OLD | NEW |
---|---|
(Empty) | |
1 /* | |
2 * Copyright (c) 2015 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 "webrtc/modules/audio_processing/vad/voice_activity_detector.h" | |
12 | |
13 #include <algorithm> | |
14 #include <vector> | |
15 | |
16 #include "testing/gtest/include/gtest/gtest.h" | |
17 #include "webrtc/test/testsupport/fileutils.h" | |
18 | |
19 namespace webrtc { | |
20 namespace { | |
21 | |
22 void GenerateNoise(std::vector<int16_t>* data) { | |
23 for (size_t i = 0; i < data->size(); ++i) { | |
24 // std::rand returns between 0 and RAND_MAX, but this will work because it | |
25 // wraps into some random place. | |
26 (*data)[i] = std::rand(); | |
27 } | |
28 } | |
29 | |
30 } // namespace | |
31 | |
32 TEST(VoiceActivityDetectorTest, ConstructorSetsDefaultValues) { | |
bloch
2015/06/17 22:27:49
Where do you test your resampling (/sample rate)?
aluebs-webrtc
2015/06/18 00:49:22
That is a good point. Added tests for 32kHz audio
| |
33 const float kDefaultVoiceValue = 1.f; | |
34 | |
35 VoiceActivityDetector vad; | |
36 | |
37 std::vector<double> p = vad.chunkwise_voice_probabilities(); | |
38 std::vector<double> rms = vad.chunkwise_rms(); | |
39 | |
40 EXPECT_EQ(p.size(), 0u); | |
41 EXPECT_EQ(rms.size(), 0u); | |
42 | |
43 EXPECT_FLOAT_EQ(vad.last_voice_probability(), kDefaultVoiceValue); | |
44 } | |
45 | |
46 TEST(VoiceActivityDetectorTest, SpeechHasHighVoiceProbabilities) { | |
47 const float kMeanSpeechProbability = 0.3f; | |
48 | |
49 VoiceActivityDetector vad; | |
50 | |
51 std::vector<int16_t> data(kLength10Ms); | |
52 float mean_probability = 0.f; | |
53 | |
54 FILE* pcm_file = | |
55 fopen(test::ResourcePath("audio_processing/agc/agc_audio", "pcm").c_str(), | |
56 "rb"); | |
57 ASSERT_TRUE(pcm_file != NULL); | |
58 | |
59 size_t num_chunks = 0; | |
60 while (fread(&data[0], sizeof(data[0]), data.size(), pcm_file) == | |
61 data.size()) { | |
62 vad.ProcessChunk(&data[0], data.size(), kSampleRateHz); | |
63 | |
64 mean_probability += vad.last_voice_probability(); | |
65 | |
66 ++num_chunks; | |
67 } | |
68 | |
69 mean_probability /= num_chunks; | |
70 | |
71 EXPECT_GT(mean_probability, kMeanSpeechProbability); | |
72 } | |
73 | |
74 TEST(VoiceActivityDetectorTest, NoiseHasLowVoiceProbabilities) { | |
75 const float kMaxNoiseProbability = 0.05f; | |
76 const size_t kNumChunks = 100u; | |
77 const size_t kNumChunksPerIsacBlock = 3; | |
78 | |
79 VoiceActivityDetector vad; | |
80 | |
81 std::vector<int16_t> data(kLength10Ms); | |
82 float max_probability = 0.f; | |
83 | |
84 std::srand(42); | |
85 | |
86 for (size_t i = 0; i < kNumChunks; ++i) { | |
87 GenerateNoise(&data); | |
88 | |
89 vad.ProcessChunk(&data[0], data.size(), kSampleRateHz); | |
90 | |
91 if (i > kNumChunksPerIsacBlock) { | |
92 max_probability = std::max(max_probability, vad.last_voice_probability()); | |
93 } | |
94 } | |
95 | |
96 EXPECT_LT(max_probability, kMaxNoiseProbability); | |
97 } | |
98 | |
99 } // namespace webrtc | |
OLD | NEW |