jamosdev
/
rhubarb-lip-sync
mirror of https://github.com/DanielSWolf/rhubarb-lip-sync.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Refactoring: Replaced audio "length" with "duration"

This commit is contained in:
Daniel Wolf 2016-09-29 11:55:36 +02:00
parent b35e05fe7c
commit f5b7971f52
9 changed files with 23 additions and 23 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
src/Exporter.cpp
View File

@ -60,7 +60,7 @@ void XMLExporter::exportShapes(const boost::filesystem::path& inputFilePath, con
// Add metadata
tree.put("rhubarbResult.metadata.soundFile", inputFilePath.string());
tree.put("rhubarbResult.metadata.duration", formatDuration(shapes.getRange().getLength()));
tree.put("rhubarbResult.metadata.duration", formatDuration(shapes.getRange().getDuration()));
// Add mouth cues
for (auto& timedShape : dummyShapeIfEmpty(shapes)) {
@ -105,7 +105,7 @@ void JSONExporter::exportShapes(const boost::filesystem::path& inputFilePath, co
outputStream << "{\n";
outputStream << " \"metadata\": {\n";
outputStream << " \"soundFile\": \"" << escapeJSONString(inputFilePath.string()) << "\",\n";
outputStream << " \"duration\": " << formatDuration(shapes.getRange().getLength()) << "\n";
outputStream << " \"duration\": " << formatDuration(shapes.getRange().getDuration()) << "\n";
outputStream << " },\n";
outputStream << " \"mouthCues\": [\n";
bool isFirst = true;

2
src/TimeRange.cpp
View File

@ -27,7 +27,7 @@ time_type TimeRange::getEnd() const {
return end;
}
time_type TimeRange::getLength() const {
time_type TimeRange::getDuration() const {
return end - start;
}

2
src/TimeRange.h
View File

@ -16,7 +16,7 @@ public:
time_type getStart() const;
time_type getEnd() const;
time_type getLength() const;
time_type getDuration() const;
bool empty() const;
void resize(const TimeRange& newRange);

2
src/audio/AudioSegment.cpp
View File

@ -6,7 +6,7 @@ using std::make_unique;
AudioSegment::AudioSegment(std::unique_ptr<AudioClip> inputClip, const TimeRange& range) :
inputClip(std::move(inputClip)),
sampleOffset(static_cast<int64_t>(range.getStart().count()) * this->inputClip->getSampleRate() / 100),
sampleCount(static_cast<int64_t>(range.getLength().count()) * this->inputClip->getSampleRate() / 100)
sampleCount(static_cast<int64_t>(range.getDuration().count()) * this->inputClip->getSampleRate() / 100)
{
if (sampleOffset < 0 || sampleOffset + sampleCount > this->inputClip->size()) {
throw std::invalid_argument("Segment extends beyond input clip.");

2
src/audio/DCOffset.cpp
View File

@ -29,7 +29,7 @@ float getDCOffset(const AudioClip& audioClip) {
flatMeanSampleCount = 3 * sampleRate;
fadingMeanSampleCount = 1 * sampleRate;
} else {
// Short audio file. Average over the entire length.
// Short audio file. Average over the entire duration.
flatMeanSampleCount = static_cast<int>(audioClip.size());
fadingMeanSampleCount = 0;
}

6
src/audio/voiceActivityDetection.cpp
View File

@ -75,10 +75,10 @@ BoundedTimeline<void> detectVoiceActivity(const AudioClip& inputAudioClip, int m
// Split audio into segments and perform parallel VAD
const int segmentCount = maxThreadCount;
centiseconds audioLength = audioClip->getTruncatedRange().getLength();
centiseconds audioDuration = audioClip->getTruncatedRange().getDuration();
vector<TimeRange> audioSegments;
for (int i = 0; i < segmentCount; ++i) {
TimeRange segmentRange = TimeRange(i * audioLength / segmentCount, (i + 1) * audioLength / segmentCount);
TimeRange segmentRange = TimeRange(i * audioDuration / segmentCount, (i + 1) * audioDuration / segmentCount);
audioSegments.push_back(segmentRange);
}
runParallel([&](const TimeRange& segmentRange, ProgressSink& segmentProgressSink) {
@ -103,7 +103,7 @@ BoundedTimeline<void> detectVoiceActivity(const AudioClip& inputAudioClip, int m
// Shorten activities. WebRTC adds a bit of buffer at the end.
const centiseconds tail(5);
for (const auto& utterance : Timeline<void>(activity)) {
if (utterance.getTimeRange().getLength() > tail && utterance.getEnd() < audioLength) {
if (utterance.getTimeRange().getDuration() > tail && utterance.getEnd() < audioDuration) {
activity.clear(utterance.getEnd() - tail, utterance.getEnd());
}
}

18
src/mouthAnimation.cpp
View File

@ -146,17 +146,17 @@ Timeline<Shape> createTweens(ContinuousTimeline<Shape> shapes) {
centiseconds tweenStart, tweenDuration;
switch (tweenTiming) {
case TweenTiming::Early: {
tweenDuration = std::min(firstTimeRange.getLength() / 3, maxTweenDuration);
tweenDuration = std::min(firstTimeRange.getDuration() / 3, maxTweenDuration);
tweenStart = firstTimeRange.getEnd() - tweenDuration;
break;
}
case TweenTiming::Centered: {
tweenDuration = std::min({ firstTimeRange.getLength() / 3, secondTimeRange.getLength() / 3, maxTweenDuration });
tweenDuration = std::min({ firstTimeRange.getDuration() / 3, secondTimeRange.getDuration() / 3, maxTweenDuration });
tweenStart = firstTimeRange.getEnd() - tweenDuration / 2;
break;
}
case TweenTiming::Late: {
tweenDuration = std::min(secondTimeRange.getLength() / 3, maxTweenDuration);
tweenDuration = std::min(secondTimeRange.getDuration() / 3, maxTweenDuration);
tweenStart = secondTimeRange.getStart();
break;
}
@ -180,7 +180,7 @@ Timeline<Shape> animatePauses(const ContinuousTimeline<Shape>& shapes) {
const centiseconds maxPausedOpenMouthDuration = 35_cs;
const TimeRange timeRange = timedShape.getTimeRange();
if (timeRange.getLength() <= maxPausedOpenMouthDuration) {
if (timeRange.getDuration() <= maxPausedOpenMouthDuration) {
result.set(timeRange, B);
}
}
@ -189,11 +189,11 @@ Timeline<Shape> animatePauses(const ContinuousTimeline<Shape>& shapes) {
for_each_adjacent(shapes.begin(), shapes.end(), [&](const Timed<Shape>& secondLast, const Timed<Shape>& last, const Timed<Shape>& pause) {
if (pause.getValue() != X) return;
centiseconds lastLength = last.getTimeRange().getLength();
centiseconds lastDuration = last.getTimeRange().getDuration();
const centiseconds minOpenDuration = 20_cs;
if (isClosed(secondLast.getValue()) && !isClosed(last.getValue()) && lastLength < minOpenDuration) {
if (isClosed(secondLast.getValue()) && !isClosed(last.getValue()) && lastDuration < minOpenDuration) {
const centiseconds minSpillDuration = 20_cs;
centiseconds spillDuration = std::min(minSpillDuration, pause.getTimeRange().getLength());
centiseconds spillDuration = std::min(minSpillDuration, pause.getTimeRange().getDuration());
result.set(pause.getStart(), pause.getStart() + spillDuration, B);
}
});
@ -214,7 +214,7 @@ ContinuousTimeline<Shape> animate(const BoundedTimeline<Phone> &phones) {
for (const auto& timedPhone : continuousPhones) {
// Animate one phone
optional<Phone> phone = timedPhone.getValue();
centiseconds duration = timedPhone.getTimeRange().getLength();
centiseconds duration = timedPhone.getTimeRange().getDuration();
Timeline<Viseme> phoneVisemes = animate(phone, duration, previousPhoneDuration);
// Result timing is relative to phone. Make absolute.
@ -236,7 +236,7 @@ ContinuousTimeline<Shape> animate(const BoundedTimeline<Phone> &phones) {
Viseme viseme = it->getValue();
// Convert viseme to phone
Shape shape = viseme.getShape(it->getTimeRange().getLength(), lastShape);
Shape shape = viseme.getShape(it->getTimeRange().getDuration(), lastShape);
shapes.set(it->getTimeRange(), shape);
lastShape = shape;

8
src/phoneRecognition.cpp
View File

@ -288,10 +288,10 @@ Timeline<void> getNoiseSounds(TimeRange utteranceTimeRange, const Timeline<Phone
}
// Remove undesired elements
const centiseconds minSoundLength = 12_cs;
const centiseconds minSoundDuration = 12_cs;
for (const auto& unknownSound : Timeline<void>(noiseSounds)) {
bool startsAtZero = unknownSound.getStart() == 0_cs;
bool tooShort = unknownSound.getTimeRange().getLength() < minSoundLength;
bool tooShort = unknownSound.getTimeRange().getDuration() < minSoundDuration;
if (startsAtZero || tooShort) {
noiseSounds.clear(unknownSound.getTimeRange());
}
@ -428,7 +428,7 @@ BoundedTimeline<Phone> recognizePhones(
};
auto getUtteranceProgressWeight = [](const Timed<void> timedUtterance) {
return timedUtterance.getTimeRange().getLength().count();
return timedUtterance.getTimeRange().getDuration().count();
};
// Perform speech recognition
@ -439,7 +439,7 @@ BoundedTimeline<Phone> recognizePhones(
// Don't use more threads than there are utterances to be processed
static_cast<int>(utterances.size()),
// Don't waste time creating additional threads (and decoders!) if the recording is short
static_cast<int>(duration_cast<std::chrono::seconds>(audioClip->getTruncatedRange().getLength()).count() / 5)
static_cast<int>(duration_cast<std::chrono::seconds>(audioClip->getTruncatedRange().getDuration()).count() / 5)
});
if (threadCount < 1) {
threadCount = 1;

2
tests/TimelineTests.cpp
View File

@ -230,7 +230,7 @@ void testSetter(std::function<void(const Timed<int>&, Timeline<int>&)> set) {
Timed<int> lastElement(centiseconds::min(), centiseconds::min(), std::numeric_limits<int>::min());
for (const auto& element : timeline) {
// No element shound have zero-length
EXPECT_LT(0_cs, element.getTimeRange().getLength());
EXPECT_LT(0_cs, element.getTimeRange().getDuration());
// No two adjacent elements should have the same value; they should have been merged
if (element.getStart() == lastElement.getEnd()) {