transverbformalities.cpp

/*------------------------------------------------------------------------
Copyright (C) 2001-2022  Tom Murphy 7 and Sophia Poirier

This file is part of Transverb.

Transverb is free software:  you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 2 of the License, or
(at your option) any later version.

Transverb 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 General Public License for more details.

You should have received a copy of the GNU General Public License
along with Transverb.  If not, see <https://www.gnu.org/licenses/>.

To contact the author, use the contact form at https://destroyfx.org
------------------------------------------------------------------------*/

#include "transverb.h"

#include <algorithm>
#include <array>
#include <cassert>
#include <cmath>
#include <cstring>
#include <numeric>

#include "dfxmisc.h"
#include "firfilter.h"


// these are macros that do boring entry point stuff for us
DFX_EFFECT_ENTRY(Transverb)
DFX_CORE_ENTRY(TransverbDSP)

using namespace dfx::TV;



Transverb::Transverb(TARGET_API_BASE_INSTANCE_TYPE inInstance)
  : DfxPlugin(inInstance, kNumParameters, kNumPresets) {

  initparameter_f(kBsize, {"buffer size", "BufSize", "BufSiz", "BfSz"}, 2700.0, 333.0, 1.0, 3000.0, DfxParam::Unit::MS);
  initparameter_f(kDrymix, {"dry mix", "DryMix", "Dry"}, 1.0, 1.0, 0.0, 1.0, DfxParam::Unit::LinearGain, DfxParam::Curve::Squared);
  initparameter_f(kMix1, {"1:mix", "1mix"}, 1.0, 1.0, 0.0, 1.0, DfxParam::Unit::LinearGain, DfxParam::Curve::Squared);
  initparameter_f(kDist1, {"1:distance", "1:dist", "1dst"}, 0.90009, 0.5, 0.0, 1.0, DfxParam::Unit::Scalar);
  initparameter_f(kSpeed1, {"1:speed", "1speed", "1spd"}, 0.0, 0.0, -3.0, 6.0, DfxParam::Unit::Octaves);
  initparameter_f(kFeed1, {"1:feedback", "1feedbk", "1fedbk", "1fdb"}, 0.0, 33.3, 0.0, 100.0, DfxParam::Unit::Percent);
  initparameter_f(kMix2, {"2:mix", "2mix"}, 0.0, 1.0, 0.0, 1.0, DfxParam::Unit::LinearGain, DfxParam::Curve::Squared);
  initparameter_f(kDist2, {"2:distance", "2:dist", "2dst"}, 0.1, 0.5, 0.0, 1.0, DfxParam::Unit::Scalar);
  initparameter_f(kSpeed2, {"2:speed", "2speed", "2spd"}, 1.0, 0.0, -3.0, 6.0, DfxParam::Unit::Octaves);
  initparameter_f(kFeed2, {"2:feedback", "2feedbk", "2fedbk", "2fdb"}, 0.0, 33.3, 0.0, 100.0, DfxParam::Unit::Percent);
  initparameter_list(kQuality, {"quality", "Qualty", "Qlty"}, kQualityMode_UltraHiFi, kQualityMode_UltraHiFi, kQualityMode_NumModes);
  initparameter_b(kTomsound, {"TOMSOUND", "TomSnd", "Tom7"}, false);
  initparameter_b(kFreeze, dfx::MakeParameterNames(dfx::kParameterNames_Freeze), false);
  initparameter_b(kAttenuateFeedbackByMixLevel, {"attenuate feedback by mix level", "AtnFdbk", "AtnFdb", "-fdb"}, false);
  initparameter_list(kDistChangeMode, {"distance change mode", "DistMod", "DstMod", "DMod"}, kDistChangeMode_Reverse, kDistChangeMode_Reverse, kDistChangeMode_Count);

  setparameterenforcevaluelimits(kBsize, true);
  for (auto const parameterID : kDistParameters) {
    setparameterenforcevaluelimits(parameterID, true);
  }

  setparametervaluestring(kQuality, kQualityMode_DirtFi, "dirt-fi");
  setparametervaluestring(kQuality, kQualityMode_HiFi, "hi-fi");
  setparametervaluestring(kQuality, kQualityMode_UltraHiFi, "ultra hi-fi");

  setparametervaluestring(kDistChangeMode, kDistChangeMode_Reverse, "reverse");
  setparametervaluestring(kDistChangeMode, kDistChangeMode_AdHocVarispeed, "ad hoc varispeed");
  setparametervaluestring(kDistChangeMode, kDistChangeMode_DistanceVarispeed, "distance varispeed");
  setparametervaluestring(kDistChangeMode, kDistChangeMode_BufferVarispeed, "buffer varispeed");
  setparametervaluestring(kDistChangeMode, kDistChangeMode_LoopingBufferVarispeed, "looping buffer varispeed");

  addparameterattributes(kFreeze, DfxParam::kAttribute_OmitFromRandomizeAll);
  addparameterattributes(kAttenuateFeedbackByMixLevel, DfxParam::kAttribute_OmitFromRandomizeAll);

  std::vector<dfx::ParameterID> mixparameters(1, kDrymix);
  for (size_t head = 0; head < kNumDelays; head++) {
    addparametergroup("head #" + std::to_string(head + 1),
      {kSpeedParameters[head], kFeedParameters[head], kDistParameters[head]});
    mixparameters.push_back(kMixParameters[head]);
  }
  addparametergroup("mix", mixparameters);


  settailsize_seconds(getparametermax_f(kBsize) * 0.001);

  setpresetname(0, PLUGIN_NAME_STRING);  // default preset name
  initPresets();

  addchannelconfig(kChannelConfig_AnyMatchedIO);  // N-in/N-out
  addchannelconfig(1, kNumDelays);  // mono-input / per-head-output

  speedModeStates.fill(kSpeedMode_Fine);
}

void Transverb::dfx_PostConstructor() {

#if TARGET_PLUGIN_USES_MIDI
  // since we don't use notes for any specialized control of Transverb,
  // allow them to be assigned to control parameters via MIDI learn
  getsettings().setAllowPitchbendEvents(true);
  getsettings().setAllowNoteEvents(true);
#endif
}



TransverbDSP::TransverbDSP(DfxPlugin* inDfxPlugin)
  : DfxPluginCore(inDfxPlugin),
    MAXBUF(static_cast<int>(getparametermax_f(kBsize) * 0.001 * getsamplerate())),
    buftemp(dynamic_cast<Transverb*>(inDfxPlugin)->getscratchbuffer()),
    firCoefficientsWindow(dfx::FIRFilter::generateKaiserWindow(kNumFIRTaps, 60.0f)) {

  buftemp.assign(MAXBUF, 0.f);

  registerSmoothedAudioValue(&drymix);

  for (auto& head : heads) {
    head.buf.assign(MAXBUF, 0.f);
    head.filter.setSampleRate(getsamplerate());
    registerSmoothedAudioValue(&head.speed);
    registerSmoothedAudioValue(&head.mix);
    registerSmoothedAudioValue(&head.feed);
  }
}


void TransverbDSP::reset() {

  std::for_each(heads.begin(), heads.end(), [](Head& head){ head.reset(); });

  firstrendersincereset = true;
}

void TransverbDSP::Head::reset() {

  smoothcount = 0;
  lastdelayval = 0.f;
  targetdist.reset();
  filter.reset();
  speedHasChanged = true;

  std::fill(buf.begin(), buf.end(), 0.f);
}


void TransverbDSP::processparameters() {

  if (auto const value = getparameterifchanged_f(kDrymix))
  {
    // balance the audio energy when mono input is fanned out to multiple output channels
    auto const dryGainScalar = std::sqrt(static_cast<double>(getplugin()->getnuminputs()) /
                                         static_cast<double>(getplugin()->getnumoutputs()));
    drymix = *value * dryGainScalar;
  }
  for (size_t head = 0; head < kNumDelays; head++)
  {
    if (auto const value = getparameterifchanged_f(kMixParameters[head]))
    {
      heads[head].mix = *value;
    }
    if (auto const value = getparameterifchanged_f(kSpeedParameters[head]))
    {
      heads[head].speed = std::pow(2., *value);
      heads[head].speedHasChanged = true;
    }
    if (auto const value = getparameterifchanged_scalar(kFeedParameters[head]))
    {
      heads[head].feed = *value;
    }
  }

  if (auto const value = getparameterifchanged_f(kBsize))
  {
    auto const entryBsize = std::exchange(bsize, std::clamp(static_cast<int>(*value * getsamplerate() * 0.001), 1, MAXBUF));
    // the commented-out logic below is the beginning of an attempt to clean up the buffer states
    // when the buffer resizes, however the tidiness maybe feels counter to the spirit of Transverb?
    if (bsize > entryBsize)
    {
      //for (auto& head : heads)
      {
        //std::fill(std::next(head.buf.begin(), entryBsize), std::next(head.buf.begin(), bsize), 0.f);
      }
    }
    else if (writer > bsize)
    {
      //auto const entryWriter = writer;
      writer %= bsize;
      //auto const copyCount = std::min(entryBsize - bsize, writer);
      //for (auto& head : heads)
      {
        //std::copy_n(std::next(head.buf.cbegin(), entryWriter - copyCount), copyCount, std::next(head.buf.begin(), writer - copyCount));
      }
    }
    auto const bsize_f = static_cast<double>(bsize);
    std::for_each(heads.begin(), heads.end(), [bsize_f](Head& head){ head.read = fmod_bipolar(head.read, bsize_f); });
  }

  distchangemode = getparameter_i(kDistChangeMode);
  for (size_t head = 0; head < kNumDelays; head++)
  {
    if (auto const dist = getparameterifchanged_f(kDistParameters[head]))
    {
      processdist(*dist, heads[head]);
    }
  }

  if (getparameterchanged(kQuality) || getparameterchanged(kTomsound))
  {
    std::for_each(heads.begin(), heads.end(), [](Head& head){ head.speedHasChanged = true; });
  }

  // stereo-split-heads mode (head 1 goes to left output and 2 to right)
  if (getplugin()->asymmetricalchannels())
  {
    static_assert(kNumDelays >= 2);
    assert(getplugin()->getnumoutputs() == 2);
    if (GetChannelNum() == 0)
    {
      heads[1].mix.setValueNow(getparametermin_f(kMix2));
    }
    else if (GetChannelNum() == 1)
    {
      heads[0].mix.setValueNow(getparametermin_f(kMix1));
    }
  }

  firstrendersincereset = false;
}

double TransverbDSP::getdist(double read) const {

  return fmod_bipolar(static_cast<double>(writer) - read, static_cast<double>(bsize));
}

void TransverbDSP::processdist(double distnormalized, Head& head) {

  auto const bsize_f = static_cast<double>(bsize);
  auto const distsamples = distnormalized * bsize_f;
  auto const targetread = fmod_bipolar(static_cast<double>(writer) - distsamples, bsize_f);

  if (firstrendersincereset)
  {
    head.read = targetread;
  }
  else
  {
    if (distchangemode == kDistChangeMode_Reverse)
    {
      head.targetdist = distsamples;
    }
    else if (auto const resamplerate = getdist(head.read) / std::max(distsamples, 1.); distchangemode == kDistChangeMode_AdHocVarispeed)
    {
      head.targetdist = distsamples;
      head.distspeedfactor = resamplerate;
      head.speedHasChanged = true;
    }
    else
    {
      constexpr double modulationthresholdsamples = 1.;
      if (std::fabs(targetread - head.read) >= modulationthresholdsamples)
      {
        auto const bsizesteps = bsize_f / resamplerate;
        bool const subslice = (distchangemode == kDistChangeMode_DistanceVarispeed);
        bool const looping = (distchangemode == kDistChangeMode_LoopingBufferVarispeed);
        auto const copylength_f = subslice ? distsamples : (looping ? bsize_f : std::min(bsize_f, bsizesteps));
        auto const copylength = static_cast<size_t>(std::lround(copylength_f));
        assert(copylength <= buftemp.size());
        assert(static_cast<int>(copylength) <= bsize);
        auto const sourcestart = subslice ? head.read : fmod_bipolar(static_cast<double>(writer) - (copylength_f * resamplerate), bsize_f);
        for (size_t i = 0; i < copylength; i++)
        {
          auto const sourcepos = std::fmod(sourcestart + (resamplerate * static_cast<double>(i)), bsize_f);
          buftemp[i] = interpolateHermite(head.buf.data(), sourcepos, bsize, writer);
        }
        auto const destinationstart = subslice ? std::lround(targetread) : mod_bipolar(writer - static_cast<int>(copylength), bsize);
        auto const copylength1 = std::min(static_cast<size_t>(bsize - destinationstart), copylength);
        auto const copylength2 = copylength - copylength1;
        std::copy_n(buftemp.cbegin(), copylength1, std::next(head.buf.begin(), destinationstart));
        std::copy_n(std::next(buftemp.cbegin(), copylength1), copylength2, head.buf.begin());
      }
      head.read = targetread;
    }
  }
}



//--------- presets --------

void Transverb::initPresets() {

	size_t i = 1;

	setpresetname(i, "phaser up");
	setpresetparameter_f(i, kBsize, 48.687074827);
	setpresetparameter_f(i, kDrymix, 0.45);
	setpresetparameter_f(i, kMix1, 0.5);
	setpresetparameter_f(i, kDist1, 0.1);
	setpresetparameter_f(i, kSpeed1, 0.048406605 / 12.);
	setpresetparameter_f(i, kFeed1, 33.5);
	setpresetparameter_f(i, kMix2, 0.0);
	setpresetparameter_f(i, kDist2, 0.0);
	setpresetparameter_f(i, kSpeed2, getparametermin_f(kSpeed2));
	setpresetparameter_f(i, kFeed2, 0.0);
	setpresetparameter_i(i, kQuality, kQualityMode_UltraHiFi);
	setpresetparameter_b(i, kTomsound, false);
	setpresetparameter_b(i, kFreeze, false);
	i++;

	setpresetname(i, "phaser down");
	setpresetparameter_f(i, kBsize, 27.0);
	setpresetparameter_f(i, kDrymix, 0.45);
	setpresetparameter_f(i, kMix1, 0.5);
	setpresetparameter_f(i, kDist1, 0.0);
	setpresetparameter_f(i, kSpeed1, -0.12 / 12.);//-0.048542333 / 12.);
	setpresetparameter_f(i, kFeed1, 38.);
	setpresetparameter_f(i, kMix2, 0.0);
	setpresetparameter_f(i, kDist2, 0.0);
	setpresetparameter_f(i, kSpeed2, getparametermin_f(kSpeed2));
	setpresetparameter_f(i, kFeed2, 0.0);
	setpresetparameter_i(i, kQuality, kQualityMode_UltraHiFi);
	setpresetparameter_b(i, kTomsound, false);
	setpresetparameter_b(i, kFreeze, false);
	i++;

	setpresetname(i, "aquinas");
	setpresetparameter_f(i, kBsize, 2605.1);
	setpresetparameter_f(i, kDrymix, 0.6276);
	setpresetparameter_f(i, kMix1, 1.0);
	setpresetparameter_f(i, kDist1, 0.007);
	setpresetparameter_f(i, kSpeed1, -4.665660556);
	setpresetparameter_f(i, kFeed1, 54.0);
	setpresetparameter_f(i, kMix2, 0.757);
	setpresetparameter_f(i, kDist2, 0.557);
	setpresetparameter_f(i, kSpeed2, 2.444534569);
	setpresetparameter_f(i, kFeed2, 34.822);
	setpresetparameter_i(i, kQuality, kQualityMode_UltraHiFi);
	setpresetparameter_b(i, kTomsound, false);
	setpresetparameter_b(i, kFreeze, false);
	i++;

	setpresetname(i, "glup drums");
	setpresetparameter_f(i, kBsize, 184.27356);
	setpresetparameter_f(i, kDrymix, 0.157);
	setpresetparameter_f(i, kMix1, 1.0);
	setpresetparameter_f(i, kDist1, 0.9055);
	setpresetparameter_f(i, kSpeed1, -8. / 12.);
	setpresetparameter_f(i, kFeed1, 97.6);
	setpresetparameter_f(i, kMix2, 0.0);
	setpresetparameter_f(i, kDist2, 0.803);
	setpresetparameter_f(i, kSpeed2, 0.978195651);
	setpresetparameter_f(i, kFeed2, 0.0);
	setpresetparameter_i(i, kQuality, kQualityMode_UltraHiFi);
	setpresetparameter_b(i, kTomsound, false);
	setpresetparameter_b(i, kFreeze, false);
	i++;

	setpresetname(i, "space invaders");
	setpresetparameter_f(i, kSpeed1, -0.23 / 12.);
	setpresetparameter_f(i, kFeed1, 73.0);
	setpresetparameter_f(i, kDist1, 0.8143);
	setpresetparameter_f(i, kSpeed2, 4.3 / 12.);
	setpresetparameter_f(i, kFeed2, 5.0225);
	setpresetparameter_f(i, kDist2, 0.4006);
	setpresetparameter_f(i, kBsize, 16.8);
	setpresetparameter_f(i, kDrymix, 0.000576);
	setpresetparameter_f(i, kMix1, 1.0);
	setpresetparameter_f(i, kMix2, 0.1225);
	setpresetparameter_i(i, kQuality, kQualityMode_UltraHiFi);
	setpresetparameter_b(i, kTomsound, false);
	setpresetparameter_b(i, kFreeze, false);
	i++;

	setpresetname(i, "mudslap (by Styrofoam)");
	setpresetparameter_f(i, kSpeed1, -0.9748);
	setpresetparameter_f(i, kFeed1, 45.6285);
	setpresetparameter_f(i, kDist1, 0.97655);
	setpresetparameter_f(i, kSpeed2, -1.0252);
	setpresetparameter_f(i, kFeed2, 46.4819);
	setpresetparameter_f(i, kDist2, 0.9808);
	setpresetparameter_f(i, kBsize, 122.4947);
	setpresetparameter_f(i, kDrymix, 0.);
	setpresetparameter_f(i, kMix1, 1.);
	setpresetparameter_f(i, kMix2, 1.);
	setpresetparameter_i(i, kQuality, kQualityMode_DirtFi);
	setpresetparameter_b(i, kTomsound, false);
	setpresetparameter_b(i, kFreeze, false);
	i++;

	setpresetname(i, "subverb (by Styrofoam)");
	setpresetparameter_f(i, kSpeed1, -2.9232);
	setpresetparameter_f(i, kFeed1, 49.6802);
	setpresetparameter_f(i, kDist1, 0.94244);
	setpresetparameter_f(i, kSpeed2, -2.90405);
	setpresetparameter_f(i, kFeed2, 47.5475);
	setpresetparameter_f(i, kDist2, 0.968);
	setpresetparameter_f(i, kBsize, 2341.3708);
	setpresetparameter_f(i, kDrymix, 0.);
	setpresetparameter_f(i, kMix1, 1.);
	setpresetparameter_f(i, kMix2, 1.);
	setpresetparameter_i(i, kQuality, kQualityMode_UltraHiFi);
	setpresetparameter_b(i, kTomsound, true);
	setpresetparameter_b(i, kFreeze, false);
	i++;

	setpresetname(i, "vocoder beat (by Styrofoam)");
	setpresetparameter_f(i, kSpeed1, 2.);
	setpresetparameter_f(i, kFeed1, 57.3561);
	setpresetparameter_f(i, kDist1, 0.9915);
	setpresetparameter_f(i, kSpeed2, 1.);
	setpresetparameter_f(i, kFeed2, 27.5053);
	setpresetparameter_f(i, kDist2, 0.9979);
	setpresetparameter_f(i, kBsize, 64.9446);
	setpresetparameter_f(i, kDrymix, 0.);
	setpresetparameter_f(i, kMix1, 1.);
	setpresetparameter_f(i, kMix2, 1.);
	setpresetparameter_i(i, kQuality, kQualityMode_DirtFi);
	setpresetparameter_b(i, kTomsound, false);
	setpresetparameter_b(i, kFreeze, false);
	i++;

	setpresetname(i, "yo pitch! (by Styrofoam)");
	setpresetparameter_f(i, kSpeed1, -2.);
	setpresetparameter_f(i, kFeed1, 80.3769);
	setpresetparameter_f(i, kDist1, 0.977638);
	setpresetparameter_f(i, kSpeed2, 2.);
	setpresetparameter_f(i, kFeed2, 18.937530198);
	setpresetparameter_f(i, kDist2, 0.989335);
	setpresetparameter_f(i, kBsize, 1938.5203);
	setpresetparameter_f(i, kDrymix, 0.);
	setpresetparameter_f(i, kMix1, 1.);
	setpresetparameter_f(i, kMix2, 0.6483);
	setpresetparameter_i(i, kQuality, kQualityMode_UltraHiFi);
	setpresetparameter_b(i, kTomsound, false);
	setpresetparameter_b(i, kFreeze, false);
	i++;

/*
	setpresetname(i, "");
	setpresetparameter_f(i, kSpeed1, );
	setpresetparameter_f(i, kFeed1, );
	setpresetparameter_f(i, kDist1, 0.0);
	setpresetparameter_f(i, kSpeed2, );
	setpresetparameter_f(i, kFeed2, );
	setpresetparameter_f(i, kDist2, 0.0);
	setpresetparameter_f(i, kBsize, );
	setpresetparameter_f(i, kDrymix, 0.);
	setpresetparameter_f(i, kMix1, 0.);
	setpresetparameter_f(i, kMix2, 0.);
	setpresetparameter_i(i, kQuality, );
	setpresetparameter_b(i, kTomsound, );
	setpresetparameter_b(i, kFreeze, false);
	i++;
*/

	// special randomizing "preset"
	setpresetname(getnumpresets() - 1, "random");
}

//-----------------------------------------------------------------------------
bool Transverb::loadpreset(size_t index)
{
	if (!presetisvalid(index))
	{
		return false;
	}

	if (getpresetname(index) == "random")
	{
		randomizeparameters();
		return true;
	}

	return DfxPlugin::loadpreset(index);
}

//--------- presets (end) --------



/* this randomizes the values of all of Transverb's parameters, sometimes in smart ways */
void Transverb::randomizeparameters()
{
	// randomize the non-mix-level parameters

	for (dfx::ParameterID i = 0; i < kDrymix; i++)
	{
		if (hasparameterattribute(i, DfxParam::kAttribute_OmitFromRandomizeAll) || hasparameterattribute(i, DfxParam::kAttribute_Unused))
		{
			continue;
		}
		// make slow speeds more probable (for fairer distribution)
		if (std::find(kSpeedParameters.cbegin(), kSpeedParameters.cend(), i) != kSpeedParameters.cend())
		{
			auto const rand = generateParameterRandomValue<double>(-1., 1.);
			auto const range = (rand < 0.) ? getparametermin_f(i) : getparametermax_f(i); 
			setparameter_f(i, std::fabs(rand) * range);
		}
		// make smaller buffer sizes more probable (because they sound better), though prevent smallest
		else if (i == kBsize)
		{
			setparameter_gen(kBsize, std::pow(generateParameterRandomValue<double>(0.07, 1.), 1.38));
		}
		else
		{
			// TODO: this is not thread-safe (wrt randomizer state) given that Transverb can execute this method
			// from the main thread (GUI) or MIDI thread (though likelihood of concurrency is quite low)
			randomizeparameter(i);
		}
	}


	// do fancy mix level randomization

	// store the current total gain sum
	auto const mixSum = getparameter_f(kDrymix) + getparameter_f(kMix1) + getparameter_f(kMix2);

	// randomize the mix parameters
	auto newDrymix = expandparametervalue(kDrymix, generateParameterRandomValue<double>());
	std::array<double, kNumDelays> newMix {};
	for (size_t head = 0; head < kNumDelays; head++)
	{
		newMix[head] = expandparametervalue(kMixParameters[head], generateParameterRandomValue<double>());
	}
	// calculate a scalar to make up for total gain changes
	auto const mixDiffScalar = mixSum / (newDrymix + std::accumulate(newMix.cbegin(), newMix.cend(), 0.));

	// apply the scalar to the new mix parameter values
	newDrymix *= mixDiffScalar;
	std::transform(newMix.cbegin(), newMix.cend(), newMix.begin(),
				   [mixDiffScalar](double value){ return value * mixDiffScalar; });

	// clip the the delay head mix values at unity gain so that we don't get mega-feedback blasts
	newDrymix = std::clamp(newDrymix, 0., getparametermax_f(kDrymix));
	std::transform(newMix.cbegin(), newMix.cend(), newMix.begin(),
				   [](double value){ return std::clamp(value, 0., 1.); });

	// set the new randomized mix parameter values as the new values
	setparameter_f(kDrymix, newDrymix);
	for (size_t head = 0; head < kNumDelays; head++)
	{
		setparameter_f(kMixParameters[head], newMix[head]);
	}


	// randomize the state parameters

	// make TOMSOUND less probable (only 1/3 of the time)
	auto const newTomsound = static_cast<bool>(generateParameterRandomValue<int64_t>(0, 2) % 2);
	setparameter_b(kTomsound, newTomsound);
	if (newTomsound)
	{
		// split probability evenly between lowest and higher qualities
		setparameter_i(kQuality, generateParameterRandomValue<int64_t>(0, kQualityMode_NumModes) % kQualityMode_NumModes);
	}
	else
	{
		// make higher qualities more probable (happen 4/5 of the time)
		setparameter_i(kQuality, generateParameterRandomValue<int64_t>(1, (kQualityMode_NumModes * 2) - 1) % kQualityMode_NumModes);
	}


	for (dfx::ParameterID i = 0; i < kNumParameters; i++)
	{
		if (hasparameterattribute(i, DfxParam::kAttribute_OmitFromRandomizeAll) || hasparameterattribute(i, DfxParam::kAttribute_Unused))
		{
			continue;
		}
		postupdate_parameter(i);  // inform any parameter listeners of the changes
	}
}

dfx::StatusCode Transverb::dfx_GetPropertyInfo(dfx::PropertyID inPropertyID, dfx::Scope inScope, unsigned int inItemIndex,
                                               size_t& outDataSize, dfx::PropertyFlags& outFlags)
{
  if (isSpeedModePropertyID(inPropertyID))
  {
    outDataSize = sizeof(uint8_t);  // single-byte representation allows for no concerns about transmission endianness
    outFlags = dfx::kPropertyFlag_Readable | dfx::kPropertyFlag_Writable;
    return dfx::kStatus_NoError;
  }
  return DfxPlugin::dfx_GetPropertyInfo(inPropertyID, inScope, inItemIndex, outDataSize, outFlags);
}

dfx::StatusCode Transverb::dfx_GetProperty(dfx::PropertyID inPropertyID, dfx::Scope inScope, unsigned int inItemIndex,
                                           void* outData)
{
  if (isSpeedModePropertyID(inPropertyID))
  {
    *static_cast<uint8_t*>(outData) = static_cast<uint8_t>(speedModeStateFromPropertyID(inPropertyID));
    return dfx::kStatus_NoError;
  }
  return DfxPlugin::dfx_GetProperty(inPropertyID, inScope, inItemIndex, outData);
}

dfx::StatusCode Transverb::dfx_SetProperty(dfx::PropertyID inPropertyID, dfx::Scope inScope, unsigned int inItemIndex,
                                           void const* inData, size_t inDataSize)
{
  if (isSpeedModePropertyID(inPropertyID))
  {
    speedModeStateFromPropertyID(inPropertyID) = *static_cast<uint8_t const*>(inData);
#ifdef TARGET_API_AUDIOUNIT
    PropertyChanged(inPropertyID, inScope, inItemIndex);
#endif
    return dfx::kStatus_NoError;
  }
  return DfxPlugin::dfx_SetProperty(inPropertyID, inScope, inItemIndex, inData, inDataSize);
}


size_t Transverb::settings_sizeOfExtendedData() const noexcept
{
	return speedModeStates.size() * sizeof(speedModeStates.front());
}

void Transverb::settings_saveExtendedData(void* outData, bool /*isPreset*/)
{
  auto speedModeStatesSerialization = speedModeStates;
  if constexpr (!DfxSettings::serializationIsNativeEndian())
  {
    dfx::ReverseBytes(speedModeStatesSerialization.data(), speedModeStatesSerialization.size());
  }
  memcpy(outData, speedModeStatesSerialization.data(), settings_sizeOfExtendedData());
}

void Transverb::settings_restoreExtendedData(void const* inData, size_t storedExtendedDataSize,
                                             unsigned int /*dataVersion*/, bool /*isPreset*/)
{
  if (storedExtendedDataSize >= settings_sizeOfExtendedData())
  {
    memcpy(speedModeStates.data(), inData, settings_sizeOfExtendedData());
    if constexpr (!DfxSettings::serializationIsNativeEndian())
    {
      dfx::ReverseBytes(speedModeStates.data(), speedModeStates.size());
    }
  }
}

void Transverb::settings_doChunkRestoreSetParameterStuff(dfx::ParameterID parameterID, float value, unsigned int dataVersion, std::optional<size_t> presetIndex)
{
	// prevent old speed mode 1 settings from applying to freeze
	if ((dataVersion < 0x00010501) && (parameterID == kFreeze))
	{
		auto const defaultValue = getparameterdefault_b(parameterID);
		if (presetIndex)
		{
			setpresetparameter_b(*presetIndex, parameterID, defaultValue);
		}
		else
		{
			setparameter_b(parameterID, defaultValue);
		}
	}

	if (dataVersion <= 0x00010502)
	{
		// invert old erroneously reversed dist parameter values
		if (std::find(kDistParameters.cbegin(), kDistParameters.cend(), parameterID) != kDistParameters.cend())
		{
			auto const valueNormalized = 1. - contractparametervalue(parameterID, value);
			if (presetIndex)
			{
				setpresetparameter_gen(*presetIndex, parameterID, valueNormalized);
			}
			else
			{
				setparameter_gen(parameterID, valueNormalized);
			}
		}

		// hi-fi mode used to behave the same as and ultra hi-fi mode in TOMSOUND
		if (parameterID == kTomsound)  // assumes that TOMSOUND parameter is restored after quality
		{
			if (presetIndex)
			{
				if (getpresetparameter_f(*presetIndex, kTomsound) && (getpresetparameter_i(*presetIndex, kQuality) == kQualityMode_HiFi))
				{
					setpresetparameter_i(*presetIndex, kQuality, kQualityMode_UltraHiFi);
				}
			}
			else
			{
				if (getparameter_b(kTomsound) && (getparameter_i(kQuality) == kQualityMode_HiFi))
				{
					setparameter_i(kQuality, kQualityMode_UltraHiFi);
				}
			}
		}

		// apply legacy feedback behavior
		// https://github.com/sophiapoirier/destroyfx/issues/61
		constexpr bool attenuate = true;
		if (presetIndex)
		{
			setpresetparameter_b(*presetIndex, kAttenuateFeedbackByMixLevel, attenuate);
		}
		else
		{
			setparameter_b(kAttenuateFeedbackByMixLevel, attenuate);
		}
	}
}



#pragma mark -

dfx::PropertyID dfx::TV::speedModeIndexToPropertyID(size_t inIndex) noexcept
{
  return kTransverbProperty_SpeedModeBase + static_cast<dfx::PropertyID>(inIndex);
}

size_t dfx::TV::speedModePropertyIDToIndex(dfx::PropertyID inPropertyID) noexcept
{
  assert(inPropertyID >= kTransverbProperty_SpeedModeBase);
  return inPropertyID - kTransverbProperty_SpeedModeBase;
}

bool dfx::TV::isSpeedModePropertyID(dfx::PropertyID inPropertyID) noexcept
{
  return (inPropertyID >= kTransverbProperty_SpeedModeBase) && (speedModePropertyIDToIndex(inPropertyID) < kNumDelays);
}