Implement wangqr Audio Changes:

- To allow for XAudio2 to work properly, we need to rework how does provider work since they only are used to be able to take in mono audio.
 - Other providers have been dumbed down to accept single channel audio since originally aegisub only accepted 1 channel audio.
 - meson.build has been modified to accommodate for xaudio, as we currently don't accept redistributable forms of xaudio, we need to work around the WinNT version.
 - There has been 1 more fix to res.rc to allow for compiling on non tagged releases.

CMakeLists.txt

@@ -497,6 +497,7 @@ set_property(SOURCE src/subtitles_provider_csri.cpp PROPERTY INCLUDE_DIRECTORIES
 if(MSVC)
     target_link_libraries (Aegisub dsound)
     add_definitions("-DWITH_DIRECTSOUND")
+    add_definitions("-DWITH_XAUDIO2")
     target_sources(Aegisub PRIVATE src/audio_player_dsound.cpp src/audio_player_dsound2.cpp src/audio_player_xaudio2.cpp)
 endif(MSVC)
 

libaegisub/audio/provider.cpp

@@ -21,13 +21,108 @@
 #include "libaegisub/log.h"
 #include "libaegisub/util.h"
 
-namespace agi {
+namespace {
-void AudioProvider::GetAudioWithVolume(void *buf, int64_t start, int64_t count, double volume) const {
-	GetAudio(buf, start, count);
 
+template<typename Source>
+class ConvertFloatToInt16 {
+	Source* src;
+public:
+	ConvertFloatToInt16(Source* src) :src(src) {}
+	int16_t operator[](size_t idx) const {
+		Source expanded = src[idx] * 32768;
+		return expanded < -32768 ? -32768 :
+			expanded > 32767 ? 32767 :
+			static_cast<int16_t>(expanded);
+	}
+};
+
+// 8 bits per sample is assumed to be unsigned with a bias of 128,
+// while everything else is assumed to be signed with zero bias
+class ConvertIntToInt16 {
+	void* src;
+	int bytes_per_sample;
+public:
+	ConvertIntToInt16(void* src, int bytes_per_sample) :src(src), bytes_per_sample(bytes_per_sample) {}
+	const int16_t& operator[](size_t idx) const {
+		return *reinterpret_cast<int16_t*>(reinterpret_cast<char*>(src) + (idx + 1) * bytes_per_sample - sizeof(int16_t));
+	}
+};
+class ConvertUInt8ToInt16 {
+	uint8_t* src;
+public:
+	ConvertUInt8ToInt16(uint8_t* src) :src(src) {}
+	int16_t operator[](size_t idx) const {
+		return int16_t(src[idx]-128) << 8;
+	}
+};
+
+template<typename Source>
+class DownmixToMono {
+	Source src;
+	int channels;
+public:
+	DownmixToMono(Source src, int channels) :src(src), channels(channels) {}
+	int16_t operator[](size_t idx) const {
+		int ret = 0;
+		// Just average the channels together
+		for (int i = 0; i < channels; ++i)
+			ret += src[idx * channels + i];
+		return ret / channels;
+	}
+};
+}
+
+namespace agi {
+	
+void AudioProvider::FillBufferInt16Mono(int16_t* buf, int64_t start, int64_t count) const {
+	if (!float_samples && bytes_per_sample == 2 && channels == 1) {
+		FillBuffer(buf, start, count);
+		return;
+	}
+	void* buff = malloc(bytes_per_sample * count * channels);
+	FillBuffer(buff, start, count);
+	if (channels == 1) {
+		if (float_samples) {
+			if (bytes_per_sample == sizeof(float))
+				for (int64_t i = 0; i < count; ++i)
+					buf[i] = ConvertFloatToInt16<float>(reinterpret_cast<float*>(buff))[i];
+			else if (bytes_per_sample == sizeof(double))
+				for (int64_t i = 0; i < count; ++i)
+					buf[i] = ConvertFloatToInt16<double>(reinterpret_cast<double*>(buff))[i];
+		}
+		else {
+			if (bytes_per_sample == sizeof(uint8_t))
+				for (int64_t i = 0; i < count; ++i)
+					buf[i] = ConvertUInt8ToInt16(reinterpret_cast<uint8_t*>(buff))[i];
+			else
+				for (int64_t i = 0; i < count; ++i)
+					buf[i] = ConvertIntToInt16(buff, bytes_per_sample)[i];
+		}
+	}
+	else {
+		if (float_samples) {
+			if (bytes_per_sample == sizeof(float))
+				for (int64_t i = 0; i < count; ++i)
+					buf[i] = DownmixToMono<ConvertFloatToInt16<float> >(ConvertFloatToInt16<float>(reinterpret_cast<float*>(buff)), channels)[i];
+			else if (bytes_per_sample == sizeof(double))
+				for (int64_t i = 0; i < count; ++i)
+					buf[i] = DownmixToMono<ConvertFloatToInt16<double> >(ConvertFloatToInt16<double>(reinterpret_cast<double*>(buff)), channels)[i];
+		}
+		else {
+			if (bytes_per_sample == sizeof(uint8_t))
+				for (int64_t i = 0; i < count; ++i)
+					buf[i] = DownmixToMono<ConvertUInt8ToInt16>(ConvertUInt8ToInt16(reinterpret_cast<uint8_t*>(buff)), channels)[i];
+			else
+				for (int64_t i = 0; i < count; ++i)
+					buf[i] = DownmixToMono<ConvertIntToInt16>(ConvertIntToInt16(buff, bytes_per_sample), channels)[i];
+		}
+	}
+	free(buff);
+}
+
+void AudioProvider::GetInt16MonoAudioWithVolume(int16_t *buf, int64_t start, int64_t count, double volume) const {
+	GetInt16MonoAudio(buf, start, count);
 	if (volume == 1.0) return;
-	if (bytes_per_sample != 2)
-		throw agi::InternalError("GetAudioWithVolume called on unconverted audio stream");
 
 	auto buffer = static_cast<int16_t *>(buf);
 	for (size_t i = 0; i < (size_t)count; ++i)
@@ -75,6 +170,39 @@ void AudioProvider::GetAudio(void *buf, int64_t start, int64_t count) const {
 	}
 }
 
+void AudioProvider::GetInt16MonoAudio(int16_t* buf, int64_t start, int64_t count) const {
+	if (start < 0) {
+		memset(buf, 0, sizeof(int16_t) * std::min(-start, count));
+		buf -= start;
+		count += start;
+		start = 0;
+	}
+
+	if (start + count > num_samples) {
+		int64_t zero_count = std::min(count, start + count - num_samples);
+		count -= zero_count;
+		memset(buf + count, 0, sizeof(int16_t) * zero_count);
+	}
+
+	if (count <= 0) return;
+
+	try {
+		FillBufferInt16Mono(buf, start, count);
+	}
+	catch (AudioDecodeError const& e) {
+		// We don't have any good way to report errors here, so just log the
+		// failure and return silence
+		LOG_E("audio_provider") << e.GetMessage();
+		memset(buf, 0, sizeof(int16_t) * count);
+		return;
+	}
+	catch (...) {
+		LOG_E("audio_provider") << "Unknown audio decoding error";
+		memset(buf, 0, sizeof(int16_t) * count);
+		return;
+	}
+}
+
 namespace {
 class writer {
 	io::Save outfile;

libaegisub/audio/provider_convert.cpp

@@ -22,119 +22,19 @@
 #include <limits>
 
 using namespace agi;
-
-/// Anything integral -> 16 bit signed machine-endian audio converter
 namespace {
-template<class Target>
+class ConvertAudioProvider final : public AudioProviderWrapper {
-class BitdepthConvertAudioProvider final : public AudioProviderWrapper {
-	int src_bytes_per_sample;
-	mutable std::vector<uint8_t> src_buf;
-
 public:
-	BitdepthConvertAudioProvider(std::unique_ptr<AudioProvider> src) : AudioProviderWrapper(std::move(src)) {
+	ConvertAudioProvider(std::unique_ptr<AudioProvider> src) : AudioProviderWrapper(std::move(src)) {
-		if (bytes_per_sample > 8)
-			throw AudioProviderError("Audio format converter: audio with bitdepths greater than 64 bits/sample is currently unsupported");
-
-		src_bytes_per_sample = bytes_per_sample;
-		bytes_per_sample = sizeof(Target);
-	}
-
-	void FillBuffer(void *buf, int64_t start, int64_t count64) const override {
-		auto count = static_cast<size_t>(count64);
-		assert(count == count64);
-
-		src_buf.resize(count * src_bytes_per_sample * channels);
-		source->GetAudio(src_buf.data(), start, count);
-
-		auto dest = static_cast<int16_t*>(buf);
-
-		for (int64_t i = 0; i < count * channels; ++i) {
-			int64_t sample = 0;
-
-			// 8 bits per sample is assumed to be unsigned with a bias of 127,
-			// while everything else is assumed to be signed with zero bias
-			if (src_bytes_per_sample == 1)
-				sample = src_buf[i] - 128;
-			else {
-				for (int j = src_bytes_per_sample; j > 0; --j) {
-					sample <<= 8;
-					sample += src_buf[i * src_bytes_per_sample + j - 1];
-				}
-			}
-
-			if (static_cast<size_t>(src_bytes_per_sample) > sizeof(Target))
-				sample /= 1LL << (src_bytes_per_sample - sizeof(Target)) * 8;
-			else if (static_cast<size_t>(src_bytes_per_sample) < sizeof(Target))
-				sample *=  1LL << (sizeof(Target) - src_bytes_per_sample ) * 8;
-
-			dest[i] = static_cast<Target>(sample);
-		}
-	}
-};
-
-/// Floating point -> 16 bit signed machine-endian audio converter
-template<class Source, class Target>
-class FloatConvertAudioProvider final : public AudioProviderWrapper {
-	mutable std::vector<Source> src_buf;
-
-public:
-	FloatConvertAudioProvider(std::unique_ptr<AudioProvider> src) : AudioProviderWrapper(std::move(src)) {
-		bytes_per_sample = sizeof(Target);
 		float_samples = false;
-	}
-
-	void FillBuffer(void *buf, int64_t start, int64_t count64) const override {
-		auto count = static_cast<size_t>(count64);
-		assert(count == count64);
-
-		src_buf.resize(count * channels);
-		source->GetAudio(&src_buf[0], start, count);
-
-		auto dest = static_cast<Target*>(buf);
-
-		for (size_t i = 0; i < static_cast<size_t>(count * channels); ++i) {
-			Source expanded;
-			if (src_buf[i] < 0)
-				expanded = static_cast<Target>(-src_buf[i] * std::numeric_limits<Target>::min());
-			else
-				expanded = static_cast<Target>(src_buf[i] * std::numeric_limits<Target>::max());
-
-			dest[i] = expanded < std::numeric_limits<Target>::min() ? std::numeric_limits<Target>::min() :
-			          expanded > std::numeric_limits<Target>::max() ? std::numeric_limits<Target>::max() :
-			                                                          static_cast<Target>(expanded);
-		}
-	}
-};
-
-/// Non-mono 16-bit signed machine-endian -> mono 16-bit signed machine endian converter
-class DownmixAudioProvider final : public AudioProviderWrapper {
-	int src_channels;
-	mutable std::vector<int16_t> src_buf;
-
-public:
-	DownmixAudioProvider(std::unique_ptr<AudioProvider> src) : AudioProviderWrapper(std::move(src)) {
-		src_channels = channels;
 		channels = 1;
+		bytes_per_sample = sizeof(int16_t);
 	}
 
-	void FillBuffer(void *buf, int64_t start, int64_t count64) const override {
+	void FillBuffer(void *buf, int64_t start, int64_t count) const override {
-		auto count = static_cast<size_t>(count64);
+		source->GetInt16MonoAudio(reinterpret_cast<int16_t*>(buf), start, count);
-		assert(count == count64);
-
-		src_buf.resize(count * src_channels);
-		source->GetAudio(&src_buf[0], start, count);
-
-		auto dst = static_cast<int16_t*>(buf);
-		// Just average the channels together
-		while (count-- > 0) {
-			int sum = 0;
-			for (int c = 0; c < src_channels; ++c)
-				sum += src_buf[count * src_channels + c];
-			dst[count] = static_cast<int16_t>(sum / src_channels);
-		}
 	}
 };
-
 /// Sample doubler with linear interpolation for the samples provider
 /// Requires 16-bit mono input
 class SampleDoublingAudioProvider final : public AudioProviderWrapper {
@@ -177,26 +77,23 @@ std::unique_ptr<AudioProvider> CreateConvertAudioProvider(std::unique_ptr<AudioP
 	// Ensure 16-bit audio with proper endianness
 	if (provider->AreSamplesFloat()) {
 		LOG_D("audio_provider") << "Converting float to S16";
-		if (provider->GetBytesPerSample() == sizeof(float))
-			provider = agi::make_unique<FloatConvertAudioProvider<float, int16_t>>(std::move(provider));
-		else
-			provider = agi::make_unique<FloatConvertAudioProvider<double, int16_t>>(std::move(provider));
 	}
 	if (provider->GetBytesPerSample() != 2) {
-		LOG_D("audio_provider") << "Converting " << provider->GetBytesPerSample() << " bytes per sample or wrong endian to S16";
+		LOG_D("audio_provider") << "Converting " << provider->GetBytesPerSample() << " bytes per sample to S16";
-		provider = agi::make_unique<BitdepthConvertAudioProvider<int16_t>>(std::move(provider));
 	}
 
 	// We currently only support mono audio
 	if (provider->GetChannels() != 1) {
 		LOG_D("audio_provider") << "Downmixing to mono from " << provider->GetChannels() << " channels";
-		provider = agi::make_unique<DownmixAudioProvider>(std::move(provider));
 	}
 
 	// Some players don't like low sample rate audio
-	while (provider->GetSampleRate() < 32000) {
+	if (provider->GetSampleRate() < 32000) {
-		LOG_D("audio_provider") << "Doubling sample rate";
+		provider = agi::make_unique<ConvertAudioProvider>(std::move(provider));
-		provider = agi::make_unique<SampleDoublingAudioProvider>(std::move(provider));
+		while (provider->GetSampleRate() < 32000) {
+			LOG_D("audio_provider") << "Doubling sample rate";
+			provider = agi::make_unique<SampleDoublingAudioProvider>(std::move(provider));
+		}
 	}
 
 	return provider;

libaegisub/audio/provider_hd.cpp

@@ -43,15 +43,15 @@ class HDAudioProvider final : public AudioProviderWrapper {
 		}
 
 		if (count > 0) {
-			start *= bytes_per_sample;
+			start *= bytes_per_sample * channels;
-			count *= bytes_per_sample;
+			count *= bytes_per_sample * channels;
 			memcpy(buf, file.read(start, count), count);
 		}
 	}
 
 	fs::path CacheFilename(fs::path const& dir) {
 		// Check free space
-		if ((uint64_t)num_samples * bytes_per_sample > fs::FreeSpace(dir))
+		if ((uint64_t)num_samples * bytes_per_sample * channels > fs::FreeSpace(dir))
 			throw AudioProviderError("Not enough free disk space in " + dir.string() + " to cache the audio");
 
 		return format("audio-%lld-%lld", time(nullptr),
@@ -61,7 +61,7 @@ class HDAudioProvider final : public AudioProviderWrapper {
 public:
 	HDAudioProvider(std::unique_ptr<AudioProvider> src, agi::fs::path const& dir)
 	: AudioProviderWrapper(std::move(src))
-	, file(dir / CacheFilename(dir), num_samples * bytes_per_sample)
+	, file(dir / CacheFilename(dir), num_samples * bytes_per_sample * channels)
 	{
 		decoded_samples = 0;
 		decoder = std::thread([&] {

libaegisub/audio/provider_ram.cpp

@@ -71,20 +71,22 @@ public:
 
 void RAMAudioProvider::FillBuffer(void *buf, int64_t start, int64_t count) const {
 	auto charbuf = static_cast<char *>(buf);
-	for (int64_t bytes_remaining = count * bytes_per_sample; bytes_remaining; ) {
+	for (int64_t bytes_remaining = count * bytes_per_sample * channels; bytes_remaining; ) {
 		if (start >= decoded_samples) {
 			memset(charbuf, 0, bytes_remaining);
 			break;
 		}
 
-		const int i = (start * bytes_per_sample) >> CacheBits;
+		const int64_t samples_per_block = CacheBlockSize / bytes_per_sample / channels;
-		const int start_offset = (start * bytes_per_sample) & (CacheBlockSize-1);
+
-		const int read_size = std::min<int>(bytes_remaining, CacheBlockSize - start_offset);
+		const size_t i = start / samples_per_block;
+		const int start_offset = (start % samples_per_block) * bytes_per_sample * channels;
+		const int read_size = std::min<int>(bytes_remaining, samples_per_block * bytes_per_sample * channels - start_offset);
 
 		memcpy(charbuf, &blockcache[i][start_offset], read_size);
 		charbuf += read_size;
 		bytes_remaining -= read_size;
-		start += read_size / bytes_per_sample;
+		start += read_size / bytes_per_sample / channels;
 	}
 }
 }

libaegisub/include/libaegisub/audio/provider.h

@@ -20,7 +20,6 @@
 #include <libaegisub/fs_fwd.h>
 
 #include <atomic>
-#include <memory>
 #include <vector>
 
 namespace agi {
@@ -37,6 +36,7 @@ protected:
 	bool float_samples = false;
 
 	virtual void FillBuffer(void *buf, int64_t start, int64_t count) const = 0;
+	virtual void FillBufferInt16Mono(int16_t* buf, int64_t start, int64_t count) const;
 
 	void ZeroFill(void *buf, int64_t count) const;
 
@@ -44,7 +44,8 @@ public:
 	virtual ~AudioProvider() = default;
 
 	void GetAudio(void *buf, int64_t start, int64_t count) const;
-	void GetAudioWithVolume(void *buf, int64_t start, int64_t count, double volume) const;
+	void GetInt16MonoAudio(int16_t* buf, int64_t start, int64_t count) const;
+	void GetInt16MonoAudioWithVolume(int16_t *buf, int64_t start, int64_t count, double volume) const;
 
 	int64_t GetNumSamples()     const { return num_samples; }
 	int64_t GetDecodedSamples() const { return decoded_samples; }
@@ -93,4 +94,4 @@ std::unique_ptr<AudioProvider> CreateHDAudioProvider(std::unique_ptr<AudioProvid
 std::unique_ptr<AudioProvider> CreateRAMAudioProvider(std::unique_ptr<AudioProvider> source_provider);
 
 void SaveAudioClip(AudioProvider const& provider, fs::path const& path, int start_time, int end_time);
-}
+}

meson.build

@@ -7,7 +7,7 @@ project('Aegisub', ['c', 'cpp'],
 cmake = import('cmake')
 
 if host_machine.system() == 'windows'
-    add_project_arguments('-DNOMINMAX', '-D_WIN32_WINNT=0x0601', language: 'cpp')
+    add_project_arguments('-DNOMINMAX', language: 'cpp')
 
     if not get_option('csri').disabled()
         add_global_arguments('-DCSRI_NO_EXPORT', language: 'c')
@@ -259,16 +259,26 @@ if host_machine.system() == 'windows'
     if not get_option('xaudio2').disabled()
         have_xaudio_h = cc.has_header('xaudio2.h')
         xaudio2_dep = cc.find_library('xaudio2', required: true)
-        if have_xaudio_h
+        if have_xaudio_h and xaudio2_dep.found()
             deps += [xaudio2_dep]
             conf.set('WITH_XAUDIO2', 1)
             dep_avail += 'XAudio2'
+            # XAudio2 needs Windows 8 or newer, so we tell meson not to define an older windows or else it can break things.
+            add_project_arguments('-D_WIN32_WINNT=0x0602', language: 'cpp')
+        else
+            # Windows 8 not required if XAudio2 fails to be found. revert for compat.
+            add_project_arguments('-D_WIN32_WINNT=0x0601', language: 'cpp')
         endif
         
         if not have_dsound_h and get_option('xaudio2').enabled()
             error('xaudio2 enabled but xaudio2.h not found')
         endif
+    else
+        # Windows 8 not required if XAudio2 is disabled. revert for compat.
+        add_project_arguments('-D_WIN32_WINNT=0x0601', language: 'cpp')
     endif
+    
+
 endif
 
 if host_machine.system() == 'darwin'

src/audio_player_alsa.cpp

@@ -127,7 +127,7 @@ void AlsaPlayer::PlaybackThread()
 
 do_setup:
 	snd_pcm_format_t pcm_format;
-	switch (provider->GetBytesPerSample())
+	switch (/*provider->GetBytesPerSample()*/ sizeof(int16_t))
 	{
 	case 1:
 		LOG_D("audio/player/alsa") << "format U8";
@@ -143,7 +143,7 @@ do_setup:
 	if (snd_pcm_set_params(pcm,
 	                       pcm_format,
 	                       SND_PCM_ACCESS_RW_INTERLEAVED,
-	                       provider->GetChannels(),
+	                       /*provider->GetChannels()*/ 1,
 	                       provider->GetSampleRate(),
 	                       1, // allow resample
 	                       100*1000 // 100 milliseconds latency
@@ -151,7 +151,8 @@ do_setup:
 		return;
 	LOG_D("audio/player/alsa") << "set pcm params";
 
-	size_t framesize = provider->GetChannels() * provider->GetBytesPerSample();
+	//size_t framesize = provider->GetChannels() * provider->GetBytesPerSample();
+	size_t framesize = sizeof(int16_t);
 
 	while (true)
 	{
@@ -175,7 +176,7 @@ do_setup:
 		{
 			auto avail = std::min(snd_pcm_avail(pcm), (snd_pcm_sframes_t)(end_position-position));
 			decode_buffer.resize(avail * framesize);
-			provider->GetAudioWithVolume(decode_buffer.data(), position, avail, volume);
+			provider->GetInt16MonoAudioWithVolume(reinterpret_cast<int16_t*>(decode_buffer.data()), position, avail, volume);
 
 			snd_pcm_sframes_t written = 0;
 			while (written <= 0)
@@ -235,7 +236,7 @@ do_setup:
 
 			{
 				decode_buffer.resize(avail * framesize);
-				provider->GetAudioWithVolume(decode_buffer.data(), position, avail, volume);
+				provider->GetInt16MonoAudioWithVolume(reinterpret_cast<int16_t*>(decode_buffer.data()), position, avail, volume);
 				snd_pcm_sframes_t written = 0;
 				while (written <= 0)
 				{
@@ -352,4 +353,4 @@ std::unique_ptr<AudioPlayer> CreateAlsaPlayer(agi::AudioProvider *provider, wxWi
 	return agi::make_unique<AlsaPlayer>(provider);
 }
 
-#endif // WITH_ALSA
+#endif // WITH_ALSA

src/audio_player_dsound.cpp

@@ -45,6 +45,7 @@
 
 #include <mmsystem.h>
 #include <dsound.h>
+#include <cguid.h>
 
 namespace {
 class DirectSoundPlayer;
@@ -111,8 +112,10 @@ DirectSoundPlayer::DirectSoundPlayer(agi::AudioProvider *provider, wxWindow *par
 	WAVEFORMATEX waveFormat;
 	waveFormat.wFormatTag = WAVE_FORMAT_PCM;
 	waveFormat.nSamplesPerSec = provider->GetSampleRate();
-	waveFormat.nChannels = provider->GetChannels();
+	//waveFormat.nChannels = provider->GetChannels();
-	waveFormat.wBitsPerSample = provider->GetBytesPerSample() * 8;
+	//waveFormat.wBitsPerSample = provider->GetBytesPerSample() * 8;
+	waveFormat.nChannels = 1;
+	waveFormat.wBitsPerSample = sizeof(int16_t) * 8;
 	waveFormat.nBlockAlign = waveFormat.nChannels * waveFormat.wBitsPerSample / 8;
 	waveFormat.nAvgBytesPerSec = waveFormat.nSamplesPerSec * waveFormat.nBlockAlign;
 	waveFormat.cbSize = sizeof(waveFormat);
@@ -160,7 +163,7 @@ bool DirectSoundPlayer::FillBuffer(bool fill) {
 	HRESULT res;
 	void *ptr1, *ptr2;
 	unsigned long int size1, size2;
-	int bytesps = provider->GetBytesPerSample();
+	int bytesps = /*provider->GetBytesPerSample()*/ sizeof(int16_t);
 
 	// To write length
 	int toWrite = 0;
@@ -223,8 +226,8 @@ RetryLock:
 	LOG_D_IF(!count1 && !count2, "audio/player/dsound1") << "DS fill: nothing";
 
 	// Get source wave
-	if (count1) provider->GetAudioWithVolume(ptr1, playPos, count1, volume);
+	if (count1) provider->GetInt16MonoAudioWithVolume(reinterpret_cast<int16_t*>(ptr1), playPos, count1, volume);
-	if (count2) provider->GetAudioWithVolume(ptr2, playPos+count1, count2, volume);
+	if (count2) provider->GetInt16MonoAudioWithVolume(reinterpret_cast<int16_t*>(ptr2), playPos+count1, count2, volume);
 	playPos += count1+count2;
 
 	buffer->Unlock(ptr1,count1*bytesps,ptr2,count2*bytesps);
@@ -254,7 +257,7 @@ void DirectSoundPlayer::Play(int64_t start,int64_t count) {
 	FillBuffer(true);
 
 	DWORD play_flag = 0;
-	if (count*provider->GetBytesPerSample() > bufSize) {
+	if (count*/*provider->GetBytesPerSample()*/sizeof(int16_t) > bufSize) {
 		// Start thread
 		thread = new DirectSoundPlayerThread(this);
 		thread->Create();
@@ -371,4 +374,4 @@ std::unique_ptr<AudioPlayer> CreateDirectSoundPlayer(agi::AudioProvider *provide
 	return agi::make_unique<DirectSoundPlayer>(provider, parent);
 }
 
-#endif // WITH_DIRECTSOUND
+#endif // WITH_DIRECTSOUND

src/audio_player_dsound2.cpp

@@ -48,6 +48,7 @@
 #include <mmsystem.h>
 #include <process.h>
 #include <dsound.h>
+#include <cguid.h>
 
 namespace {
 class DirectSoundPlayer2Thread;
@@ -319,8 +320,10 @@ void DirectSoundPlayer2Thread::Run()
 	WAVEFORMATEX waveFormat;
 	waveFormat.wFormatTag = WAVE_FORMAT_PCM;
 	waveFormat.nSamplesPerSec = provider->GetSampleRate();
-	waveFormat.nChannels = provider->GetChannels();
+	//waveFormat.nChannels = provider->GetChannels();
-	waveFormat.wBitsPerSample = provider->GetBytesPerSample() * 8;
+	//waveFormat.wBitsPerSample = provider->GetBytesPerSample() * 8;
+	waveFormat.nChannels = 1;
+	waveFormat.wBitsPerSample = sizeof(int16_t) * 8;
 	waveFormat.nBlockAlign = waveFormat.nChannels * waveFormat.wBitsPerSample / 8;
 	waveFormat.nAvgBytesPerSec = waveFormat.nSamplesPerSec * waveFormat.nBlockAlign;
 	waveFormat.cbSize = sizeof(waveFormat);
@@ -369,7 +372,7 @@ void DirectSoundPlayer2Thread::Run()
 	DWORD buffer_offset = 0;
 	bool playback_should_be_running = false;
 	int current_latency = wanted_latency;
-	const DWORD wanted_latency_bytes = wanted_latency*waveFormat.nSamplesPerSec*provider->GetBytesPerSample()/1000;
+	const DWORD wanted_latency_bytes = wanted_latency*waveFormat.nSamplesPerSec*/*provider->GetBytesPerSample()*/sizeof(int16_t)/1000;
 
 	while (running)
 	{
@@ -422,7 +425,7 @@ void DirectSoundPlayer2Thread::Run()
 				if (bytes_filled < wanted_latency_bytes)
 				{
 					// Very short playback length, do without streaming playback
-					current_latency = (bytes_filled*1000) / (waveFormat.nSamplesPerSec*provider->GetBytesPerSample());
+					current_latency = (bytes_filled*1000) / (waveFormat.nSamplesPerSec*/*provider->GetBytesPerSample()*/sizeof(int16_t));
 					if (FAILED(bfr->Play(0, 0, 0)))
 						REPORT_ERROR("Could not start single-buffer playback.")
 				}
@@ -553,7 +556,7 @@ do_fill_buffer:
 				else if (bytes_filled < wanted_latency_bytes)
 				{
 					// Didn't fill as much as we wanted to, let's get back to filling sooner than normal
-					current_latency = (bytes_filled*1000) / (waveFormat.nSamplesPerSec*provider->GetBytesPerSample());
+					current_latency = (bytes_filled*1000) / (waveFormat.nSamplesPerSec*/*provider->GetBytesPerSample()*/sizeof(int16_t));
 				}
 				else
 				{
@@ -577,7 +580,7 @@ DWORD DirectSoundPlayer2Thread::FillAndUnlockBuffers(void *buf1, DWORD buf1sz, v
 {
 	// Assume buffers have been locked and are ready to be filled
 
-	DWORD bytes_per_frame = provider->GetChannels() * provider->GetBytesPerSample();
+	DWORD bytes_per_frame = /*provider->GetChannels() * provider->GetBytesPerSample()*/sizeof(int16_t);
 	DWORD buf1szf = buf1sz / bytes_per_frame;
 	DWORD buf2szf = buf2sz / bytes_per_frame;
 
@@ -608,7 +611,7 @@ DWORD DirectSoundPlayer2Thread::FillAndUnlockBuffers(void *buf1, DWORD buf1sz, v
 			buf2sz = 0;
 		}
 
-		provider->GetAudioWithVolume(buf1, input_frame, buf1szf, volume);
+		provider->GetInt16MonoAudioWithVolume(reinterpret_cast<int16_t*>(buf1), input_frame, buf1szf, volume);
 
 		input_frame += buf1szf;
 	}
@@ -621,7 +624,7 @@ DWORD DirectSoundPlayer2Thread::FillAndUnlockBuffers(void *buf1, DWORD buf1sz, v
 			buf2sz = buf2szf * bytes_per_frame;
 		}
 
-		provider->GetAudioWithVolume(buf2, input_frame, buf2szf, volume);
+		provider->GetInt16MonoAudioWithVolume(reinterpret_cast<int16_t*>(buf2), input_frame, buf2szf, volume);
 
 		input_frame += buf2szf;
 	}
@@ -932,4 +935,4 @@ std::unique_ptr<AudioPlayer> CreateDirectSound2Player(agi::AudioProvider *provid
 	return agi::make_unique<DirectSoundPlayer2>(provider, parent);
 }
 
-#endif // WITH_DIRECTSOUND
+#endif // WITH_DIRECTSOUND

src/audio_player_openal.cpp

@@ -125,7 +125,7 @@ public:
 OpenALPlayer::OpenALPlayer(agi::AudioProvider *provider)
 : AudioPlayer(provider)
 , samplerate(provider->GetSampleRate())
-, bpf(provider->GetChannels() * provider->GetBytesPerSample())
+, bpf(/*provider->GetChannels() * provider->GetBytesPerSample()*/sizeof(int16_t))
 {
 	device = alcOpenDevice(nullptr);
 	if (!device) throw AudioPlayerOpenError("Failed opening default OpenAL device");
@@ -241,7 +241,7 @@ void OpenALPlayer::FillBuffers(ALsizei count)
 
 		if (fill_len > 0)
 			// Get fill_len frames of audio
-			provider->GetAudioWithVolume(&decode_buffer[0], cur_frame, fill_len, volume);
+			provider->GetInt16MonoAudioWithVolume(reinterpret_cast<int16_t*>(decode_buffer.data()), cur_frame, fill_len, volume);
 		if ((size_t)fill_len * bpf < decode_buffer.size())
 			// And zerofill the rest
 			memset(&decode_buffer[fill_len * bpf], 0, decode_buffer.size() - fill_len * bpf);
@@ -308,4 +308,4 @@ std::unique_ptr<AudioPlayer> CreateOpenALPlayer(agi::AudioProvider *provider, wx
 	return agi::make_unique<OpenALPlayer>(provider);
 }
 
-#endif // WITH_OPENAL
+#endif // WITH_OPENAL

src/audio_player_oss.cpp

@@ -131,7 +131,7 @@ public:
 
         while (!TestDestroy() && parent->cur_frame < parent->end_frame) {
             int rsize = std::min(wsize, parent->end_frame - parent->cur_frame);
-            parent->provider->GetAudioWithVolume(buf, parent->cur_frame,
+            parent->provider->GetInt16MonoAudioWithVolume(reinterpret_cast<int16_t*>(buf), parent->cur_frame,
                                                  rsize, parent->volume);
             int written = ::write(parent->dspdev, buf, rsize * parent->bpf);
             parent->cur_frame += written / parent->bpf;
@@ -146,7 +146,7 @@ public:
 
 void OSSPlayer::OpenStream()
 {
-    bpf = provider->GetChannels() * provider->GetBytesPerSample();
+    bpf = /*provider->GetChannels() * provider->GetBytesPerSample()*/sizeof(int16_t);
 
     // Open device
     wxString device = to_wx(OPT_GET("Player/Audio/OSS/Device")->GetString());
@@ -162,14 +162,14 @@ void OSSPlayer::OpenStream()
 #endif
 
     // Set number of channels
-    int channels = provider->GetChannels();
+    int channels = /*provider->GetChannels()*/1;
     if (ioctl(dspdev, SNDCTL_DSP_CHANNELS, &channels) < 0) {
         throw AudioPlayerOpenError("OSS player: setting channels failed");
     }
 
     // Set sample format
     int sample_format;
-    switch (provider->GetBytesPerSample()) {
+    switch (/*provider->GetBytesPerSample()*/sizeof(int16_t)) {
         case 1:
             sample_format = AFMT_S8;
             break;
@@ -283,4 +283,4 @@ std::unique_ptr<AudioPlayer> CreateOSSPlayer(agi::AudioProvider *provider, wxWin
     return agi::make_unique<OSSPlayer>(provider);
 }
 
-#endif // WITH_OSS
+#endif // WITH_OSS

src/audio_player_portaudio.cpp

@@ -140,7 +140,7 @@ void PortAudioPlayer::OpenStream() {
 		const PaDeviceInfo *device_info = Pa_GetDeviceInfo((*device_ids)[i]);
 		PaStreamParameters pa_output_p;
 		pa_output_p.device = (*device_ids)[i];
-		pa_output_p.channelCount = provider->GetChannels();
+		pa_output_p.channelCount = /*provider->GetChannels()*/ 1;
 		pa_output_p.sampleFormat = paInt16;
 		pa_output_p.suggestedLatency = device_info->defaultLowOutputLatency;
 		pa_output_p.hostApiSpecificStreamInfo = nullptr;
@@ -222,7 +222,7 @@ int PortAudioPlayer::paCallback(const void *inputBuffer, void *outputBuffer,
 
 	// Play something
 	if (lenAvailable > 0) {
-		player->provider->GetAudioWithVolume(outputBuffer, player->current, lenAvailable, player->GetVolume());
+		player->provider->GetInt16MonoAudioWithVolume(reinterpret_cast<int16_t*>(outputBuffer), player->current, lenAvailable, player->GetVolume());
 
 		// Set play position
 		player->current += lenAvailable;
@@ -283,4 +283,4 @@ std::unique_ptr<AudioPlayer> CreatePortAudioPlayer(agi::AudioProvider *provider,
 	return agi::make_unique<PortAudioPlayer>(provider);
 }
 
-#endif // WITH_PORTAUDIO
+#endif // WITH_PORTAUDIO

src/audio_player_pulse.cpp

@@ -133,11 +133,11 @@ PulseAudioPlayer::PulseAudioPlayer(agi::AudioProvider *provider) : AudioPlayer(p
 	}
 
 	// Set up stream
-	bpf = provider->GetChannels() * provider->GetBytesPerSample();
+	bpf = /*provider->GetChannels() * provider->GetBytesPerSample()*/sizeof(int16_t);
 	pa_sample_spec ss;
 	ss.format = PA_SAMPLE_S16LE; // FIXME
 	ss.rate = provider->GetSampleRate();
-	ss.channels = provider->GetChannels();
+	ss.channels = /*provider->GetChannels()*/1;
 	pa_channel_map map;
 	pa_channel_map_init_auto(&map, ss.channels, PA_CHANNEL_MAP_DEFAULT);
 
@@ -308,7 +308,7 @@ void PulseAudioPlayer::pa_stream_write(pa_stream *p, size_t length, PulseAudioPl
 	unsigned long maxframes = thread->end_frame - thread->cur_frame;
 	if (frames > maxframes) frames = maxframes;
 	void *buf = malloc(frames * bpf);
-	thread->provider->GetAudioWithVolume(buf, thread->cur_frame, frames, thread->volume);
+	thread->provider->GetInt16MonoAudioWithVolume(reinterpret_cast<int16_t*>(buf), thread->cur_frame, frames, thread->volume);
 	::pa_stream_write(p, buf, frames*bpf, free, 0, PA_SEEK_RELATIVE);
 	thread->cur_frame += frames;
 }
@@ -324,4 +324,4 @@ void PulseAudioPlayer::pa_stream_notify(pa_stream *p, PulseAudioPlayer *thread)
 std::unique_ptr<AudioPlayer> CreatePulseAudioPlayer(agi::AudioProvider *provider, wxWindow *) {
 	return agi::make_unique<PulseAudioPlayer>(provider);
 }
-#endif // WITH_LIBPULSE
+#endif // WITH_LIBPULSE

src/meson.build

@@ -236,7 +236,7 @@ opt_src = [
     ['OSS', 'audio_player_oss.cpp'],
     ['DirectSound', ['audio_player_dsound.cpp',
                      'audio_player_dsound2.cpp']],
-
+    ['XAudio2', 'audio_player_xaudio2.cpp'],
     ['FFMS2', ['audio_provider_ffmpegsource.cpp',
                'video_provider_ffmpegsource.cpp',
                'ffmpegsource_common.cpp']],

src/res/res.rc

@@ -39,8 +39,13 @@ eyedropper_cursor CURSOR "../bitmaps/windows/eyedropper.cur"
 #endif
 
 VS_VERSION_INFO VERSIONINFO
+#ifdef TAGGED_RELEASE
 FILEVERSION    	RESOURCE_BASE_VERSION, BUILD_GIT_VERSION_NUMBER
 PRODUCTVERSION 	RESOURCE_BASE_VERSION, 0
+#else
+FILEVERSION    	BUILD_GIT_VERSION_NUMBER, BUILD_GIT_VERSION_NUMBER
+PRODUCTVERSION 	BUILD_GIT_VERSION_NUMBER, 0
+#endif
 FILEFLAGSMASK  	VS_FFI_FILEFLAGSMASK
 FILEFLAGS      	(AGI_RC_FLAG_DEBUG|AGI_RC_FLAG_PRERELEASE)
 FILEOS         	VOS__WINDOWS32

tests/tests/audio.cpp

@@ -172,21 +172,21 @@ TEST(lagi_audio, save_audio_clip_out_of_audio_range) {
 
 TEST(lagi_audio, get_with_volume) {
 	TestAudioProvider<> provider;
-	uint16_t buff[4];
+	int16_t buff[4];
 
-	provider.GetAudioWithVolume(buff, 0, 4, 1.0);
+	provider.GetInt16MonoAudioWithVolume(buff, 0, 4, 1.0);
 	EXPECT_EQ(0, buff[0]);
 	EXPECT_EQ(1, buff[1]);
 	EXPECT_EQ(2, buff[2]);
 	EXPECT_EQ(3, buff[3]);
 
-	provider.GetAudioWithVolume(buff, 0, 4, 0.0);
+	provider.GetInt16MonoAudioWithVolume(buff, 0, 4, 0.0);
 	EXPECT_EQ(0, buff[0]);
 	EXPECT_EQ(0, buff[1]);
 	EXPECT_EQ(0, buff[2]);
 	EXPECT_EQ(0, buff[3]);
 
-	provider.GetAudioWithVolume(buff, 0, 4, 2.0);
+	provider.GetInt16MonoAudioWithVolume(buff, 0, 4, 2.0);
 	EXPECT_EQ(0, buff[0]);
 	EXPECT_EQ(2, buff[1]);
 	EXPECT_EQ(4, buff[2]);
@@ -195,8 +195,8 @@ TEST(lagi_audio, get_with_volume) {
 
 TEST(lagi_audio, volume_should_clamp_rather_than_wrap) {
 	TestAudioProvider<> provider;
-	uint16_t buff[1];
+	int16_t buff[1];
-	provider.GetAudioWithVolume(buff, 30000, 1, 2.0);
+	provider.GetInt16MonoAudioWithVolume(buff, 30000, 1, 2.0);
 	EXPECT_EQ(SHRT_MAX, buff[0]);
 }
 
@@ -232,7 +232,7 @@ TEST(lagi_audio, convert_8bit) {
 	auto provider = agi::CreateConvertAudioProvider(agi::make_unique<TestAudioProvider<uint8_t>>());
 
 	int16_t data[256];
-	provider->GetAudio(data, 0, 256);
+	provider->GetInt16MonoAudio(data, 0, 256);
 	for (int i = 0; i < 256; ++i)
 		ASSERT_EQ((i - 128) * 256, data[i]);
 }
@@ -243,13 +243,13 @@ TEST(lagi_audio, convert_32bit) {
 	auto provider = agi::CreateConvertAudioProvider(std::move(src));
 
 	int16_t sample;
-	provider->GetAudio(&sample, 0, 1);
+	provider->GetInt16MonoAudio(&sample, 0, 1);
 	EXPECT_EQ(SHRT_MIN, sample);
 
-	provider->GetAudio(&sample, 1LL << 31, 1);
+	provider->GetInt16MonoAudio(&sample, 1LL << 31, 1);
 	EXPECT_EQ(0, sample);
 
-	provider->GetAudio(&sample, (1LL << 32) - 1, 1);
+	provider->GetInt16MonoAudio(&sample, (1LL << 32) - 1, 1);
 	EXPECT_EQ(SHRT_MAX, sample);
 }
 
@@ -310,10 +310,10 @@ TEST(lagi_audio, stereo_downmix) {
 	};
 
 	auto provider = agi::CreateConvertAudioProvider(agi::make_unique<AudioProvider>());
-	EXPECT_EQ(1, provider->GetChannels());
+	EXPECT_EQ(2, provider->GetChannels());
 
 	int16_t samples[100];
-	provider->GetAudio(samples, 0, 100);
+	provider->GetInt16MonoAudio(samples, 0, 100);
 	for (int i = 0; i < 100; ++i)
 		EXPECT_EQ(i, samples[i]);
 }
@@ -333,27 +333,27 @@ struct FloatAudioProvider : agi::AudioProvider {
 		auto out = static_cast<Float *>(buf);
 		for (int64_t end = start + count; start < end; ++start) {
 			auto shifted = start + SHRT_MIN;
-			*out++ = (Float)(1.0 * shifted / (shifted < 0 ? -SHRT_MIN : SHRT_MAX));
+			*out++ = (Float)(shifted) / (-SHRT_MIN);
 		}
 	}
 };
 
 TEST(lagi_audio, float_conversion) {
 	auto provider = agi::CreateConvertAudioProvider(agi::make_unique<FloatAudioProvider<float>>());
-	EXPECT_FALSE(provider->AreSamplesFloat());
+	EXPECT_TRUE(provider->AreSamplesFloat());
 
 	int16_t samples[1 << 16];
-	provider->GetAudio(samples, 0, 1 << 16);
+	provider->GetInt16MonoAudio(samples, 0, 1 << 16);
 	for (int i = 0; i < (1 << 16); ++i)
 		ASSERT_EQ(i + SHRT_MIN, samples[i]);
 }
 
 TEST(lagi_audio, double_conversion) {
 	auto provider = agi::CreateConvertAudioProvider(agi::make_unique<FloatAudioProvider<double>>());
-	EXPECT_FALSE(provider->AreSamplesFloat());
+	EXPECT_TRUE(provider->AreSamplesFloat());
 
 	int16_t samples[1 << 16];
-	provider->GetAudio(samples, 0, 1 << 16);
+	provider->GetInt16MonoAudio(samples, 0, 1 << 16);
 	for (int i = 0; i < (1 << 16); ++i)
 		ASSERT_EQ(i + SHRT_MIN, samples[i]);
 }
@@ -551,4 +551,4 @@ TEST(lagi_audio, wave64_truncated) {
 	}
 
 	agi::fs::Remove(path);
-}
+}