summaryrefslogtreecommitdiffstats
path: root/dom/media/gtest/TestAudioInputProcessing.cpp
blob: d21c37a900f0044ac53e50376430ac662c01a1ea (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
/* -*- Mode: C++; tab-width: 2; indent-tabs-mode: nil; c-basic-offset: 2 -*-*/
/* This Source Code Form is subject to the terms of the Mozilla Public
 * License, v. 2.0. If a copy of the MPL was not distributed with this file,
 * You can obtain one at http://mozilla.org/MPL/2.0/. */

#include "gmock/gmock.h"
#include "gtest/gtest.h"

#include "AudioGenerator.h"
#include "MediaEngineWebRTCAudio.h"
#include "MediaTrackGraphImpl.h"
#include "PrincipalHandle.h"
#include "mozilla/Attributes.h"
#include "mozilla/NullPrincipal.h"
#include "mozilla/UniquePtr.h"
#include "mozilla/Unused.h"
#include "nsContentUtils.h"
#include "nsTArray.h"

using namespace mozilla;
using testing::NiceMock;
using testing::Return;

class MockGraph : public MediaTrackGraphImpl {
 public:
  explicit MockGraph(TrackRate aRate)
      : MediaTrackGraphImpl(0, aRate, nullptr, AbstractThread::MainThread()) {
    ON_CALL(*this, OnGraphThread).WillByDefault(Return(true));
  }

  void Init(uint32_t aChannels) {
    MediaTrackGraphImpl::Init(OFFLINE_THREAD_DRIVER, DIRECT_DRIVER, aChannels);

    MonitorAutoLock lock(mMonitor);
    // We don't need a graph driver.  Advance to
    // LIFECYCLE_WAITING_FOR_TRACK_DESTRUCTION so that the driver never
    // starts.  Graph control messages run as in shutdown, synchronously.
    // This permits the main thread part of track initialization through
    // AudioProcessingTrack::Create().
    mLifecycleState = LIFECYCLE_WAITING_FOR_TRACK_DESTRUCTION;
#ifdef DEBUG
    mCanRunMessagesSynchronously = true;
#endif
    // Remove this graph's driver since it holds a ref. We are still kept
    // alive by the self-ref. Destroy() must be called to break that cycle if
    // no tracks are created and destroyed.
    mDriver = nullptr;
  }

  MOCK_CONST_METHOD0(OnGraphThread, bool());

 protected:
  ~MockGraph() = default;
};

// AudioInputProcessing will put extra frames as pre-buffering data to avoid
// glitchs in non pass-through mode. The main goal of the test is to check how
// many frames left in the AudioInputProcessing's mSegment in various situations
// after input data has been processed.
TEST(TestAudioInputProcessing, Buffering)
{
  const TrackRate rate = 8000;  // So packet size is 80
  const uint32_t channels = 1;
  auto graph = MakeRefPtr<NiceMock<MockGraph>>(rate);
  graph->Init(channels);
  RefPtr track = AudioProcessingTrack::Create(graph);

  auto aip = MakeRefPtr<AudioInputProcessing>(channels);

  const size_t frames = 72;

  AudioGenerator<AudioDataValue> generator(channels, rate);
  GraphTime processedTime;
  GraphTime nextTime;
  AudioSegment output;
  MediaEnginePrefs settings;
  settings.mChannels = channels;
  // pref "media.getusermedia.agc2_forced" defaults to true.
  // mAgc would need to be set to something other than kAdaptiveAnalog
  // for mobile, as asserted in AudioInputProcessing::ConfigForPrefs,
  // if gain_controller1 were used.
  settings.mAgc2Forced = true;

  // Toggle pass-through mode without starting
  {
    EXPECT_EQ(aip->IsPassThrough(graph), true);
    EXPECT_EQ(aip->NumBufferedFrames(graph), 0);

    settings.mAgcOn = true;
    aip->ApplySettings(graph, nullptr, settings);
    EXPECT_EQ(aip->IsPassThrough(graph), false);
    EXPECT_EQ(aip->NumBufferedFrames(graph), 0);

    settings.mAgcOn = false;
    aip->ApplySettings(graph, nullptr, settings);
    EXPECT_EQ(aip->NumBufferedFrames(graph), 0);
  }

  {
    // Need (nextTime - processedTime) = 128 - 0 = 128 frames this round.
    // aip has not started and set to processing mode yet, so output will be
    // filled with silence data directly.
    processedTime = 0;
    nextTime = MediaTrackGraphImpl::RoundUpToEndOfAudioBlock(frames);

    AudioSegment input;
    generator.Generate(input, nextTime - processedTime);

    aip->Process(track, processedTime, nextTime, &input, &output);
    EXPECT_EQ(input.GetDuration(), nextTime - processedTime);
    EXPECT_EQ(output.GetDuration(), nextTime);
    EXPECT_EQ(aip->NumBufferedFrames(graph), 0);
  }

  // Set aip to processing/non-pass-through mode
  settings.mAgcOn = true;
  aip->ApplySettings(graph, nullptr, settings);
  {
    // Need (nextTime - processedTime) = 256 - 128 = 128 frames this round.
    // aip has not started yet, so output will be filled with silence data
    // directly.
    processedTime = nextTime;
    nextTime = MediaTrackGraphImpl::RoundUpToEndOfAudioBlock(2 * frames);

    AudioSegment input;
    generator.Generate(input, nextTime - processedTime);

    aip->Process(track, processedTime, nextTime, &input, &output);
    EXPECT_EQ(input.GetDuration(), nextTime - processedTime);
    EXPECT_EQ(output.GetDuration(), nextTime);
    EXPECT_EQ(aip->NumBufferedFrames(graph), 0);
  }

  // aip has been set to processing mode and is started.
  aip->Start(graph);
  {
    // Need (nextTime - processedTime) = 256 - 256 = 0 frames this round.
    // Process() will return early on 0 frames of input.
    // Pre-buffering is not triggered.
    processedTime = nextTime;
    nextTime = MediaTrackGraphImpl::RoundUpToEndOfAudioBlock(3 * frames);

    AudioSegment input;
    generator.Generate(input, nextTime - processedTime);

    aip->Process(track, processedTime, nextTime, &input, &output);
    EXPECT_EQ(input.GetDuration(), nextTime - processedTime);
    EXPECT_EQ(output.GetDuration(), nextTime);
    EXPECT_EQ(aip->NumBufferedFrames(graph), 0);
  }

  {
    // Need (nextTime - processedTime) = 384 - 256 = 128 frames this round.
    // On receipt of the these first frames, aip will insert 80 frames
    // into its internal buffer as pre-buffering.
    // Process() will take 128 frames from input, packetize and process
    // these frames into floor(128/80) = 1 80-frame packet (48 frames left in
    // packetizer), insert packets into aip's internal buffer, then move 128
    // frames the internal buffer to output, leaving 80 + 80 - 128 = 32 frames
    // in aip's internal buffer.
    processedTime = nextTime;
    nextTime = MediaTrackGraphImpl::RoundUpToEndOfAudioBlock(4 * frames);

    AudioSegment input;
    generator.Generate(input, nextTime - processedTime);

    aip->Process(track, processedTime, nextTime, &input, &output);
    EXPECT_EQ(input.GetDuration(), nextTime - processedTime);
    EXPECT_EQ(output.GetDuration(), nextTime);
    EXPECT_EQ(aip->NumBufferedFrames(graph), 32);
  }

  {
    // Need (nextTime - processedTime) = 384 - 384 = 0 frames this round.
    processedTime = nextTime;
    nextTime = MediaTrackGraphImpl::RoundUpToEndOfAudioBlock(5 * frames);

    AudioSegment input;
    generator.Generate(input, nextTime - processedTime);

    aip->Process(track, processedTime, nextTime, &input, &output);
    EXPECT_EQ(input.GetDuration(), nextTime - processedTime);
    EXPECT_EQ(output.GetDuration(), nextTime);
    EXPECT_EQ(aip->NumBufferedFrames(graph), 32);
  }

  {
    // Need (nextTime - processedTime) = 512 - 384 = 128 frames this round.
    // The Process() aip will take 128 frames from input, packetize and process
    // these frames into floor(128+48/80) = 2 80-frame packet (16 frames left in
    // packetizer), insert packets into aip's internal buffer, then move 128
    // frames the internal buffer to output, leaving 32 + 2*80 - 128 = 64 frames
    // in aip's internal buffer.
    processedTime = nextTime;
    nextTime = MediaTrackGraphImpl::RoundUpToEndOfAudioBlock(6 * frames);

    AudioSegment input;
    generator.Generate(input, nextTime - processedTime);

    aip->Process(track, processedTime, nextTime, &input, &output);
    EXPECT_EQ(input.GetDuration(), nextTime - processedTime);
    EXPECT_EQ(output.GetDuration(), nextTime);
    EXPECT_EQ(aip->NumBufferedFrames(graph), 64);
  }

  // Set aip to pass-through mode
  settings.mAgcOn = false;
  aip->ApplySettings(graph, nullptr, settings);
  {
    // Need (nextTime - processedTime) = 512 - 512 = 0 frames this round.
    // No buffering in pass-through mode
    processedTime = nextTime;
    nextTime = MediaTrackGraphImpl::RoundUpToEndOfAudioBlock(7 * frames);

    AudioSegment input;
    generator.Generate(input, nextTime - processedTime);

    aip->Process(track, processedTime, nextTime, &input, &output);
    EXPECT_EQ(input.GetDuration(), nextTime - processedTime);
    EXPECT_EQ(output.GetDuration(), processedTime);
    EXPECT_EQ(aip->NumBufferedFrames(graph), 0);
  }

  aip->Stop(graph);
  track->Destroy();
}

TEST(TestAudioInputProcessing, ProcessDataWithDifferentPrincipals)
{
  const TrackRate rate = 48000;  // so # of output frames from packetizer is 480
  const uint32_t channels = 2;
  auto graph = MakeRefPtr<NiceMock<MockGraph>>(rate);
  graph->Init(channels);
  RefPtr track = AudioProcessingTrack::Create(graph);

  auto aip = MakeRefPtr<AudioInputProcessing>(channels);
  AudioGenerator<AudioDataValue> generator(channels, rate);

  RefPtr<nsIPrincipal> dummy_principal =
      NullPrincipal::CreateWithoutOriginAttributes();
  const PrincipalHandle principal1 = MakePrincipalHandle(dummy_principal.get());
  const PrincipalHandle principal2 =
      MakePrincipalHandle(nsContentUtils::GetSystemPrincipal());

  // Total 4800 frames. It's easier to test with frames of multiples of 480.
  nsTArray<std::pair<TrackTime, PrincipalHandle>> framesWithPrincipal = {
      {100, principal1},
      {200, PRINCIPAL_HANDLE_NONE},
      {300, principal2},
      {400, principal1},
      {440, PRINCIPAL_HANDLE_NONE},
      // 3 packet-size above.
      {480, principal1},
      {480, principal2},
      {480, PRINCIPAL_HANDLE_NONE},
      // 3 packet-size above.
      {500, principal2},
      {490, principal1},
      {600, principal1},
      {330, principal1}
      // 4 packet-size above.
  };

  // Generate 4800 frames of data with different principals.
  AudioSegment input;
  {
    for (const auto& [duration, principal] : framesWithPrincipal) {
      AudioSegment data;
      generator.Generate(data, duration);
      for (AudioSegment::ChunkIterator it(data); !it.IsEnded(); it.Next()) {
        it->mPrincipalHandle = principal;
      }

      input.AppendFrom(&data);
    }
  }

  auto verifyPrincipals = [&](const AudioSegment& data) {
    TrackTime start = 0;
    for (const auto& [duration, principal] : framesWithPrincipal) {
      const TrackTime end = start + duration;

      AudioSegment slice;
      slice.AppendSlice(data, start, end);
      start = end;

      for (AudioSegment::ChunkIterator it(slice); !it.IsEnded(); it.Next()) {
        EXPECT_EQ(it->mPrincipalHandle, principal);
      }
    }
  };

  // Check the principals in audio-processing mode.
  MediaEnginePrefs settings;
  settings.mChannels = channels;
  settings.mAgcOn = true;
  settings.mAgc2Forced = true;
  aip->ApplySettings(graph, nullptr, settings);
  EXPECT_EQ(aip->IsPassThrough(graph), false);
  aip->Start(graph);
  {
    AudioSegment output;
    {
      AudioSegment data;
      aip->Process(track, 0, 4800, &input, &data);
      EXPECT_EQ(input.GetDuration(), 4800);
      EXPECT_EQ(data.GetDuration(), 4800);

      // Extract another 480 frames to account for delay from pre-buffering.
      EXPECT_EQ(aip->NumBufferedFrames(graph), 480);
      AudioSegment dummy;
      dummy.AppendNullData(480);
      aip->Process(track, 0, 480, &dummy, &data);
      EXPECT_EQ(dummy.GetDuration(), 480);
      EXPECT_EQ(data.GetDuration(), 480 + 4800);

      // Ignore the pre-buffering silence.
      output.AppendSlice(data, 480, 480 + 4800);
    }

    verifyPrincipals(output);
  }

  // Check the principals in pass-through mode.
  settings.mAgcOn = false;
  aip->ApplySettings(graph, nullptr, settings);
  EXPECT_EQ(aip->IsPassThrough(graph), true);
  {
    AudioSegment output;
    aip->Process(track, 0, 4800, &input, &output);
    EXPECT_EQ(input.GetDuration(), 4800);
    EXPECT_EQ(output.GetDuration(), 4800);

    verifyPrincipals(output);
  }

  aip->Stop(graph);
  track->Destroy();
}

TEST(TestAudioInputProcessing, Downmixing)
{
  const TrackRate rate = 44100;
  const uint32_t channels = 4;
  auto graph = MakeRefPtr<NiceMock<MockGraph>>(rate);
  graph->Init(channels);
  RefPtr track = AudioProcessingTrack::Create(graph);

  auto aip = MakeRefPtr<AudioInputProcessing>(channels);

  const size_t frames = 44100;

  AudioGenerator<AudioDataValue> generator(channels, rate);
  GraphTime processedTime;
  GraphTime nextTime;

  MediaEnginePrefs settings;
  settings.mChannels = channels;
  settings.mAgcOn = true;
  settings.mAgc2Forced = true;
  aip->ApplySettings(graph, nullptr, settings);
  EXPECT_EQ(aip->IsPassThrough(graph), false);
  aip->Start(graph);

  processedTime = 0;
  nextTime = MediaTrackGraphImpl::RoundUpToEndOfAudioBlock(frames);

  {
    AudioSegment input;
    AudioSegment output;
    generator.Generate(input, nextTime - processedTime);

    // Intentionally reduce the amplitude of the generated sine wave so there's
    // no chance the max amplitude reaches 1.0, but not enough so that 4
    // channels summed together won't clip.
    input.ApplyVolume(0.9);

    // Process is going to see that it has 4 channels of input, and is going to
    // downmix to mono, scaling the input by 1/4 in the process.
    // We can't compare the input and output signal because the sine is going to
    // be mangledui
    aip->Process(track, processedTime, nextTime, &input, &output);
    EXPECT_EQ(input.GetDuration(), nextTime - processedTime);
    EXPECT_EQ(output.GetDuration(), nextTime);
    EXPECT_EQ(output.MaxChannelCount(), 1u);

    // Verify that it doesn't clip: the input signal has likely been mangled by
    // the various processing passes, but at least it shouldn't clip. We know we
    // always have floating point audio here, regardless of the sample-type used
    // by Gecko.
    for (AudioSegment::ChunkIterator iterOutput(output); !iterOutput.IsEnded();
         iterOutput.Next()) {
      const float* const output = iterOutput->ChannelData<float>()[0];
      for (uint32_t i = 0; i < iterOutput->GetDuration(); i++) {
        // Very conservative here, it's likely that the AGC lowers the volume a
        // lot.
        EXPECT_LE(std::abs(output[i]), 0.95);
      }
    }
  }

  // Now, repeat the test in pass-through mode, checking we get the unmodified
  // 4 channels.
  settings.mAgcOn = false;
  aip->ApplySettings(graph, nullptr, settings);
  EXPECT_EQ(aip->IsPassThrough(graph), true);

  AudioSegment input, output;
  processedTime = nextTime;
  nextTime += MediaTrackGraphImpl::RoundUpToEndOfAudioBlock(frames);
  generator.Generate(input, nextTime - processedTime);

  aip->Process(track, processedTime, nextTime, &input, &output);
  EXPECT_EQ(input.GetDuration(), nextTime - processedTime);
  EXPECT_EQ(output.GetDuration(), nextTime - processedTime);
  // This time, no downmix: 4 channels of input, 4 channels of output
  EXPECT_EQ(output.MaxChannelCount(), 4u);

  nsTArray<AudioDataValue> inputLinearized, outputLinearized;
  input.WriteToInterleavedBuffer(inputLinearized, input.MaxChannelCount());
  output.WriteToInterleavedBuffer(outputLinearized, output.MaxChannelCount());

  // The data should be passed through, and exactly equal.
  for (uint32_t i = 0; i < frames * channels; i++) {
    EXPECT_EQ(inputLinearized[i], outputLinearized[i]);
  }

  aip->Stop(graph);
  track->Destroy();
}