void
MediaAudioTest::testCreation() {
const int32_t numSamples = 1024;
const int32_t sampleRate = 22050;
const int32_t channels = 8;
const AudioChannel::Layout layout = AudioChannel::CH_LAYOUT_7POINT1;
const AudioFormat::Type format = AudioFormat::SAMPLE_FMT_S16P;
// now let's test invalid methods.
RefPointer<MediaAudio> audio;
{
LoggerStack stack;
stack.setGlobalLevel(Logger::LEVEL_ERROR, false);
TS_ASSERT_THROWS(MediaAudio::make(-1, sampleRate, channels, layout, format),
HumbleInvalidArgument);
}
{
LoggerStack stack;
stack.setGlobalLevel(Logger::LEVEL_ERROR, false);
TS_ASSERT_THROWS(MediaAudio::make(numSamples, -1, channels, layout, format),
HumbleInvalidArgument);
}
{
LoggerStack stack;
stack.setGlobalLevel(Logger::LEVEL_ERROR, false);
TS_ASSERT_THROWS(MediaAudio::make(numSamples, sampleRate, -1, layout, format),
HumbleInvalidArgument);
}
{
LoggerStack stack;
stack.setGlobalLevel(Logger::LEVEL_ERROR, false);
TS_ASSERT_THROWS(MediaAudio::make(numSamples, sampleRate, channels + 1, layout,
format), HumbleInvalidArgument);
}
{
LoggerStack stack;
stack.setGlobalLevel(Logger::LEVEL_ERROR, false);
TS_ASSERT_THROWS(MediaAudio::make(numSamples, sampleRate, channels, layout,
AudioFormat::SAMPLE_FMT_NONE), HumbleInvalidArgument);
}
// And this should be valid
audio = MediaAudio::make(numSamples, sampleRate, channels + 1,
AudioChannel::CH_LAYOUT_UNKNOWN, format);
TS_ASSERT(audio);
audio = MediaAudio::make(numSamples, sampleRate, channels, layout, format);
TS_ASSERT(audio);
// now let's try getting the data
RefPointer<Buffer> buf;
TS_ASSERT_EQUALS(numSamples, audio->getMaxNumSamples());
TS_ASSERT_EQUALS(numSamples, audio->getNumSamples());
TS_ASSERT(!audio->isComplete());
TS_ASSERT_EQUALS(channels, audio->getChannels());
TS_ASSERT_EQUALS(channels, audio->getNumDataPlanes());
TS_ASSERT_EQUALS(layout, audio->getChannelLayout());
TS_ASSERT_EQUALS(sampleRate, audio->getSampleRate());
TS_ASSERT_EQUALS(format, audio->getFormat());
{
LoggerStack stack;
stack.setGlobalLevel(Logger::LEVEL_ERROR, false);
TS_ASSERT_THROWS(
audio->setNumSamples(audio->getMaxNumSamples()+1),
HumbleInvalidArgument);
TS_ASSERT_THROWS(audio->setNumSamples(-1), HumbleInvalidArgument);
TS_ASSERT_THROWS(audio->setNumSamples(0), HumbleInvalidArgument);
}
audio->setNumSamples(1);
TS_ASSERT_EQUALS(1, audio->getNumSamples());
audio->setComplete(true);
TS_ASSERT(audio->isComplete());
for (int i = 0; i < channels; i++) {
buf = audio->getData(i);
TS_ASSERT(buf);
TS_ASSERT_EQUALS(audio->getDataPlaneSize(i),
audio->getNumSamples()*AudioFormat::getBytesPerSample(audio->getFormat())*(audio->isPlanar()?1:audio->getChannels()));
}
// now let's try packed audio
audio = MediaAudio::make(numSamples, sampleRate, channels, layout,
AudioFormat::getPackedSampleFormat(format));
TS_ASSERT(audio);
TS_ASSERT_EQUALS(AudioFormat::getPackedSampleFormat(format),
audio->getFormat());
TS_ASSERT_EQUALS(channels, audio->getChannels());
TS_ASSERT_EQUALS(1, audio->getNumDataPlanes());
buf = audio->getData(0);
TS_ASSERT(buf);
for (int i = 1; i < channels; i++) {
LoggerStack stack;
stack.setGlobalLevel(Logger::LEVEL_ERROR, false);
TS_ASSERT_THROWS(audio->getData(i), HumbleInvalidArgument);
}
}
void
MediaAudioTest::testCopy() {
const int32_t numSamples = 155; // I choose an odd number because HV will align up to 32
const int32_t sampleRate = 22050;
const int32_t channels = 15; // choose a large # of channels to make sure we expand the Frame
const AudioChannel::Layout layout = AudioChannel::CH_LAYOUT_UNKNOWN;
const AudioFormat::Type format = AudioFormat::SAMPLE_FMT_DBLP;
int32_t bufSize = AudioFormat::getBufferSizeNeeded(numSamples, channels,
format);
// test that there is rounding up
int32_t minSize = AudioFormat::getBytesPerSample(format) * numSamples
* channels;
TS_ASSERT_LESS_THAN(minSize, bufSize);
RefPointer<Buffer> src = Buffer::make(0, bufSize);
double* srcData = (double*) src->getBytes(0, bufSize);
// now, let's go nuts!
for (size_t i = 0; i < bufSize / sizeof(double); i++) {
srcData[i] = i;
}
RefPointer<MediaAudio> audio;
audio = MediaAudio::make(src.value(), numSamples, sampleRate, channels,
layout, format);
TS_ASSERT(audio);
TS_ASSERT_EQUALS(channels, audio->getNumDataPlanes());
bool tests[] =
{ true, false };
for (size_t i = 0; i < sizeof(tests) / sizeof(*tests); i++) {
// now let's make a copy
RefPointer<MediaAudio> copy = MediaAudio::make(audio.value(), tests[i]);
TS_ASSERT_EQUALS(copy->getMaxNumSamples(), audio->getMaxNumSamples());
TS_ASSERT_EQUALS(copy->getNumSamples(), audio->getNumSamples());
TS_ASSERT_EQUALS(copy->getChannels(), audio->getChannels());
TS_ASSERT_EQUALS(copy->getNumDataPlanes(), audio->getNumDataPlanes());
TS_ASSERT_EQUALS(copy->getChannelLayout(), audio->getChannelLayout());
TS_ASSERT_EQUALS(copy->getSampleRate(), audio->getSampleRate());
TS_ASSERT_EQUALS(copy->getFormat(), audio->getFormat());
for (int32_t j = 0; j < audio->getNumDataPlanes(); j++) {
RefPointer<Buffer> srcBuf = audio->getData(j);
RefPointer<Buffer> dstBuf = copy->getData(j);
int32_t planeSize = srcBuf->getBufferSize();
TS_ASSERT_EQUALS(planeSize, dstBuf->getBufferSize());
uint8_t* srcBytes = (uint8_t*) srcBuf->getBytes(0, planeSize);
uint8_t* dstBytes = (uint8_t*) dstBuf->getBytes(0, planeSize);
if (tests[i]) {
for (int32_t k = 0; k < planeSize; k++) {
// should be byte-by-byte the same
TS_ASSERT_EQUALS(srcBytes[k], dstBytes[k]);
}
} else {
TS_ASSERT_EQUALS(srcBytes, dstBytes);
}
}
}
}
void
StreamCoderTest :: testTimestamps()
{
LoggerStack stack;
stack.setGlobalLevel(Logger::LEVEL_WARN, false);
Helper hr;
hr.setupReading("testfile_h264_mp4a_tmcd.mov");
RefPointer<IStreamCoder> ac;
RefPointer<IStreamCoder> vc;
if (hr.first_input_audio_stream >= 0)
ac = hr.coders[hr.first_input_audio_stream];
if (hr.first_input_video_stream >= 0)
vc = hr.coders[hr.first_input_video_stream];
hw->setupWriting("StreamCoderTest_7_output.flv",
vc ? "flv" : 0,
ac ? "libmp3lame" : 0,
"flv",
vc ? vc->getPixelType() : IPixelFormat::YUV420P,
vc ? vc->getWidth() : 0,
vc ? vc->getHeight() : 0,
false,
ac ? ac->getSampleRate() : 0,
ac ? ac->getChannels() : 0,
true);
RefPointer<IVideoPicture> frame;
if (vc) {
frame = IVideoPicture::make(
vc->getPixelType(),
vc->getWidth(),
vc->getHeight());
}
RefPointer<IAudioSamples> samples;
if (ac) {
samples = IAudioSamples::make(10000,
ac->getChannels());
}
int retval = -1;
int numReads = 0;
int maxReads = 10000;
while ((retval = hr.readNextDecodedObject(samples.value(), frame.value())) > 0
&& numReads < maxReads)
{
int result = retval;
numReads++;
VS_TUT_ENSURE("should only return 1 or 2", retval == 1 || retval == 2);
if (retval == 1)
{
VS_TUT_ENSURE("something should be complete...", samples->isComplete());
retval = hw->writeSamples(samples.value());
VS_TUT_ENSURE("could not write audio", retval >= 0);
}
if (retval == 2)
{
VS_TUT_ENSURE("something should be complete...", frame->isComplete());
retval = hw->writeFrame(frame.value());
VS_TUT_ENSURE("could not write video", retval >= 0);
}
{
RefPointer<IRational> timebase = hr.coders[hr.packet->getStreamIndex()]->getTimeBase();
ICodec::Type type = hr.coders[hr.packet->getStreamIndex()]->getCodecType();
(void) type;
(void) result;
VS_LOG_TRACE("Packet stream: %d.%d; pts: %lld; dts: %lld; spts: %lld; fpts: %lld; tb: %d/%d; ret: %d,"
"samps: %d",
hr.packet->getStreamIndex(),
type,
hr.packet->getPts(),
hr.packet->getDts(),
samples ? samples->getPts() : -1,
frame ? frame->getPts() : -1,
timebase->getNumerator(), timebase->getDenominator(),
result,
samples ? samples->getNumSamples() : 0);
}
}
VS_TUT_ENSURE("should return some data",
numReads > 0);
}
void
AudioResamplerTest :: testResamplingAudio()
{
RefPointer<IAudioResampler> resampler = IAudioResampler::make(2, 1,
44100, 22050);
RefPointer<IAudioSamples> samples = 0;
RefPointer<IAudioSamples> resamples = 0;
int numSamples = 0;
int numPackets = 0;
int retval = -1;
samples = IAudioSamples::make(1024, 1);
VS_TUT_ENSURE("got no samples", samples);
resamples = IAudioSamples::make(1024, 2);
VS_TUT_ENSURE("got no samples", samples);
h->setupReading(h->SAMPLE_FILE);
RefPointer<IPacket> packet = IPacket::make();
hw->setupWriting("AudioResamplerTest_3_output.flv", 0, "libmp3lame", "flv");
int outStream = hw->first_output_audio_stream;
VS_TUT_ENSURE("Could not find an audio stream in the output", outStream >= 0);
int inStream = h->first_input_audio_stream;
VS_TUT_ENSURE("Could not find an audio stream in the input", inStream >= 0);
RefPointer<IStreamCoder> ic = h->coders[inStream];
RefPointer<IStreamCoder> oc = hw->coders[outStream];
RefPointer<IPacket> opacket = IPacket::make();
VS_TUT_ENSURE("! opacket", opacket);
// Set the output coder correctly.
int outChannels = 2;
int outRate = 44100;
resampler = IAudioResampler::make(outChannels, ic->getChannels(),
outRate, ic->getSampleRate());
oc->setSampleRate(outRate);
oc->setChannels(outChannels);
oc->setBitRate(ic->getBitRate());
int maxSamples = 10 * ic->getSampleRate(); // 10 seconds
retval = ic->open();
VS_TUT_ENSURE("Could not open input coder", retval >= 0);
retval = oc->open();
VS_TUT_ENSURE("Could not open output coder", retval >= 0);
// write header
retval = hw->container->writeHeader();
VS_TUT_ENSURE("could not write header", retval >= 0);
while (h->container->readNextPacket(packet.value()) == 0
&& numSamples < maxSamples)
{
if (packet->getStreamIndex() == inStream)
{
int offset = 0;
numPackets++;
while (offset < packet->getSize())
{
retval = ic->decodeAudio(
samples.value(),
packet.value(),
offset);
VS_TUT_ENSURE("could not decode any audio",
retval > 0);
offset += retval;
VS_TUT_ENSURE("could not write any samples",
samples->getNumSamples() > 0);
numSamples += samples->getNumSamples();
resamples = IAudioSamples::make((samples->getNumSamples()*2),2);
// now, resample the audio
retval = resampler->resample(resamples.value(), samples.value(), 0);
VS_TUT_ENSURE("could not resample", retval > 0);
VS_TUT_ENSURE("no resamples", resamples->getNumSamples() > 0);
VS_TUT_ENSURE_EQUALS("wrong sample rate", resamples->getSampleRate(),
outRate);
VS_TUT_ENSURE_EQUALS("wrong channels", resamples->getChannels(),
outChannels);
// now, write out the packets.
unsigned int samplesConsumed = 0;
do {
retval = oc->encodeAudio(opacket.value(), resamples.value(),
samplesConsumed);
VS_TUT_ENSURE("Could not encode any audio", retval >= 0);
samplesConsumed += (unsigned int)retval;
VS_LOG_TRACE("packet: %d; is: %d; os: %d",
numPackets, numSamples, samplesConsumed);
if (opacket->isComplete())
{
VS_TUT_ENSURE("could not encode audio", opacket->getSize() > 0);
RefPointer<IBuffer> encodedBuffer = opacket->getData();
VS_TUT_ENSURE("no encoded data", encodedBuffer);
VS_TUT_ENSURE("less data than there should be",
encodedBuffer->getBufferSize() >=
opacket->getSize());
retval = hw->container->writePacket(opacket.value());
VS_TUT_ENSURE("could not write packet", retval >= 0);
}
// keep going until we've encoded all samples in this buffer.
} while (samplesConsumed < resamples->getNumSamples());
}
}
}
// sigh; it turns out that to flush the encoding buffers you need to
// ask the encoder to encode a NULL set of samples. So, let's do that.
retval = oc->encodeAudio(opacket.value(), 0, 0);
VS_TUT_ENSURE("Could not encode any audio", retval >= 0);
if (retval > 0)
{
retval = hw->container->writePacket(opacket.value());
VS_TUT_ENSURE("could not write packet", retval >= 0);
}
retval = hw->container->writeTrailer();
VS_TUT_ENSURE("! writeTrailer", retval >= 0);
retval = ic->close();
VS_TUT_ENSURE("! close", retval >= 0);
retval = oc->close();
VS_TUT_ENSURE("! close", retval >= 0);
VS_TUT_ENSURE("could not get any audio packets", numPackets > 0);
VS_TUT_ENSURE("could not decode any audio", numSamples > 0);
}
void
EncoderTest::testRegression36Internal (const Codec::ID codecId,
const int32_t numSamples,
const int32_t sampleRate,
const int32_t channels,
const AudioChannel::Layout channelLayout,
const AudioFormat::Type audioFormat,
const int64_t bitRate,
const char* testOutputName)
{
VS_LOG_DEBUG("Output filename: %s", testOutputName);
RefPointer<Codec> codec = Codec::findEncodingCodec (codecId);
RefPointer<Encoder> encoder = Encoder::make (codec.value ());
RefPointer<FilterGraph> graph = FilterGraph::make ();
RefPointer<MediaAudio> audio = MediaAudio::make (numSamples, sampleRate,
channels, channelLayout,
audioFormat);
// set the encoder properties we need
encoder->setSampleRate (audio->getSampleRate ());
encoder->setSampleFormat (audio->getFormat ());
encoder->setChannelLayout (audio->getChannelLayout ());
encoder->setChannels (audio->getChannels ());
encoder->setProperty ("b", (int64_t) (bitRate));
RefPointer<Rational> tb = Rational::make (1, sampleRate);
encoder->setTimeBase (tb.value ());
// create an output muxer
RefPointer<Muxer> muxer = Muxer::make (testOutputName, 0, 0);
RefPointer<MuxerFormat> format = muxer->getFormat ();
if (format->getFlag (MuxerFormat::GLOBAL_HEADER))
encoder->setFlag (Encoder::FLAG_GLOBAL_HEADER, true);
// open the encoder
encoder->open (0, 0);
RefPointer<FilterAudioSink> fsink = graph->addAudioSink (
"out", audio->getSampleRate (), audio->getChannelLayout (),
audio->getFormat ());
// Generate a 220 Hz sine wave with a 880 Hz beep each second, for 10 seconds.
graph->open ("sine=frequency=220:beep_factor=4:duration=11[out]");
fsink->setFrameSize (numSamples);
// add a stream for the encoded packets
{
RefPointer<MuxerStream> stream = muxer->addNewStream (encoder.value ());
} // and open the muxer
muxer->open (0, 0);
// now we're (in theory) ready to start writing data.
int32_t numCompletePackets = 0;
RefPointer<MediaPacket> packet;
// Get one audio packet that is larger than the frame-size.
fsink->getAudio (audio.value ());
TS_ASSERT(audio->isComplete ());
TS_ASSERT_EQUALS(audio->getNumSamples (), sampleRate);
audio->setTimeStamp (0);
// let's encode
packet = MediaPacket::make ();
encoder->encodeAudio (packet.value (), audio.value ());
if (packet->isComplete ())
{
muxer->write (packet.value (), false);
}
// now flush the encoder
do
{
packet = MediaPacket::make ();
encoder->encodeAudio (packet.value (), 0);
if (packet->isComplete ())
{
muxer->write (packet.value (), false);
++numCompletePackets;
}
}
while (packet->isComplete ());
muxer->close ();
if (!(codec->getCapabilities() & Codec::CAP_VARIABLE_FRAME_SIZE)) {
const int32_t numExpectedPackets = audio->getNumSamples()
/ encoder->getFrameSize();
VS_LOG_DEBUG("%ld vs %ld; framesize: %ld", numCompletePackets, numExpectedPackets, encoder->getFrameSize());
TS_ASSERT(numCompletePackets > 10);
}
}
void
EncoderTest::testEncodeAudio() {
Logger::setGlobalIsLogging(Logger::LEVEL_TRACE, false);
LoggerStack stack;
stack.setGlobalLevel(Logger::LEVEL_INFO, false);
const bool isMemCheck = getenv("VS_TEST_MEMCHECK") ? true : false;
const int32_t sampleRate = 44100;
const int32_t maxSamples = isMemCheck ? sampleRate*0.5 : sampleRate*10;
const int32_t numSamples = 1024;
const AudioChannel::Layout channelLayout = AudioChannel::CH_LAYOUT_STEREO;
const int32_t channels = AudioChannel::getNumChannelsInLayout(channelLayout);
const AudioFormat::Type audioFormat = AudioFormat::SAMPLE_FMT_S16;
RefPointer<Codec> codec = Codec::findEncodingCodec(Codec::CODEC_ID_AAC);
RefPointer<Encoder> encoder = Encoder::make(codec.value());
RefPointer<FilterGraph> graph = FilterGraph::make();
RefPointer<MediaAudio> audio = MediaAudio::make(numSamples, sampleRate, channels, channelLayout,
audioFormat);
// set the encoder properties we need
encoder->setSampleRate(audio->getSampleRate());
encoder->setSampleFormat(audio->getFormat());
encoder->setChannelLayout(audio->getChannelLayout());
encoder->setChannels(audio->getChannels());
encoder->setProperty("b", (int64_t)64000); // bitrate
RefPointer<Rational> tb = Rational::make(1,25);
encoder->setTimeBase(tb.value());
// create an output muxer
RefPointer<Muxer> muxer = Muxer::make("EncoderTest_encodeAudio.mp4", 0, 0);
RefPointer<MuxerFormat> format = muxer->getFormat();
if (format->getFlag(MuxerFormat::GLOBAL_HEADER))
encoder->setFlag(Encoder::FLAG_GLOBAL_HEADER, true);
// open the encoder
encoder->open(0, 0);
RefPointer<FilterAudioSink> fsink = graph->addAudioSink("out", audio->getSampleRate(), audio->getChannelLayout(), audio->getFormat());
// Generate a 220 Hz sine wave with a 880 Hz beep each second, for 10 seconds.
graph->open("sine=frequency=660:beep_factor=4:duration=11[out]");
// Generate an amplitude modulated signal
//graph->open("aevalsrc=sin(10*2*PI*t)*sin(880*2*PI*t)[out]");
// add a stream for the encoded packets
{
RefPointer<MuxerStream> stream = muxer->addNewStream(encoder.value());
}
// and open the muxer
muxer->open(0, 0);
// now we're (in theory) ready to start writing data.
int32_t numFrames = 0;
RefPointer<MediaPacket> packet;
while(fsink->getAudio(audio.value()) >= 0 && audio->isComplete() && numFrames*audio->getNumSamples() < maxSamples) {
audio->setTimeStamp(numFrames*audio->getNumSamples());
// let's encode
packet = MediaPacket::make();
encoder->encodeAudio(packet.value(), audio.value());
if (packet->isComplete()) {
muxer->write(packet.value(), false);
}
++numFrames;
}
// now flush the encoder
do {
packet = MediaPacket::make();
encoder->encodeAudio(packet.value(), 0);
if (packet->isComplete()) {
muxer->write(packet.value(), false);
}
} while (packet->isComplete());
muxer->close();
}