OpenRCT2/src/openrct2-ui/audio/AudioMixer.cpp

425 lines
13 KiB
C++

/*****************************************************************************
* Copyright (c) 2014-2024 OpenRCT2 developers
*
* For a complete list of all authors, please refer to contributors.md
* Interested in contributing? Visit https://github.com/OpenRCT2/OpenRCT2
*
* OpenRCT2 is licensed under the GNU General Public License version 3.
*****************************************************************************/
#include "AudioMixer.h"
#include <algorithm>
#include <iterator>
#include <openrct2/OpenRCT2.h>
#include <openrct2/config/Config.h>
#include <speex/speex_resampler.h>
using namespace OpenRCT2::Audio;
AudioMixer::~AudioMixer()
{
Close();
}
void AudioMixer::Init(const char* device)
{
Close();
SDL_AudioSpec want = {};
want.freq = 22050;
want.format = AUDIO_S16SYS;
want.channels = 2;
want.samples = 2048;
want.callback = [](void* arg, uint8_t* dst, int32_t length) -> void {
auto* mixer = static_cast<AudioMixer*>(arg);
mixer->GetNextAudioChunk(dst, static_cast<size_t>(length));
mixer->RemoveReleasedSources();
};
want.userdata = this;
SDL_AudioSpec have;
_deviceId = SDL_OpenAudioDevice(device, 0, &want, &have, 0);
_format.format = have.format;
_format.channels = have.channels;
_format.freq = have.freq;
SDL_PauseAudioDevice(_deviceId, 0);
}
void AudioMixer::Close()
{
// Free channels
Lock();
_channels.clear();
Unlock();
SDL_CloseAudioDevice(_deviceId);
// Free buffers
_channelBuffer.clear();
_channelBuffer.shrink_to_fit();
_convertBuffer.clear();
_convertBuffer.shrink_to_fit();
_effectBuffer.clear();
_effectBuffer.shrink_to_fit();
}
void AudioMixer::Lock()
{
SDL_LockAudioDevice(_deviceId);
}
void AudioMixer::Unlock()
{
SDL_UnlockAudioDevice(_deviceId);
}
std::shared_ptr<IAudioChannel> AudioMixer::Play(IAudioSource* source, int32_t loop, bool deleteondone)
{
Lock();
auto channel = std::shared_ptr<ISDLAudioChannel>(AudioChannel::Create());
if (channel != nullptr)
{
channel->Play(source, loop);
channel->SetDeleteOnDone(deleteondone);
_channels.push_back(channel);
}
Unlock();
return channel;
}
void AudioMixer::SetVolume(float volume)
{
_volume = volume;
}
SDLAudioSource* AudioMixer::AddSource(std::unique_ptr<SDLAudioSource> source)
{
std::lock_guard<std::mutex> guard(_mutex);
if (source != nullptr)
{
_sources.push_back(std::move(source));
return _sources.back().get();
}
return nullptr;
}
void AudioMixer::RemoveReleasedSources()
{
std::lock_guard<std::mutex> guard(_mutex);
_sources.erase(
std::remove_if(
_sources.begin(), _sources.end(),
[](std::unique_ptr<SDLAudioSource>& source) {
{
return source->IsReleased();
}
}),
_sources.end());
}
const AudioFormat& AudioMixer::GetFormat() const
{
return _format;
}
void AudioMixer::GetNextAudioChunk(uint8_t* dst, size_t length)
{
UpdateAdjustedSound();
// Zero the output buffer
std::fill_n(dst, length, 0);
// Mix channels onto output buffer
auto it = _channels.begin();
while (it != _channels.end())
{
auto& channel = *it;
auto channelSource = channel->GetSource();
auto channelSourceReleased = channelSource == nullptr || channelSource->IsReleased();
if (channelSourceReleased || (channel->IsDone() && channel->DeleteOnDone()) || channel->IsStopping())
{
channel->SetDone(true);
it = _channels.erase(it);
}
else
{
auto group = channel->GetGroup();
if ((group != MixerGroup::Sound || gConfigSound.SoundEnabled) && gConfigSound.MasterSoundEnabled
&& gConfigSound.MasterVolume != 0)
{
MixChannel(channel.get(), dst, length);
}
it++;
}
}
}
void AudioMixer::UpdateAdjustedSound()
{
// Did the volume level get changed? Recalculate level in this case.
if (_settingSoundVolume != gConfigSound.SoundVolume)
{
_settingSoundVolume = gConfigSound.SoundVolume;
_adjustSoundVolume = powf(static_cast<float>(_settingSoundVolume) / 100.f, 10.f / 6.f);
}
if (_settingMusicVolume != gConfigSound.AudioFocus)
{
_settingMusicVolume = gConfigSound.AudioFocus;
_adjustMusicVolume = powf(static_cast<float>(_settingMusicVolume) / 100.f, 10.f / 6.f);
}
}
void AudioMixer::MixChannel(ISDLAudioChannel* channel, uint8_t* data, size_t length)
{
int32_t byteRate = _format.GetByteRate();
auto numSamples = static_cast<int32_t>(length / byteRate);
double rate = 1;
if (_format.format == AUDIO_S16SYS)
{
rate = channel->GetRate();
}
bool mustConvert = false;
SDL_AudioCVT cvt;
cvt.len_ratio = 1;
AudioFormat streamformat = channel->GetFormat();
if (streamformat != _format)
{
if (SDL_BuildAudioCVT(
&cvt, streamformat.format, streamformat.channels, streamformat.freq, _format.format, _format.channels,
_format.freq)
== -1)
{
// Unable to convert channel data
return;
}
mustConvert = true;
}
// Read raw PCM from channel
int32_t readSamples = numSamples * rate;
auto readLength = static_cast<size_t>(readSamples / cvt.len_ratio) * byteRate;
_channelBuffer.resize(readLength);
size_t bytesRead = channel->Read(_channelBuffer.data(), readLength);
// Convert data to required format if necessary
void* buffer = nullptr;
size_t bufferLen = 0;
if (mustConvert)
{
if (Convert(&cvt, _channelBuffer.data(), bytesRead))
{
buffer = cvt.buf;
bufferLen = cvt.len_cvt;
}
else
{
return;
}
}
else
{
buffer = _channelBuffer.data();
bufferLen = bytesRead;
}
// Apply effects
if (rate != 1)
{
auto inRate = static_cast<int32_t>(bufferLen / byteRate);
int32_t outRate = numSamples;
if (bytesRead != readLength)
{
inRate = _format.freq;
outRate = _format.freq * (1 / rate);
}
_effectBuffer.resize(length);
bufferLen = ApplyResample(channel, buffer, static_cast<int32_t>(bufferLen / byteRate), numSamples, inRate, outRate);
buffer = _effectBuffer.data();
}
// Apply panning and volume
ApplyPan(channel, buffer, bufferLen, byteRate);
int32_t mixVolume = ApplyVolume(channel, buffer, bufferLen);
// Finally mix on to destination buffer
size_t dstLength = std::min(length, bufferLen);
SDL_MixAudioFormat(data, static_cast<const uint8_t*>(buffer), _format.format, static_cast<uint32_t>(dstLength), mixVolume);
channel->UpdateOldVolume();
}
/**
* Resample the given buffer into _effectBuffer.
* Assumes that srcBuffer is the same format as _format.
*/
size_t AudioMixer::ApplyResample(
ISDLAudioChannel* channel, const void* srcBuffer, int32_t srcSamples, int32_t dstSamples, int32_t inRate, int32_t outRate)
{
int32_t byteRate = _format.GetByteRate();
// Create resampler
SpeexResamplerState* resampler = channel->GetResampler();
if (resampler == nullptr)
{
resampler = speex_resampler_init(_format.channels, _format.freq, _format.freq, 0, nullptr);
channel->SetResampler(resampler);
}
speex_resampler_set_rate(resampler, inRate, outRate);
uint32_t inLen = srcSamples;
uint32_t outLen = dstSamples;
speex_resampler_process_interleaved_int(
resampler, static_cast<const spx_int16_t*>(srcBuffer), &inLen, reinterpret_cast<spx_int16_t*>(_effectBuffer.data()),
&outLen);
return outLen * byteRate;
}
void AudioMixer::ApplyPan(const IAudioChannel* channel, void* buffer, size_t len, size_t sampleSize)
{
if (channel->GetPan() != 0.5f && _format.channels == 2)
{
switch (_format.format)
{
case AUDIO_S16SYS:
EffectPanS16(channel, static_cast<int16_t*>(buffer), static_cast<int32_t>(len / sampleSize));
break;
case AUDIO_U8:
EffectPanU8(channel, static_cast<uint8_t*>(buffer), static_cast<int32_t>(len / sampleSize));
break;
}
}
}
int32_t AudioMixer::ApplyVolume(const IAudioChannel* channel, void* buffer, size_t len)
{
float volumeAdjust = _volume;
volumeAdjust *= gConfigSound.MasterSoundEnabled ? (static_cast<float>(gConfigSound.MasterVolume) / 100.0f) : 0.0f;
switch (channel->GetGroup())
{
case MixerGroup::Sound:
volumeAdjust *= _adjustSoundVolume;
// Cap sound volume on title screen so music is more audible
if (gScreenFlags & SCREEN_FLAGS_TITLE_DEMO)
{
volumeAdjust = std::min(volumeAdjust, 0.75f);
}
break;
case MixerGroup::RideMusic:
case MixerGroup::TitleMusic:
volumeAdjust *= _adjustMusicVolume;
break;
}
int32_t startVolume = channel->GetOldVolume() * volumeAdjust;
int32_t endVolume = channel->GetVolume() * volumeAdjust;
if (channel->IsStopping())
{
endVolume = 0;
}
int32_t mixVolume = channel->GetVolume() * volumeAdjust;
if (startVolume != endVolume)
{
// Set to max since we are adjusting the volume ourselves
mixVolume = MIXER_VOLUME_MAX;
// Fade between volume levels to smooth out sound and minimize clicks from sudden volume changes
int32_t fadeLength = static_cast<int32_t>(len) / _format.BytesPerSample();
switch (_format.format)
{
case AUDIO_S16SYS:
EffectFadeS16(static_cast<int16_t*>(buffer), fadeLength, startVolume, endVolume);
break;
case AUDIO_U8:
EffectFadeU8(static_cast<uint8_t*>(buffer), fadeLength, startVolume, endVolume);
break;
}
}
return mixVolume;
}
void AudioMixer::EffectPanS16(const IAudioChannel* channel, int16_t* data, int32_t length)
{
const float dt = 1.0f / static_cast<float>(length * 2.0f);
float volumeL = channel->GetOldVolumeL();
float volumeR = channel->GetOldVolumeR();
const float d_left = dt * (channel->GetVolumeL() - channel->GetOldVolumeL());
const float d_right = dt * (channel->GetVolumeR() - channel->GetOldVolumeR());
for (int32_t i = 0; i < length * 2; i += 2)
{
data[i + 0] = static_cast<int16_t>(volumeL * static_cast<float>(data[i + 0]));
data[i + 1] = static_cast<int16_t>(volumeR * static_cast<float>(data[i + 1]));
volumeL += d_left;
volumeR += d_right;
}
}
void AudioMixer::EffectPanU8(const IAudioChannel* channel, uint8_t* data, int32_t length)
{
float volumeL = channel->GetVolumeL();
float volumeR = channel->GetVolumeR();
float oldVolumeL = channel->GetOldVolumeL();
float oldVolumeR = channel->GetOldVolumeR();
for (int32_t i = 0; i < length * 2; i += 2)
{
float t = static_cast<float>(i) / static_cast<float>(length * 2.0f);
data[i] = static_cast<uint8_t>(data[i] * ((1.0 - t) * oldVolumeL + t * volumeL));
data[i + 1] = static_cast<uint8_t>(data[i + 1] * ((1.0 - t) * oldVolumeR + t * volumeR));
}
}
void AudioMixer::EffectFadeS16(int16_t* data, int32_t length, int32_t startvolume, int32_t endvolume)
{
static_assert(SDL_MIX_MAXVOLUME == MIXER_VOLUME_MAX, "Max volume differs between OpenRCT2 and SDL2");
float startvolume_f = static_cast<float>(startvolume) / SDL_MIX_MAXVOLUME;
float endvolume_f = static_cast<float>(endvolume) / SDL_MIX_MAXVOLUME;
for (int32_t i = 0; i < length; i++)
{
float t = static_cast<float>(i) / length;
data[i] = static_cast<int16_t>(data[i] * ((1.0f - t) * startvolume_f + t * endvolume_f));
}
}
void AudioMixer::EffectFadeU8(uint8_t* data, int32_t length, int32_t startvolume, int32_t endvolume)
{
static_assert(SDL_MIX_MAXVOLUME == MIXER_VOLUME_MAX, "Max volume differs between OpenRCT2 and SDL2");
float startvolume_f = static_cast<float>(startvolume) / SDL_MIX_MAXVOLUME;
float endvolume_f = static_cast<float>(endvolume) / SDL_MIX_MAXVOLUME;
for (int32_t i = 0; i < length; i++)
{
float t = static_cast<float>(i) / length;
data[i] = static_cast<uint8_t>(data[i] * ((1.0f - t) * startvolume_f + t * endvolume_f));
}
}
bool AudioMixer::Convert(SDL_AudioCVT* cvt, const void* src, size_t len)
{
// tofix: there seems to be an issue with converting audio using SDL_ConvertAudio in the callback vs preconverted,
// can cause pops and static depending on sample rate and channels
bool result = false;
if (len != 0 && cvt->len_mult != 0)
{
size_t reqConvertBufferCapacity = len * cvt->len_mult;
_convertBuffer.resize(reqConvertBufferCapacity);
std::copy_n(static_cast<const uint8_t*>(src), len, _convertBuffer.data());
cvt->len = static_cast<int32_t>(len);
cvt->buf = static_cast<uint8_t*>(_convertBuffer.data());
if (SDL_ConvertAudio(cvt) >= 0)
{
result = true;
}
}
return result;
}