal); | |
| 362 frame_info[0] = AudioMixer::Source::AudioFrameInfo::kMuted; | |
| 363 std::fill(frames[0].data_, frames[0].data_ + kDefaultSampleRateHz / 100, | |
| 364 std::numeric_limits<int16_t>::max()); | |
| 365 std::vector<bool> expected_status(kAudioSources, true); | |
| 366 expected_status[0] = false; | |
| 367 | |
| 368 MixAndCompare(frames, frame_info, expected_status); | |
| 369 } | |
| 370 } // namespace webrtc | |
| OLD | NEW |