Pull the Voice Activity Detector out from the AGC

webrtc/modules/audio_processing/vad/voice_activity_detector_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 "webrtc/modules/audio_processing/vad/voice_activity_detector.h"
+
+#include <algorithm>
+#include <vector>
+
+#include "testing/gtest/include/gtest/gtest.h"
+#include "webrtc/test/testsupport/fileutils.h"
+
+namespace webrtc {
+namespace {
+
+void GenerateNoise(std::vector<int16_t>* data) {
+  for (size_t i = 0; i < data->size(); ++i) {
+    // std::rand returns between 0 and RAND_MAX, but this will work because it
+    // wraps into some random place.
+    (*data)[i] = std::rand();
+  }
+}
+
+}  // namespace
+
+TEST(VoiceActivityDetectorTest, ConstructorSetsDefaultValues) {

[bloch, 2015/06/17 22:27:49]

Where do you test your resampling (/sample rate)? I know you're testing
higher-level functionality here, but is there somewhere you're testing edge
cases for these method parameters?

[aluebs-webrtc, 2015/06/18 00:49:22]

That is a good point. Added tests for 32kHz audio and noise. I know those fseek
are horrible, but these files have a lot of silences and clicks I want to avoid
to get a higher voice probability. It is that or adding special files to the
resources. WDYT?

+  const float kDefaultVoiceValue = 1.f;
+
+  VoiceActivityDetector vad;
+
+  std::vector<double> p = vad.chunkwise_voice_probabilities();
+  std::vector<double> rms = vad.chunkwise_rms();
+
+  EXPECT_EQ(p.size(), 0u);
+  EXPECT_EQ(rms.size(), 0u);
+
+  EXPECT_FLOAT_EQ(vad.last_voice_probability(), kDefaultVoiceValue);
+}
+
+TEST(VoiceActivityDetectorTest, SpeechHasHighVoiceProbabilities) {
+  const float kMeanSpeechProbability = 0.3f;
+
+  VoiceActivityDetector vad;
+
+  std::vector<int16_t> data(kLength10Ms);
+  float mean_probability = 0.f;
+
+  FILE* pcm_file =
+      fopen(test::ResourcePath("audio_processing/agc/agc_audio", "pcm").c_str(),
+            "rb");
+  ASSERT_TRUE(pcm_file != NULL);
+
+  size_t num_chunks = 0;
+  while (fread(&data[0], sizeof(data[0]), data.size(), pcm_file) ==
+         data.size()) {
+    vad.ProcessChunk(&data[0], data.size(), kSampleRateHz);
+
+    mean_probability += vad.last_voice_probability();
+
+    ++num_chunks;
+  }
+
+  mean_probability /= num_chunks;
+
+  EXPECT_GT(mean_probability, kMeanSpeechProbability);
+}
+
+TEST(VoiceActivityDetectorTest, NoiseHasLowVoiceProbabilities) {
+  const float kMaxNoiseProbability = 0.05f;
+  const size_t kNumChunks = 100u;
+  const size_t kNumChunksPerIsacBlock = 3;
+
+  VoiceActivityDetector vad;
+
+  std::vector<int16_t> data(kLength10Ms);
+  float max_probability = 0.f;
+
+  std::srand(42);
+
+  for (size_t i = 0; i < kNumChunks; ++i) {
+    GenerateNoise(&data);
+
+    vad.ProcessChunk(&data[0], data.size(), kSampleRateHz);
+
+    if (i > kNumChunksPerIsacBlock) {
+      max_probability = std::max(max_probability, vad.last_voice_probability());
+    }
+  }
+
+  EXPECT_LT(max_probability, kMaxNoiseProbability);
+}
+
+}  // namespace webrtc