AudioProvider *AudioProviderFactory::GetProvider(wxString const& filename) {
provider_creator creator;
AudioProvider *provider = nullptr;
provider = creator.try_create("Dummy audio provider", [&]() { return new DummyAudioProvider(filename); });
// Try a PCM provider first
if (!provider && !OPT_GET("Provider/Audio/PCM/Disable")->GetBool())
provider = creator.try_create("PCM audio provider", [&]() { return CreatePCMAudioProvider(filename); });
if (!provider) {
std::vector<std::string> list = GetClasses(OPT_GET("Audio/Provider")->GetString());
if (list.empty()) throw agi::NoAudioProvidersError("No audio providers are available.", 0);
for (auto const& name : list) {
provider = creator.try_create(name, [&]() { return Create(name, filename); });
if (provider) break;
}
}
if (!provider) {
if (creator.found_audio)
throw agi::AudioProviderOpenError(creator.msg, 0);
if (creator.found_file)
throw agi::AudioDataNotFoundError(creator.msg, 0);
throw agi::FileNotFoundError(from_wx(filename));
}
bool needsCache = provider->NeedsCache();
// Give it a converter if needed
if (provider->GetBytesPerSample() != 2 || provider->GetSampleRate() < 32000 || provider->GetChannels() != 1)
provider = CreateConvertAudioProvider(provider);
// Change provider to RAM/HD cache if needed
int cache = OPT_GET("Audio/Cache/Type")->GetInt();
if (!cache || !needsCache)
return new LockAudioProvider(provider);
DialogProgress progress(wxGetApp().frame, _("Load audio"));
// Convert to RAM
if (cache == 1) return new RAMAudioProvider(provider, &progress);
// Convert to HD
if (cache == 2) return new HDAudioProvider(provider, &progress);
throw agi::AudioCacheOpenError("Unknown caching method", 0);
}
/// @brief Open stream
///
void DirectSoundPlayer::OpenStream() {
// Get provider
AudioProvider *provider = GetProvider();
// Initialize the DirectSound object
HRESULT res;
res = DirectSoundCreate8(&DSDEVID_DefaultPlayback,&directSound,NULL); // TODO: support selecting audio device
if (FAILED(res)) throw _T("Failed initializing DirectSound");
// Set DirectSound parameters
AegisubApp *app = (AegisubApp*) wxTheApp;
directSound->SetCooperativeLevel((HWND)app->frame->GetHandle(),DSSCL_PRIORITY);
// Create the wave format structure
WAVEFORMATEX waveFormat;
waveFormat.wFormatTag = WAVE_FORMAT_PCM;
waveFormat.nSamplesPerSec = provider->GetSampleRate();
waveFormat.nChannels = provider->GetChannels();
waveFormat.wBitsPerSample = provider->GetBytesPerSample() * 8;
waveFormat.nBlockAlign = waveFormat.nChannels * waveFormat.wBitsPerSample / 8;
waveFormat.nAvgBytesPerSec = waveFormat.nSamplesPerSec * waveFormat.nBlockAlign;
waveFormat.cbSize = sizeof(waveFormat);
// Create the buffer initializer
int aim = waveFormat.nAvgBytesPerSec * 15/100; // 150 ms buffer
int min = DSBSIZE_MIN;
int max = DSBSIZE_MAX;
bufSize = std::min(std::max(min,aim),max);
DSBUFFERDESC desc;
desc.dwSize = sizeof(DSBUFFERDESC);
desc.dwFlags = DSBCAPS_GETCURRENTPOSITION2 | DSBCAPS_GLOBALFOCUS;
desc.dwBufferBytes = bufSize;
desc.dwReserved = 0;
desc.lpwfxFormat = &waveFormat;
desc.guid3DAlgorithm = GUID_NULL;
// Create the buffer
IDirectSoundBuffer *buf;
res = directSound->CreateSoundBuffer(&desc,&buf,NULL);
if (res != DS_OK) throw _T("Failed creating DirectSound buffer");
// Copy interface to buffer
res = buf->QueryInterface(IID_IDirectSoundBuffer8,(LPVOID*) &buffer);
if (res != S_OK) throw _T("Failed casting interface to IDirectSoundBuffer8");
// Set data
offset = 0;
}
//////////////////////
// PortAudio callback
int PortAudioPlayer::paCallback(void *inputBuffer, void *outputBuffer, unsigned long framesPerBuffer, PaTimestamp outTime, void *userData) {
// Get provider
PortAudioPlayer *player = (PortAudioPlayer *) userData;
AudioProvider *provider = player->GetProvider();
int end = 0;
// Calculate how much left
__int64 lenAvailable = player->endPos - player->playPos;
uint64_t avail = 0;
if (lenAvailable > 0) {
avail = lenAvailable;
if (avail > framesPerBuffer) {
lenAvailable = framesPerBuffer;
avail = lenAvailable;
}
}
else {
lenAvailable = 0;
avail = 0;
}
// Play something
if (lenAvailable > 0) {
provider->GetAudio(outputBuffer,player->playPos,lenAvailable);
}
// Pad end with blank
if (avail < (uint64_t) framesPerBuffer) {
//provider->softStop = true;
}
// Set volume
short *output = (short*) outputBuffer;
for (unsigned int i=0;i<avail;i++) output[i] = MID(-(1<<15),int(output[i] * player->GetVolume()),(1<<15)-1);
// Fill rest with blank
for (unsigned int i=avail;i<framesPerBuffer;i++) output[i]=0;
// Set play position (and real one)
player->playPos += framesPerBuffer;
player->realPlayPos = (__int64)(Pa_StreamTime(player->stream) - player->paStart) + player->startPos;
// Cap to start if lower
return end;
}
/// @brief See if we need to downmix the number of channels
/// @param source_provider
///
AudioProvider *CreateConvertAudioProvider(AudioProvider *source_provider) {
AudioProvider *provider = source_provider;
// Aegisub requires 16 bit samples,
// some audio players break with low samplerates,
// everything breaks with wrong-ended samples.
if (provider->GetBytesPerSample() != 2 ||
provider->GetSampleRate() < 32000 ||
!provider->AreSamplesNativeEndian())
{
// @todo add support for more bitdepths (i.e. 24- and 32-bit audio)
if (provider->GetBytesPerSample() > 2)
throw AudioOpenError("Audio format converter: audio with bitdepths greater than 16 bits/sample is currently unsupported");
provider = new ConvertAudioProvider(provider);
}
// We also require mono audio for historical reasons
if (provider->GetChannels() != 1)
{
provider = new DownmixingAudioProvider(provider);
}
return provider;
}
AudioProvider *AudioProviderFactory::GetProvider(wxString const& filename, int cache) {
AudioProvider *provider = 0;
bool found_file = false;
bool found_audio = false;
std::string msg;
if (!OPT_GET("Provider/Audio/PCM/Disable")->GetBool()) {
// Try a PCM provider first
try {
provider = CreatePCMAudioProvider(filename);
LOG_D("audio_provider") << "Using PCM provider";
}
catch (agi::FileNotFoundError const& err) {
msg = "PCM audio provider: " + err.GetMessage() + " not found.\n";
}
catch (agi::AudioOpenError const& err) {
found_file = true;
msg += err.GetChainedMessage() + "\n";
}
}
if (!provider) {
std::vector<std::string> list = GetClasses(OPT_GET("Audio/Provider")->GetString());
if (list.empty()) throw agi::NoAudioProvidersError("No audio providers are available.", 0);
for (size_t i = 0; i < list.size() ; ++i) {
try {
provider = Create(list[i], filename);
if (provider) {
LOG_D("audio_provider") << "Using audio provider: " << list[i];
break;
}
}
catch (agi::FileNotFoundError const& err) {
msg += list[i] + ": " + err.GetMessage() + " not found.\n";
}
catch (agi::AudioDataNotFoundError const& err) {
found_file = true;
msg += list[i] + ": " + err.GetChainedMessage() + "\n";
}
catch (agi::AudioOpenError const& err) {
found_audio = true;
found_file = true;
msg += list[i] + ": " + err.GetChainedMessage() + "\n";
}
}
}
if (!provider) {
if (found_audio)
throw agi::AudioProviderOpenError(msg, 0);
if (found_file)
throw agi::AudioDataNotFoundError(msg, 0);
throw agi::FileNotFoundError(STD_STR(filename));
}
bool needsCache = provider->NeedsCache();
// Give it a converter if needed
if (provider->GetBytesPerSample() != 2 || provider->GetSampleRate() < 32000 || provider->GetChannels() != 1)
provider = CreateConvertAudioProvider(provider);
// Change provider to RAM/HD cache if needed
if (cache == -1) cache = OPT_GET("Audio/Cache/Type")->GetInt();
if (!cache || !needsCache) {
return new LockAudioProvider(provider);
}
DialogProgress progress(wxGetApp().frame, _("Load audio"));
// Convert to RAM
if (cache == 1) return new RAMAudioProvider(provider, &progress);
// Convert to HD
if (cache == 2) return new HDAudioProvider(provider, &progress);
throw agi::AudioCacheOpenError("Unknown caching method", 0);
}
/// @brief Fill buffer
/// @param fill
/// @return
///
bool DirectSoundPlayer::FillBuffer(bool fill) {
if (playPos >= endPos) return false;
// Variables
HRESULT res;
void *ptr1, *ptr2;
unsigned long int size1, size2;
AudioProvider *provider = GetProvider();
int bytesps = provider->GetBytesPerSample();
// To write length
int toWrite = 0;
if (fill) {
toWrite = bufSize;
}
else {
DWORD bufplay;
res = buffer->GetCurrentPosition(&bufplay, NULL);
if (FAILED(res)) return false;
toWrite = (int)bufplay - (int)offset;
if (toWrite < 0) toWrite += bufSize;
}
if (toWrite == 0) return true;
// Make sure we only get as many samples as are available
if (playPos + toWrite/bytesps > endPos) {
toWrite = (endPos - playPos) * bytesps;
}
// If we're going to fill the entire buffer (ie. at start of playback) start by zeroing it out
// If it's not zeroed out we might have a playback selection shorter than the buffer
// and then everything after the playback selection will be junk, which we don't want played.
if (fill) {
RetryClear:
res = buffer->Lock(0, bufSize, &ptr1, &size1, &ptr2, &size2, 0);
if (res == DSERR_BUFFERLOST) {
buffer->Restore();
goto RetryClear;
}
memset(ptr1, 0, size1);
memset(ptr2, 0, size2);
buffer->Unlock(ptr1, size1, ptr2, size2);
}
// Lock buffer
RetryLock:
if (fill) {
res = buffer->Lock(offset, toWrite, &ptr1, &size1, &ptr2, &size2, 0);
}
else {
res = buffer->Lock(offset, toWrite, &ptr1, &size1, &ptr2, &size2, 0);//DSBLOCK_FROMWRITECURSOR);
}
// Buffer lost?
if (res == DSERR_BUFFERLOST) {
LOG_D("audio/player/dsound1") << "lost dsound buffer";
buffer->Restore();
goto RetryLock;
}
// Error
if (FAILED(res)) return false;
// Convert size to number of samples
unsigned long int count1 = size1 / bytesps;
unsigned long int count2 = size2 / bytesps;
LOG_D_IF(count1, "audio/player/dsound1") << "DS fill: " << (unsigned long)playPos << " -> " << (unsigned long)playPos+count1;
LOG_D_IF(count2, "audio/player/dsound1") << "DS fill: " << (unsigned long)playPos+count1 << " -> " << (unsigned long)playPos+count1+count2;
LOG_D_IF(!count1 && !count2, "audio/player/dsound1") << "DS fill: nothing";
// Get source wave
if (count1) provider->GetAudioWithVolume(ptr1, playPos, count1, volume);
if (count2) provider->GetAudioWithVolume(ptr2, playPos+count1, count2, volume);
playPos += count1+count2;
// Unlock
buffer->Unlock(ptr1,count1*bytesps,ptr2,count2*bytesps);
// Update offset
offset = (offset + count1*bytesps + count2*bytesps) % bufSize;
return playPos < endPos;
}
void SaveAudioClip(AudioProvider const& provider, fs::path const& path, int start_time, int end_time) {
const auto max_samples = provider.GetNumSamples();
const auto start_sample = std::min(max_samples, ((int64_t)start_time * provider.GetSampleRate() + 999) / 1000);
const auto end_sample = util::mid(start_sample, ((int64_t)end_time * provider.GetSampleRate() + 999) / 1000, max_samples);
const size_t bytes_per_sample = provider.GetBytesPerSample() * provider.GetChannels();
const size_t bufsize = (end_sample - start_sample) * bytes_per_sample;
writer out{path};
out.write("RIFF");
out.write<int32_t>(bufsize + 36);
out.write("WAVEfmt ");
out.write<int32_t>(16); // Size of chunk
out.write<int16_t>(1); // compression format (PCM)
out.write<int16_t>(provider.GetChannels());
out.write<int32_t>(provider.GetSampleRate());
out.write<int32_t>(provider.GetSampleRate() * provider.GetChannels() * provider.GetBytesPerSample());
out.write<int16_t>(provider.GetChannels() * provider.GetBytesPerSample());
out.write<int16_t>(provider.GetBytesPerSample() * 8);
out.write("data");
out.write<int32_t>(bufsize);
// samples per read
size_t spr = 65536 / bytes_per_sample;
std::vector<char> buf;
for (int64_t i = start_sample; i < end_sample; i += spr) {
spr = std::min<size_t>(spr, end_sample - i);
buf.resize(spr * bytes_per_sample);
provider.GetAudio(&buf[0], i, spr);
out.write(buf);
}
}
