зеркало из https://github.com/mozilla/gecko-dev.git
Bug 1649817 - Verify audio buffers in drift correction tests, r=padenot
In AudioDriftCorrection gtests, the audio in the buffers is simulated by a sine tone of specific amplitude and frequency. This paych implements a way to verify automatically that the expected signal is being received in the output. Differential Revision: https://phabricator.services.mozilla.com/D81988
This commit is contained in:
Alex Chronopoulos
Родитель
53bed70ef3
Коммит
70eee99400
|
|
@ -199,13 +199,159 @@ TEST(TestClockDrift, SmallBufferedFrames)
|
|||
EXPECT_FLOAT_EQ(c.GetCorrection(), 1.1);
|
||||
}
|
||||
|
||||
template <typename T>
|
||||
class AudioToneGenerator {
|
||||
public:
|
||||
static_assert(std::is_same<T, int16_t>::value ||
|
||||
std::is_same<T, float>::value);
|
||||
|
||||
explicit AudioToneGenerator(int32_t aRate) : mRate(aRate) {
|
||||
MOZ_ASSERT(mRate > 0);
|
||||
}
|
||||
|
||||
void Write(AudioChunk& aChunk) {
|
||||
float time = mTime;
|
||||
for (uint32_t i = 0; i < aChunk.ChannelCount(); ++i) {
|
||||
mTime = time; // reset the time for each channel
|
||||
T* buffer = aChunk.ChannelDataForWrite<T>(0);
|
||||
Write(buffer, aChunk.GetDuration());
|
||||
}
|
||||
}
|
||||
|
||||
void Write(T* aBuffer, int32_t aFrames) {
|
||||
for (int i = 0; i < aFrames; ++i) {
|
||||
double value = Amplitude() * sin(2 * M_PI * mFrequency * mTime + mPhase);
|
||||
aBuffer[i] = static_cast<T>(value);
|
||||
mTime += mDeltaTime;
|
||||
}
|
||||
}
|
||||
|
||||
T MaxMagnitudeDifference() {
|
||||
return static_cast<T>(Amplitude() *
|
||||
sin(2 * M_PI * mFrequency * mDeltaTime + mPhase));
|
||||
}
|
||||
|
||||
const int32_t mRate;
|
||||
const int32_t mFrequency = 100;
|
||||
const float mPhase = 0.0;
|
||||
const float mDeltaTime = 1.0f / mRate;
|
||||
|
||||
static T Amplitude() {
|
||||
if (std::is_same<T, int16_t>::value) {
|
||||
return 19384; // int16_t::max / 2
|
||||
}
|
||||
return 0.5f; // 1.0f / 2
|
||||
}
|
||||
|
||||
private:
|
||||
float mTime = 0.0;
|
||||
};
|
||||
|
||||
template <typename T>
|
||||
class AudioToneVerifier {
|
||||
public:
|
||||
explicit AudioToneVerifier(int32_t aRate)
|
||||
: mRate(aRate), mExpectedTone(aRate) {
|
||||
MOZ_ASSERT(mRate > 0);
|
||||
}
|
||||
|
||||
// Only the mono channel is taken into account.
|
||||
void AppendData(const AudioSegment& segment) {
|
||||
for (AudioSegment::ConstChunkIterator iter(segment); !iter.IsEnded();
|
||||
iter.Next()) {
|
||||
const AudioChunk& c = *iter;
|
||||
CheckBuffer(c.ChannelData<T>()[0], c.GetDuration());
|
||||
}
|
||||
}
|
||||
|
||||
int32_t EstimatedFreq() const {
|
||||
if (mTotalFramesSoFar == PreSilenceSamples() || mZeroCrossCount <= 1) {
|
||||
return 0.0f;
|
||||
}
|
||||
MOZ_ASSERT(mZeroCrossCount > 1);
|
||||
return mRate / (mSumPeriodInSamples / (mZeroCrossCount - 1));
|
||||
}
|
||||
|
||||
int64_t PreSilenceSamples() const {
|
||||
// The first sample of the sinewave is zero.
|
||||
MOZ_ASSERT(mPreSilenceSamples >= 1);
|
||||
return mPreSilenceSamples - 1;
|
||||
}
|
||||
|
||||
int32_t CountDiscontinuities() const {
|
||||
// Every discontinuity is counted twice one on current sample and one more
|
||||
// on previous sample.
|
||||
return mDiscontinuitiesCount / 2;
|
||||
}
|
||||
|
||||
private:
|
||||
void CheckBuffer(const T* aBuffer, int32_t aSamples) {
|
||||
for (int i = 0; i < aSamples; ++i) {
|
||||
++mTotalFramesSoFar;
|
||||
// Avoid pre-silence
|
||||
if (!CountPreSilence(aBuffer[i])) {
|
||||
CountZeroCrossing(aBuffer[i]);
|
||||
CountDiscontinuities(aBuffer[i]);
|
||||
}
|
||||
|
||||
mPrevious = aBuffer[i];
|
||||
}
|
||||
}
|
||||
|
||||
bool CountPreSilence(T aCurrentSample) {
|
||||
if (IsZero(aCurrentSample) && mPreSilenceSamples == mTotalFramesSoFar - 1) {
|
||||
++mPreSilenceSamples;
|
||||
return true;
|
||||
}
|
||||
return false;
|
||||
}
|
||||
|
||||
// Positive to negative direction
|
||||
void CountZeroCrossing(T aCurrentSample) {
|
||||
if (mPrevious > 0 && aCurrentSample <= 0) {
|
||||
if (mZeroCrossCount++) {
|
||||
MOZ_ASSERT(mZeroCrossCount > 1);
|
||||
mSumPeriodInSamples += mTotalFramesSoFar - mLastZeroCrossPosition;
|
||||
}
|
||||
mLastZeroCrossPosition = mTotalFramesSoFar;
|
||||
}
|
||||
}
|
||||
|
||||
void CountDiscontinuities(T aCurrentSample) {
|
||||
mDiscontinuitiesCount += fabs(fabs(aCurrentSample) - fabs(mPrevious)) >
|
||||
2 * mExpectedTone.MaxMagnitudeDifference();
|
||||
}
|
||||
|
||||
bool IsZero(float a) { return fabs(a) < 1e-8; }
|
||||
bool IsZero(short a) { return a == 0; }
|
||||
|
||||
private:
|
||||
const int32_t mRate;
|
||||
AudioToneGenerator<T> mExpectedTone;
|
||||
|
||||
int32_t mZeroCrossCount = 0;
|
||||
int64_t mLastZeroCrossPosition = 0;
|
||||
int64_t mSumPeriodInSamples = 0;
|
||||
|
||||
int64_t mTotalFramesSoFar = 0;
|
||||
int64_t mPreSilenceSamples = 0;
|
||||
|
||||
int32_t mDiscontinuitiesCount = 0;
|
||||
// This is needed to connect previous the previous buffers.
|
||||
T mPrevious = {};
|
||||
};
|
||||
|
||||
// Print the mono channel of a segment.
|
||||
void printAudioSegment(const AudioSegment& segment) {
|
||||
for (AudioSegment::ConstChunkIterator iter(segment); !iter.IsEnded();
|
||||
iter.Next()) {
|
||||
const AudioChunk& c = *iter;
|
||||
const float* buffer = c.ChannelData<float>()[0];
|
||||
for (int i = 0; i < c.GetDuration(); ++i) {
|
||||
printf("%f\n", buffer[i]);
|
||||
if (c.mBufferFormat == AUDIO_FORMAT_FLOAT32) {
|
||||
printf("%f\n", c.ChannelData<float>()[0][i]);
|
||||
} else {
|
||||
printf("%d\n", c.ChannelData<int16_t>()[0][i]);
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
|
@ -219,11 +365,9 @@ void testAudioCorrection(int32_t aSourceRate, int32_t aTargetRate) {
|
|||
const int32_t sampleRateReceiver = aTargetRate;
|
||||
AudioDriftCorrection ad(sampleRateTransmitter, sampleRateReceiver);
|
||||
|
||||
const float amplitude = 0.5;
|
||||
const float frequency = 10;
|
||||
const float phase = 0.0;
|
||||
float time = 0.0;
|
||||
const float deltaTime = 1.0f / sampleRateTransmitter;
|
||||
AudioToneGenerator<AudioDataValue> tone(sampleRateTransmitter);
|
||||
AudioToneVerifier<AudioDataValue> inToneVerifier(sampleRateTransmitter);
|
||||
AudioToneVerifier<AudioDataValue> outToneVerifier(sampleRateReceiver);
|
||||
|
||||
int32_t sourceFrames;
|
||||
const int32_t targetFrames = sampleRateReceiver / 100;
|
||||
|
|
@ -239,16 +383,12 @@ void testAudioCorrection(int32_t aSourceRate, int32_t aTargetRate) {
|
|||
|
||||
for (int n = 0; n < 250; ++n) {
|
||||
// Create the input (sine tone)
|
||||
AudioChunk chunk =
|
||||
CreateAudioChunk<float>(sourceFrames, 1, AUDIO_FORMAT_FLOAT32);
|
||||
float* monoBuffer = chunk.ChannelDataForWrite<float>(0);
|
||||
for (int i = 0; i < sourceFrames; ++i) {
|
||||
double value = amplitude * sin(2 * M_PI * frequency * time + phase);
|
||||
monoBuffer[i] = static_cast<float>(value);
|
||||
time += deltaTime;
|
||||
}
|
||||
AudioChunk chunk = CreateAudioChunk<AudioDataValue>(sourceFrames, 1,
|
||||
AUDIO_OUTPUT_FORMAT);
|
||||
tone.Write(chunk);
|
||||
AudioSegment inSegment;
|
||||
inSegment.AppendAndConsumeChunk(&chunk);
|
||||
inToneVerifier.AppendData(inSegment);
|
||||
// Print the input for debugging
|
||||
// printAudioSegment(inSegment);
|
||||
|
||||
|
|
@ -257,8 +397,18 @@ void testAudioCorrection(int32_t aSourceRate, int32_t aTargetRate) {
|
|||
EXPECT_EQ(outSegment.GetDuration(), targetFrames);
|
||||
// Print the output for debugging
|
||||
// printAudioSegment(outSegment);
|
||||
outToneVerifier.AppendData(outSegment);
|
||||
}
|
||||
}
|
||||
EXPECT_EQ(inToneVerifier.EstimatedFreq(), tone.mFrequency);
|
||||
EXPECT_EQ(inToneVerifier.PreSilenceSamples(), 0);
|
||||
EXPECT_EQ(inToneVerifier.CountDiscontinuities(), 0);
|
||||
|
||||
EXPECT_EQ(outToneVerifier.EstimatedFreq(), tone.mFrequency);
|
||||
// The expected pre-silence is 50ms plus the resampling, this is roughly more
|
||||
// than 2000 frames for the samples rates being used here
|
||||
EXPECT_GT(outToneVerifier.PreSilenceSamples(), 2000);
|
||||
EXPECT_EQ(outToneVerifier.CountDiscontinuities(), 0);
|
||||
}
|
||||
|
||||
TEST(TestAudioDriftCorrection, Basic)
|
||||
|
|
@ -273,11 +423,9 @@ void testMonoToStereoInput(int aSourceRate, int aTargetRate) {
|
|||
const int32_t sampleRateReceiver = aTargetRate;
|
||||
AudioDriftCorrection ad(sampleRateTransmitter, sampleRateReceiver);
|
||||
|
||||
const float amplitude = 0.5;
|
||||
const float frequency = 10;
|
||||
const float phase = 0.0;
|
||||
float time = 0.0;
|
||||
const float deltaTime = 1.0f / sampleRateTransmitter;
|
||||
AudioToneGenerator<AudioDataValue> tone(sampleRateTransmitter);
|
||||
AudioToneVerifier<AudioDataValue> inToneVerify(sampleRateTransmitter);
|
||||
AudioToneVerifier<AudioDataValue> outToneVerify(sampleRateReceiver);
|
||||
|
||||
int32_t sourceFrames;
|
||||
const int32_t targetFrames = sampleRateReceiver / 100;
|
||||
|
|
@ -293,25 +441,18 @@ void testMonoToStereoInput(int aSourceRate, int aTargetRate) {
|
|||
|
||||
for (int n = 0; n < 250; ++n) {
|
||||
// Create the input (sine tone)
|
||||
AudioChunk chunk =
|
||||
CreateAudioChunk<float>(sourceFrames / 2, 1, AUDIO_FORMAT_FLOAT32);
|
||||
float* monoBuffer = chunk.ChannelDataForWrite<float>(0);
|
||||
for (int i = 0; i < chunk.GetDuration(); ++i) {
|
||||
double value = amplitude * sin(2 * M_PI * frequency * time + phase);
|
||||
monoBuffer[i] = static_cast<float>(value);
|
||||
time += deltaTime;
|
||||
}
|
||||
AudioChunk chunk2 =
|
||||
CreateAudioChunk<float>(sourceFrames / 2, 2, AUDIO_FORMAT_FLOAT32);
|
||||
for (int i = 0; i < chunk2.GetDuration(); ++i) {
|
||||
double value = amplitude * sin(2 * M_PI * frequency * time + phase);
|
||||
chunk2.ChannelDataForWrite<float>(0)[i] =
|
||||
chunk2.ChannelDataForWrite<float>(1)[i] = static_cast<float>(value);
|
||||
time += deltaTime;
|
||||
}
|
||||
AudioChunk chunk = CreateAudioChunk<AudioDataValue>(sourceFrames / 2, 1,
|
||||
AUDIO_OUTPUT_FORMAT);
|
||||
tone.Write(chunk);
|
||||
|
||||
AudioChunk chunk2 = CreateAudioChunk<AudioDataValue>(sourceFrames / 2, 2,
|
||||
AUDIO_OUTPUT_FORMAT);
|
||||
tone.Write(chunk2);
|
||||
|
||||
AudioSegment inSegment;
|
||||
inSegment.AppendAndConsumeChunk(&chunk);
|
||||
inSegment.AppendAndConsumeChunk(&chunk2);
|
||||
inToneVerify.AppendData(inSegment);
|
||||
// Print the input for debugging
|
||||
// printAudioSegment(inSegment);
|
||||
|
||||
|
|
@ -320,8 +461,21 @@ void testMonoToStereoInput(int aSourceRate, int aTargetRate) {
|
|||
EXPECT_EQ(outSegment.GetDuration(), targetFrames);
|
||||
// Print the output for debugging
|
||||
// printAudioSegment(outSegment);
|
||||
outToneVerify.AppendData(outSegment);
|
||||
}
|
||||
}
|
||||
EXPECT_EQ(inToneVerify.EstimatedFreq(), tone.mFrequency);
|
||||
EXPECT_EQ(inToneVerify.PreSilenceSamples(), 0);
|
||||
EXPECT_EQ(inToneVerify.CountDiscontinuities(), 0);
|
||||
|
||||
EXPECT_GT(outToneVerify.CountDiscontinuities(), 0)
|
||||
<< "Expect discontinuities";
|
||||
EXPECT_NE(outToneVerify.EstimatedFreq(), tone.mFrequency)
|
||||
<< "Estimation is not accurate due to discontinuities";
|
||||
// The expected pre-silence is 50ms plus the resampling. However, due to
|
||||
// discontinuities pre-silence is expected only in the first iteration which
|
||||
// is routhly a little more than 400 frames for the chosen sample rates.
|
||||
EXPECT_GT(outToneVerify.PreSilenceSamples(), 400);
|
||||
}
|
||||
|
||||
TEST(TestAudioDriftCorrection, MonoToStereoInput)
|
||||
|
|
|
|||
Загрузка…
Ссылка в новой задаче