mirror/JUCE
1
0
Fork 0
mirror of https://github.com/juce-framework/JUCE.git synced 2026-01-31 03:00:05 +00:00
Code Activity

Updated the juce demo audio latency detector to use a more accurate, impulse based detection method

Browse source
This commit is contained in:
Julian Storer 2009-10-01 19:58:41 +01:00
parent b171ae8400
commit 9623f6b108
1 changed files with 62 additions and 35 deletions
Download patch file Download diff file Expand all files Collapse all files

97
extras/juce demo/src/demos/AudioDemoLatencyPage.cpp
View file

@ -37,7 +37,6 @@ public:
recordedSound (1, 1),
playingSampleNum (0),
recordedSampleNum (-1),
sineWaveFrequency (300.0),
isRunning (false),
resultsBox (resultsBox_)
{
@ -53,6 +52,7 @@ public:
startTimer (50);
const ScopedLock sl (lock);
createTestSound();
recordedSound.clear();
playingSampleNum = recordedSampleNum = 0;
isRunning = true;
@ -73,13 +73,13 @@ public:
if (latencySamples >= 0)
{
message << "\n\nLatency test results:\n"
<< latencySamples << " samples (" << (latencySamples * 1000.0 / sampleRate) << " milliseconds)\n"
<< latencySamples << " samples (" << String (latencySamples * 1000.0 / sampleRate, 1) << " milliseconds)\n"
<< "The audio device reports an input latency of "
<< deviceInputLatency << " samples, output latency of "
<< deviceOutputLatency << " samples."
<< "\nSo the corrected latency = "
<< (latencySamples - deviceInputLatency - deviceOutputLatency)
<< " samples (" << ((latencySamples - deviceInputLatency - deviceOutputLatency) * 1000.0 / sampleRate)
<< " samples (" << String ((latencySamples - deviceInputLatency - deviceOutputLatency) * 1000.0 / sampleRate, 2)
<< " milliseconds)";
}
else
@ -102,8 +102,6 @@ public:
deviceOutputLatency = device->getOutputLatencyInSamples();
playingSampleNum = recordedSampleNum = 0;
createTestSound();
recordedSound.setSize (1, (int) (1.5 * sampleRate));
recordedSound.clear();
}
@ -161,12 +159,13 @@ private:
AudioSampleBuffer testSound, recordedSound;
int playingSampleNum, recordedSampleNum;
CriticalSection lock;
const double sineWaveFrequency;
double sampleRate;
bool isRunning;
TextEditor* resultsBox;
int deviceInputLatency, deviceOutputLatency;
Array <int> spikes;
void createTestSound()
{
const int length = ((int) sampleRate) / 4;
@ -174,61 +173,89 @@ private:
testSound.clear();
float* s = testSound.getSampleData (0, 0);
const double scale = 2.0 * double_Pi / (sampleRate / sineWaveFrequency);
int n = 0;
IntegerElementComparator<int> comp;
spikes.clear();
for (int i = 512; i < length; ++i)
s[i] = 0.95f * sinf ((float) (scale * n++));
for (int i = 0; i < 50; ++i)
{
const int spikePos = Random::getSystemRandom().nextInt (length - 20) + 10;
spikes.addSorted (comp, spikePos);
testSound.applyGainRamp (0, length - length / 3, length / 3, 1.0f, 0.0f);
s [spikePos] = 0.99f;
s [spikePos + 1] = -0.99f;
}
}
int findStartSampleOfSineWave (const AudioSampleBuffer& buffer, const double sampleRate) const
// Searches a buffer for a set of spikes that matches those in the test sound
int findOffsetOfSpikes (const AudioSampleBuffer& buffer) const
{
const float minSpikeLevel = 5.0f;
const double smooth = 0.975;
const float* s = buffer.getSampleData (0, 0);
const int spikeDriftAllowed = 5;
const double damping = 0.995;
double avG = 0;
Array <int> spikesFound (100);
double runningAverage = 0;
int lastSpike = 0;
for (int i = 0; i < buffer.getNumSamples(); i += 2)
for (int i = 0; i < buffer.getNumSamples() - 10; ++i)
{
const int num = jmin (512, buffer.getNumSamples() - i);
const float samp = fabsf (s[i]);
const double g = calcGoertzel (s + i, num, sampleRate, sineWaveFrequency);
avG = avG * damping + (1.0 - damping) * g * g;
if (samp > runningAverage * minSpikeLevel && i > lastSpike + 20)
{
lastSpike = i;
spikesFound.add (i);
}
if (avG > 3)
return i;
runningAverage = runningAverage * smooth + (1.0 - smooth) * samp;
}
return -1;
}
int bestMatch = -1;
int bestNumMatches = spikes.size() / 3; // the minimum number of matches required
static double calcGoertzel (const float* const samples, const int numSamples,
const double sampleRate, const double frequency) throw()
{
double n = 0, n1 = 0;
const double pi2freqOverRate = (2.0 * double_Pi) * frequency / sampleRate;
const double multiplier = 2.0 * cos (pi2freqOverRate);
if (spikesFound.size() < bestNumMatches)
return -1;
for (int i = 0; i < numSamples; ++i)
for (int offsetToTest = 0; offsetToTest < buffer.getNumSamples() - 2048; ++offsetToTest)
{
const double n2 = n1;
n1 = n;
n = multiplier * n1 - n2 + samples[i];
int numMatchesHere = 0;
int foundIndex = 0;
for (int refIndex = 0; refIndex < spikes.size(); ++refIndex)
{
const int referenceSpike = spikes.getUnchecked (refIndex) + offsetToTest;
int spike = 0;
while ((spike = spikesFound.getUnchecked (foundIndex)) < referenceSpike - spikeDriftAllowed
&& foundIndex < spikesFound.size() - 1)
++foundIndex;
if (spike >= referenceSpike - spikeDriftAllowed && spike <= referenceSpike + spikeDriftAllowed)
++numMatchesHere;
}
if (numMatchesHere > bestNumMatches)
{
bestNumMatches = numMatchesHere;
bestMatch = offsetToTest;
if (numMatchesHere == spikes.size())
break;
}
}
return n - n1 * exp (-pi2freqOverRate);
return bestMatch;
}
int calculateLatencySamples() const
{
// Detect the sound in both our test sound and the recording of it, and measure the difference
// in their start times..
const int referenceStart = findStartSampleOfSineWave (testSound, sampleRate);
const int referenceStart = findOffsetOfSpikes (testSound);
jassert (referenceStart >= 0);
const int recordedStart = findStartSampleOfSineWave (recordedSound, sampleRate);
const int recordedStart = findOffsetOfSpikes (recordedSound);
return (recordedStart < 0) ? -1 : (recordedStart - referenceStart);
}