2021-04-28 12:43:51 +08:00
|
|
|
/**
|
|
|
|
* OpenAL cross platform audio library
|
|
|
|
* Copyright (C) 2013 by Mike Gorchak
|
|
|
|
* This library is free software; you can redistribute it and/or
|
|
|
|
* modify it under the terms of the GNU Library General Public
|
|
|
|
* License as published by the Free Software Foundation; either
|
|
|
|
* version 2 of the License, or (at your option) any later version.
|
|
|
|
*
|
|
|
|
* This library is distributed in the hope that it will be useful,
|
|
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
|
|
|
* Library General Public License for more details.
|
|
|
|
*
|
|
|
|
* You should have received a copy of the GNU Library General Public
|
|
|
|
* License along with this library; if not, write to the
|
|
|
|
* Free Software Foundation, Inc.,
|
|
|
|
* 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
|
|
|
|
* Or go to http://www.gnu.org/copyleft/lgpl.html
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include "config.h"
|
|
|
|
|
|
|
|
#include <algorithm>
|
2021-05-14 10:15:42 +08:00
|
|
|
#include <array>
|
2021-04-28 12:43:51 +08:00
|
|
|
#include <climits>
|
|
|
|
#include <cstdlib>
|
|
|
|
#include <iterator>
|
|
|
|
|
2021-05-14 10:15:42 +08:00
|
|
|
#include "alc/effects/base.h"
|
2021-04-28 12:43:51 +08:00
|
|
|
#include "almalloc.h"
|
2022-04-25 12:02:45 +08:00
|
|
|
#include "alnumbers.h"
|
2021-04-28 12:43:51 +08:00
|
|
|
#include "alnumeric.h"
|
|
|
|
#include "alspan.h"
|
2021-05-14 10:15:42 +08:00
|
|
|
#include "core/bufferline.h"
|
|
|
|
#include "core/context.h"
|
|
|
|
#include "core/devformat.h"
|
|
|
|
#include "core/device.h"
|
2022-04-25 12:02:45 +08:00
|
|
|
#include "core/effectslot.h"
|
2021-05-14 10:15:42 +08:00
|
|
|
#include "core/mixer.h"
|
|
|
|
#include "core/mixer/defs.h"
|
|
|
|
#include "core/resampler_limits.h"
|
|
|
|
#include "intrusive_ptr.h"
|
2021-04-28 12:43:51 +08:00
|
|
|
#include "opthelpers.h"
|
|
|
|
#include "vector.h"
|
|
|
|
|
|
|
|
|
|
|
|
namespace {
|
|
|
|
|
2021-05-14 10:15:42 +08:00
|
|
|
using uint = unsigned int;
|
|
|
|
|
2021-04-28 12:43:51 +08:00
|
|
|
#define MAX_UPDATE_SAMPLES 256
|
|
|
|
|
|
|
|
struct ChorusState final : public EffectState {
|
|
|
|
al::vector<float,16> mSampleBuffer;
|
|
|
|
uint mOffset{0};
|
|
|
|
|
|
|
|
uint mLfoOffset{0};
|
|
|
|
uint mLfoRange{1};
|
|
|
|
float mLfoScale{0.0f};
|
|
|
|
uint mLfoDisp{0};
|
|
|
|
|
|
|
|
/* Gains for left and right sides */
|
|
|
|
struct {
|
|
|
|
float Current[MAX_OUTPUT_CHANNELS]{};
|
|
|
|
float Target[MAX_OUTPUT_CHANNELS]{};
|
|
|
|
} mGains[2];
|
|
|
|
|
|
|
|
/* effect parameters */
|
|
|
|
ChorusWaveform mWaveform{};
|
|
|
|
int mDelay{0};
|
|
|
|
float mDepth{0.0f};
|
|
|
|
float mFeedback{0.0f};
|
|
|
|
|
|
|
|
void getTriangleDelays(uint (*delays)[MAX_UPDATE_SAMPLES], const size_t todo);
|
|
|
|
void getSinusoidDelays(uint (*delays)[MAX_UPDATE_SAMPLES], const size_t todo);
|
|
|
|
|
2021-05-14 10:15:42 +08:00
|
|
|
void deviceUpdate(const DeviceBase *device, const Buffer &buffer) override;
|
|
|
|
void update(const ContextBase *context, const EffectSlot *slot, const EffectProps *props,
|
2021-04-28 12:43:51 +08:00
|
|
|
const EffectTarget target) override;
|
|
|
|
void process(const size_t samplesToDo, const al::span<const FloatBufferLine> samplesIn,
|
|
|
|
const al::span<FloatBufferLine> samplesOut) override;
|
|
|
|
|
|
|
|
DEF_NEWDEL(ChorusState)
|
|
|
|
};
|
|
|
|
|
2021-05-14 10:15:42 +08:00
|
|
|
void ChorusState::deviceUpdate(const DeviceBase *Device, const Buffer&)
|
2021-04-28 12:43:51 +08:00
|
|
|
{
|
2021-05-14 10:15:42 +08:00
|
|
|
constexpr float max_delay{maxf(ChorusMaxDelay, FlangerMaxDelay)};
|
2021-04-28 12:43:51 +08:00
|
|
|
|
|
|
|
const auto frequency = static_cast<float>(Device->Frequency);
|
|
|
|
const size_t maxlen{NextPowerOf2(float2uint(max_delay*2.0f*frequency) + 1u)};
|
|
|
|
if(maxlen != mSampleBuffer.size())
|
|
|
|
al::vector<float,16>(maxlen).swap(mSampleBuffer);
|
|
|
|
|
|
|
|
std::fill(mSampleBuffer.begin(), mSampleBuffer.end(), 0.0f);
|
|
|
|
for(auto &e : mGains)
|
|
|
|
{
|
|
|
|
std::fill(std::begin(e.Current), std::end(e.Current), 0.0f);
|
|
|
|
std::fill(std::begin(e.Target), std::end(e.Target), 0.0f);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2021-05-14 10:15:42 +08:00
|
|
|
void ChorusState::update(const ContextBase *Context, const EffectSlot *Slot,
|
2021-04-28 12:43:51 +08:00
|
|
|
const EffectProps *props, const EffectTarget target)
|
|
|
|
{
|
|
|
|
constexpr int mindelay{(MaxResamplerPadding>>1) << MixerFracBits};
|
|
|
|
|
|
|
|
/* The LFO depth is scaled to be relative to the sample delay. Clamp the
|
|
|
|
* delay and depth to allow enough padding for resampling.
|
|
|
|
*/
|
2021-05-14 10:15:42 +08:00
|
|
|
const DeviceBase *device{Context->mDevice};
|
2021-04-28 12:43:51 +08:00
|
|
|
const auto frequency = static_cast<float>(device->Frequency);
|
|
|
|
|
|
|
|
mWaveform = props->Chorus.Waveform;
|
|
|
|
|
|
|
|
mDelay = maxi(float2int(props->Chorus.Delay*frequency*MixerFracOne + 0.5f), mindelay);
|
|
|
|
mDepth = minf(props->Chorus.Depth * static_cast<float>(mDelay),
|
|
|
|
static_cast<float>(mDelay - mindelay));
|
|
|
|
|
|
|
|
mFeedback = props->Chorus.Feedback;
|
|
|
|
|
|
|
|
/* Gains for left and right sides */
|
|
|
|
const auto lcoeffs = CalcDirectionCoeffs({-1.0f, 0.0f, 0.0f}, 0.0f);
|
|
|
|
const auto rcoeffs = CalcDirectionCoeffs({ 1.0f, 0.0f, 0.0f}, 0.0f);
|
|
|
|
|
|
|
|
mOutTarget = target.Main->Buffer;
|
|
|
|
ComputePanGains(target.Main, lcoeffs.data(), Slot->Gain, mGains[0].Target);
|
|
|
|
ComputePanGains(target.Main, rcoeffs.data(), Slot->Gain, mGains[1].Target);
|
|
|
|
|
|
|
|
float rate{props->Chorus.Rate};
|
|
|
|
if(!(rate > 0.0f))
|
|
|
|
{
|
|
|
|
mLfoOffset = 0;
|
|
|
|
mLfoRange = 1;
|
|
|
|
mLfoScale = 0.0f;
|
|
|
|
mLfoDisp = 0;
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
/* Calculate LFO coefficient (number of samples per cycle). Limit the
|
|
|
|
* max range to avoid overflow when calculating the displacement.
|
|
|
|
*/
|
|
|
|
uint lfo_range{float2uint(minf(frequency/rate + 0.5f, float{INT_MAX/360 - 180}))};
|
|
|
|
|
|
|
|
mLfoOffset = mLfoOffset * lfo_range / mLfoRange;
|
|
|
|
mLfoRange = lfo_range;
|
|
|
|
switch(mWaveform)
|
|
|
|
{
|
|
|
|
case ChorusWaveform::Triangle:
|
|
|
|
mLfoScale = 4.0f / static_cast<float>(mLfoRange);
|
|
|
|
break;
|
|
|
|
case ChorusWaveform::Sinusoid:
|
2022-04-25 12:02:45 +08:00
|
|
|
mLfoScale = al::numbers::pi_v<float>*2.0f / static_cast<float>(mLfoRange);
|
2021-04-28 12:43:51 +08:00
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Calculate lfo phase displacement */
|
|
|
|
int phase{props->Chorus.Phase};
|
|
|
|
if(phase < 0) phase = 360 + phase;
|
|
|
|
mLfoDisp = (mLfoRange*static_cast<uint>(phase) + 180) / 360;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
void ChorusState::getTriangleDelays(uint (*delays)[MAX_UPDATE_SAMPLES], const size_t todo)
|
|
|
|
{
|
|
|
|
const uint lfo_range{mLfoRange};
|
|
|
|
const float lfo_scale{mLfoScale};
|
|
|
|
const float depth{mDepth};
|
|
|
|
const int delay{mDelay};
|
|
|
|
|
|
|
|
ASSUME(lfo_range > 0);
|
|
|
|
ASSUME(todo > 0);
|
|
|
|
|
|
|
|
uint offset{mLfoOffset};
|
|
|
|
auto gen_lfo = [&offset,lfo_range,lfo_scale,depth,delay]() -> uint
|
|
|
|
{
|
|
|
|
offset = (offset+1)%lfo_range;
|
|
|
|
const float offset_norm{static_cast<float>(offset) * lfo_scale};
|
|
|
|
return static_cast<uint>(fastf2i((1.0f-std::abs(2.0f-offset_norm)) * depth) + delay);
|
|
|
|
};
|
|
|
|
std::generate_n(delays[0], todo, gen_lfo);
|
|
|
|
|
|
|
|
offset = (mLfoOffset+mLfoDisp) % lfo_range;
|
|
|
|
std::generate_n(delays[1], todo, gen_lfo);
|
|
|
|
|
|
|
|
mLfoOffset = static_cast<uint>(mLfoOffset+todo) % lfo_range;
|
|
|
|
}
|
|
|
|
|
|
|
|
void ChorusState::getSinusoidDelays(uint (*delays)[MAX_UPDATE_SAMPLES], const size_t todo)
|
|
|
|
{
|
|
|
|
const uint lfo_range{mLfoRange};
|
|
|
|
const float lfo_scale{mLfoScale};
|
|
|
|
const float depth{mDepth};
|
|
|
|
const int delay{mDelay};
|
|
|
|
|
|
|
|
ASSUME(lfo_range > 0);
|
|
|
|
ASSUME(todo > 0);
|
|
|
|
|
|
|
|
uint offset{mLfoOffset};
|
|
|
|
auto gen_lfo = [&offset,lfo_range,lfo_scale,depth,delay]() -> uint
|
|
|
|
{
|
|
|
|
offset = (offset+1)%lfo_range;
|
|
|
|
const float offset_norm{static_cast<float>(offset) * lfo_scale};
|
|
|
|
return static_cast<uint>(fastf2i(std::sin(offset_norm)*depth) + delay);
|
|
|
|
};
|
|
|
|
std::generate_n(delays[0], todo, gen_lfo);
|
|
|
|
|
|
|
|
offset = (mLfoOffset+mLfoDisp) % lfo_range;
|
|
|
|
std::generate_n(delays[1], todo, gen_lfo);
|
|
|
|
|
|
|
|
mLfoOffset = static_cast<uint>(mLfoOffset+todo) % lfo_range;
|
|
|
|
}
|
|
|
|
|
|
|
|
void ChorusState::process(const size_t samplesToDo, const al::span<const FloatBufferLine> samplesIn, const al::span<FloatBufferLine> samplesOut)
|
|
|
|
{
|
|
|
|
const size_t bufmask{mSampleBuffer.size()-1};
|
|
|
|
const float feedback{mFeedback};
|
|
|
|
const uint avgdelay{(static_cast<uint>(mDelay) + (MixerFracOne>>1)) >> MixerFracBits};
|
|
|
|
float *RESTRICT delaybuf{mSampleBuffer.data()};
|
|
|
|
uint offset{mOffset};
|
|
|
|
|
|
|
|
for(size_t base{0u};base < samplesToDo;)
|
|
|
|
{
|
|
|
|
const size_t todo{minz(MAX_UPDATE_SAMPLES, samplesToDo-base)};
|
|
|
|
|
|
|
|
uint moddelays[2][MAX_UPDATE_SAMPLES];
|
|
|
|
if(mWaveform == ChorusWaveform::Sinusoid)
|
|
|
|
getSinusoidDelays(moddelays, todo);
|
|
|
|
else /*if(mWaveform == ChorusWaveform::Triangle)*/
|
|
|
|
getTriangleDelays(moddelays, todo);
|
|
|
|
|
|
|
|
alignas(16) float temps[2][MAX_UPDATE_SAMPLES];
|
|
|
|
for(size_t i{0u};i < todo;++i)
|
|
|
|
{
|
|
|
|
// Feed the buffer's input first (necessary for delays < 1).
|
|
|
|
delaybuf[offset&bufmask] = samplesIn[0][base+i];
|
|
|
|
|
|
|
|
// Tap for the left output.
|
|
|
|
uint delay{offset - (moddelays[0][i]>>MixerFracBits)};
|
|
|
|
float mu{static_cast<float>(moddelays[0][i]&MixerFracMask) * (1.0f/MixerFracOne)};
|
|
|
|
temps[0][i] = cubic(delaybuf[(delay+1) & bufmask], delaybuf[(delay ) & bufmask],
|
|
|
|
delaybuf[(delay-1) & bufmask], delaybuf[(delay-2) & bufmask], mu);
|
|
|
|
|
|
|
|
// Tap for the right output.
|
|
|
|
delay = offset - (moddelays[1][i]>>MixerFracBits);
|
|
|
|
mu = static_cast<float>(moddelays[1][i]&MixerFracMask) * (1.0f/MixerFracOne);
|
|
|
|
temps[1][i] = cubic(delaybuf[(delay+1) & bufmask], delaybuf[(delay ) & bufmask],
|
|
|
|
delaybuf[(delay-1) & bufmask], delaybuf[(delay-2) & bufmask], mu);
|
|
|
|
|
|
|
|
// Accumulate feedback from the average delay of the taps.
|
|
|
|
delaybuf[offset&bufmask] += delaybuf[(offset-avgdelay) & bufmask] * feedback;
|
|
|
|
++offset;
|
|
|
|
}
|
|
|
|
|
2021-05-14 10:15:42 +08:00
|
|
|
for(size_t c{0};c < 2;++c)
|
2021-04-28 12:43:51 +08:00
|
|
|
MixSamples({temps[c], todo}, samplesOut, mGains[c].Current, mGains[c].Target,
|
|
|
|
samplesToDo-base, base);
|
|
|
|
|
|
|
|
base += todo;
|
|
|
|
}
|
|
|
|
|
|
|
|
mOffset = offset;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
struct ChorusStateFactory final : public EffectStateFactory {
|
|
|
|
al::intrusive_ptr<EffectState> create() override
|
|
|
|
{ return al::intrusive_ptr<EffectState>{new ChorusState{}}; }
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
|
|
/* Flanger is basically a chorus with a really short delay. They can both use
|
|
|
|
* the same processing functions, so piggyback flanger on the chorus functions.
|
|
|
|
*/
|
|
|
|
struct FlangerStateFactory final : public EffectStateFactory {
|
|
|
|
al::intrusive_ptr<EffectState> create() override
|
|
|
|
{ return al::intrusive_ptr<EffectState>{new ChorusState{}}; }
|
|
|
|
};
|
|
|
|
|
|
|
|
} // namespace
|
|
|
|
|
|
|
|
EffectStateFactory *ChorusStateFactory_getFactory()
|
|
|
|
{
|
|
|
|
static ChorusStateFactory ChorusFactory{};
|
|
|
|
return &ChorusFactory;
|
|
|
|
}
|
|
|
|
|
|
|
|
EffectStateFactory *FlangerStateFactory_getFactory()
|
|
|
|
{
|
|
|
|
static FlangerStateFactory FlangerFactory{};
|
|
|
|
return &FlangerFactory;
|
|
|
|
}
|