Bug fixed for unix error "readlink /proc/self/fd/0" on MacOS.

[Faustine.git] / interpreter / preprocessor / faust-0.9.47mr3 / architecture / effect.lib

declare name "Faust Audio Effect Library";
declare author "Julius O. Smith (jos at ccrma.stanford.edu)";
declare copyright "Julius O. Smith III";
declare version "1.33";
declare license "STK-4.3"; // Synthesis Tool Kit 4.3 (MIT style license)
declare reference "https://ccrma.stanford.edu/realsimple/faust_strings/";

import("filter.lib"); // dcblocker*, lowpass, filterbank, ...

// The following utilities (or equivalents) could go in music.lib:

//----------------------- midikey2hz,pianokey2hz ------------------------
midikey2hz(x) = 440.0*pow(2.0, (x-69.0)/12); // MIDI key 69 = A440
pianokey2hz(x) = 440.0*pow(2.0, (x-49.0)/12); // piano key 49 = A440

//---------------- cross2, bypass1, bypass2, select2stereo --------------
//
cross2 = _,_,_,_ <: _,!,_,!,!,_,!,_;

bypass1(bpc,e) = _ <: select2(bpc,(inswitch:e),_)
                 with {inswitch = select2(bpc,_,0);};

bypass2(bpc,e) = _,_ <: ((inswitch:e),_,_) : select2stereo(bpc) with {
  inswitch = _,_ : (select2(bpc,_,0), select2(bpc,_,0)) : _,_;
};

select2stereo(bpc) = cross2 : select2(bpc), select2(bpc) : _,_;

//---------------------- levelfilter, levelfilterN ----------------------
// Dynamic level lowpass filter:
//
// USAGE: levelfilter(L,freq), where
//  L    = desired level (in dB) at Nyquist limit (SR/2), e.g., -60
//  freq = corner frequency (-3dB point) usually set to fundamental freq
//
// REFERENCE:
// https://ccrma.stanford.edu/realsimple/faust_strings/Dynamic_Level_Lowpass_Filter.html
//
levelfilter(L,freq,x) = (L * L0 * x) + ((1.0-L) * lp2out(x))
with {
  L0 = pow(L,1/3);
  Lw = PI*freq/SR; // = w1 T / 2
  Lgain = Lw / (1.0 + Lw);
  Lpole2 = (1.0 - Lw) / (1.0 + Lw);
  lp2out = *(Lgain) : + ~ *(Lpole2);
};

levelfilterN(N,freq,L) = seq(i,N,levelfilter((L/N),freq));

//------------------------- speakerbp -------------------------------
// Dirt-simple speaker simulator (overall bandpass eq with observed
// roll-offs above and below the passband).
//
// Low-frequency speaker model = +12 dB/octave slope breaking to
// flat near f1. Implemented using two dc blockers in series.
//
// High-frequency model = -24 dB/octave slope implemented using a
// fourth-order Butterworth lowpass.
//
// Example based on measured Celestion G12 (12" speaker):
// speakerbp(130,5000);
//
// Requires filter.lib
//
speakerbp(f1,f2) = dcblockerat(f1) : dcblockerat(f1) : lowpass(4,f2);


//--------------------- cubicnl(drive,offset) -----------------------
// Cubic nonlinearity distortion
//
// USAGE: cubicnl(drive,offset), where
//   drive  = distortion amount, between 0 and 1
//   offset = constant added before nonlinearity to give even harmonics
//            Note: offset can introduce a nonzero mean - feed
//            cubicnl output to dcblocker to remove this.
//
// REFERENCES:
//   https://ccrma.stanford.edu/~jos/pasp/Cubic_Soft_Clipper.html
//   https://ccrma.stanford.edu/~jos/pasp/Nonlinear_Distortion.html
//
cubicnl(drive,offset) = *(pregain) : +(offset) : clip(-1,1) : cubic
with {
    pregain = pow(10.0,2*drive);
    clip(lo,hi) = min(hi) : max(lo);
    cubic(x) = x - x*x*x/3;
    postgain = max(1.0,1.0/pregain); // unity gain when nearly linear
};

cubicnl_nodc(drive,offset) = cubicnl(drive,offset) : dcblocker;

//--------------------------- cubicnl_demo --------------------------
// USAGE: _ : cubicnl_demo : _;
//
cubicnl_demo = bypass1(bp,
   cubicnl_nodc(drive:smooth(0.999),offset:smooth(0.999)))
with {
   cnl_group(x)  = vgroup("CUBIC NONLINEARITY cubicnl
        [tooltip: Reference:
         https://ccrma.stanford.edu/~jos/pasp/Cubic_Soft_Clipper.html]", x);
//   bypass_group(x) = cnl_group(hgroup("[0]", x));
   slider_group(x)  = cnl_group(hgroup("[1]", x));
//   bp = bypass_group(checkbox("[0] Bypass
   bp = slider_group(checkbox("[0] Bypass
        [tooltip: When this is checked, the nonlinearity has no effect]"));
//   drive = slider_group(vslider("[1] Drive [style: knob]
   drive = slider_group(hslider("[1] Drive
                       [tooltip: Amount of distortion]",
                       0, 0, 1, 0.01));
//   offset = slider_group(vslider("[2] Offset [style: knob]
   offset = slider_group(hslider("[2] Offset
                       [tooltip: Brings in even harmonics]",
                       0, 0, 1, 0.01));
};

//------------------------- moog_vcf(res,fr) ---------------------------
// Moog "Voltage Controlled Filter" (VCF) in "analog" form
//
// USAGE: moog_vcf(res,fr), where
//   fr = corner-resonance frequency in Hz ( less than SR/6.3 or so )
//   res  = Normalized amount of corner-resonance between 0 and 1
//        (0 is no resonance, 1 is maximum)
// Requires filter.lib.
//
// DESCRIPTION: Moog VCF implemented using the same logical block diagram
//   as the classic analog circuit.  As such, it neglects the one-sample
//   delay associated with the feedback path around the four one-poles.
//   This extra delay alters the response, especially at high frequencies
//   (see reference [1] for details).
//   See moog_vcf_2b below for a more accurate implementation.
//
// REFERENCES:
//   [1] https://ccrma.stanford.edu/~stilti/papers/moogvcf.pdf
//   [2] https://ccrma.stanford.edu/~jos/pasp/vegf.html
//
moog_vcf(res,fr) = (+ : seq(i,4,pole(p)) : *(unitygain(p))) ~ *(mk)
with {
     p = 1.0 - fr * 2.0 * PI / SR; // good approximation for fr << SR
     unitygain(p) = pow(1.0-p,4.0); // one-pole unity-gain scaling
     mk = -4.0*max(0,min(res,0.999999)); // need mk > -4 for stability
};

//----------------------- moog_vcf_2b[n] ---------------------------
// Moog "Voltage Controlled Filter" (VCF) as two biquads
//
// USAGE:
//   moog_vcf_2b(res,fr)
//   moog_vcf_2bn(res,fr)
// where
//   fr = corner-resonance frequency in Hz
//   res  = Normalized amount of corner-resonance between 0 and 1
//        (0 is min resonance, 1 is maximum)
//
// DESCRIPTION: Implementation of the ideal Moog VCF transfer
//   function factored into second-order sections.  As a result, it is
//   more accurate than moog_vcf above, but its coefficient formulas are
//   more complex when one or both parameters are varied.  Here, res
//   is the fourth root of that in moog_vcf, so, as the sampling rate
//   approaches infinity, moog_vcf(res,fr) becomes equivalent
//   to moog_vcf_2b[n](res^4,fr) (when res and fr are constant).
//
//   moog_vcf_2b  uses two direct-form biquads (tf2)
//   moog_vcf_2bn uses two protected normalized-ladder biquads (tf2np)
//
// REQUIRES: filter.lib
//
moog_vcf_2b(res,fr) = tf2s(0,0,b0,a11,a01,w1) : tf2s(0,0,b0,a12,a02,w1)
with {
 s = 1; // minus the open-loop location of all four poles
 frl = max(20,min(10000,fr)); // limit fr to reasonable 20-10k Hz range
 w1 = 2*PI*frl; // frequency-scaling parameter for bilinear xform
 // Equivalent: w1 = 1; s = 2*PI*frl;
 kmax = sqrt(2)*0.999; // 0.999 gives stability margin (tf2 is unprotected)
 k = min(kmax,sqrt(2)*res); // fourth root of Moog VCF feedback gain
 b0 = s^2;
 s2k = sqrt(2) * k;
 a11 = s * (2 + s2k);
 a12 = s * (2 - s2k);
 a01 = b0 * (1 + s2k + k^2);
 a02 = b0 * (1 - s2k + k^2);
};

moog_vcf_2bn(res,fr) = tf2snp(0,0,b0,a11,a01,w1) : tf2snp(0,0,b0,a12,a02,w1)
with {
 s = 1; // minus the open-loop location of all four poles
 w1 = 2*PI*max(fr,20); // frequency-scaling parameter for bilinear xform
 k = sqrt(2)*0.999*res; // fourth root of Moog VCF feedback gain
 b0 = s^2;
 s2k = sqrt(2) * k;
 a11 = s * (2 + s2k);
 a12 = s * (2 - s2k);
 a01 = b0 * (1 + s2k + k^2);
 a02 = b0 * (1 - s2k + k^2);
};

//------------------------- moog_vcf_demo ---------------------------
// Illustrate and compare all three Moog VCF implementations above
// (called by <faust>/examples/vcf_wah_pedals.dsp).
//
// USAGE: _ : moog_vcf_demo : _;

moog_vcf_demo = bypass1(bp,vcf) with {
   mvcf_group(x)  = hgroup("MOOG VCF (Voltage Controlled Filter)
      [tooltip: See Faust's effect.lib for info and references]",x);

   meter_group(x) = mvcf_group(vgroup("[0]",x));
   cb_group(x) = meter_group(hgroup("[0]",x));

   bp = cb_group(checkbox("[0] Bypass  [tooltip: When this is checked, the Moog VCF has no effect]"));
   archsw = cb_group(checkbox("[1] Use Biquads
   [tooltip: Select moog_vcf_2b (two-biquad) implementation, instead of the default moog_vcf (analog style) implementation]"));
   bqsw = cb_group(checkbox("[2] Normalized Ladders
   [tooltip: If using biquads, make them normalized ladders (moog_vcf_2bn)]"));

   freq = mvcf_group(hslider("[1] Corner Frequency [unit:PK] [style:knob]
   [tooltip: The VCF resonates at the corner frequency (specified in PianoKey (PK) units, with A440 = 49 PK).  The VCF response is flat below the corner frequency, and rolls off -24 dB per octave above.]",
   25, 1, 88, 0.01) : pianokey2hz) : smooth(0.999);

   res = mvcf_group(hslider("[2] Corner Resonance [style:knob]
   [tooltip: Amount of resonance near VCF corner frequency (specified between 0 and 1)]",
   0.9, 0, 1, 0.01));

   outgain  = meter_group(hslider("[1] VCF Output Level [unit:dB]
   [tooltip: output level in decibels]",
   5, -60, 20, 0.1)) : smooth(0.999)
   : component("music.lib").db2linear;

   vcfbq =  _ <: select2(bqsw, moog_vcf_2b(res,freq), moog_vcf_2bn(res,freq));
   vcfarch =  _ <: select2(archsw, moog_vcf(res^4,freq), vcfbq);
   vcf = vcfarch : *(outgain);
};

//-------------------------- wah4(fr) -------------------------------
// Wah effect, 4th order
// USAGE: wah4(fr), where fr = resonance frequency in Hz
// REFERENCE "https://ccrma.stanford.edu/~jos/pasp/vegf.html";
//
wah4(fr) = 4*moog_vcf((3.2/4),fr:smooth(0.999));

//------------------------- wah4_demo ---------------------------
// USAGE: _ : wah4_demo : _;

wah4_demo = bypass1(bp, wah4(fr)) with {
  wah4_group(x)  = hgroup("WAH4
       [tooltip: Fourth-order wah effect made using moog_vcf]", x);
  bp = wah4_group(checkbox("[0] Bypass
       [tooltip: When this is checked, the wah pedal has no effect]"));
  fr = wah4_group(hslider("[1] Resonance Frequency
       [tooltip: wah resonance frequency in Hz]",
     200,100,2000,1));
// Avoid dc with the moog_vcf (amplitude too high when freq comes up from dc)
// Also, avoid very high resonance frequencies (e.g., 5kHz or above).
};

//------------------------ autowah(level) -----------------------------
// Auto-wah effect
// USAGE: _ : autowah(level) : _;
// where level = amount of effect desired (0 to 1).
//
autowah(level,x) = level * crybaby(amp_follower(0.1,x),x) + (1.0-level)*x;

//-------------------------- crybaby(wah) -----------------------------
// Digitized CryBaby wah pedal
// USAGE: _ : crybaby(wah) : _;
// where wah = "pedal angle" from 0 to 1.
// REFERENCE: https://ccrma.stanford.edu/~jos/pasp/vegf.html
//
crybaby(wah) = *(gs) : tf2(1,-1,0,a1s,a2s)
with {
  Q  = pow(2.0,(2.0*(1.0-wah)+1.0)); // Resonance "quality factor"
  fr = 450.0*pow(2.0,2.3*wah);       // Resonance tuning
  g  = 0.1*pow(4.0,wah);             // gain (optional)

  // Biquad fit using z = exp(s T) ~ 1 + sT for low frequencies:
  frn = fr/SR; // Normalized pole frequency (cycles per sample)
  R = 1 - PI*frn/Q; // pole radius
  theta = 2*PI*frn; // pole angle
  a1 = 0-2.0*R*cos(theta); // biquad coeff
  a2 = R*R;                // biquad coeff

  // dezippering of slider-driven signals:
  s = 0.999; // smoothing parameter (one-pole pole location)
  a1s = a1 : smooth(s);
  a2s = a2 : smooth(s);
  gs =  g  : smooth(s);

  tf2 = component("filter.lib").tf2;
};

//------------------------- crybaby_demo ---------------------------
// USAGE: _ : crybaby_demo : _ ;

crybaby_demo = bypass1(bp, crybaby(wah)) with {
   crybaby_group(x)  = hgroup("CRYBABY [tooltip: Reference: https://ccrma.stanford.edu/~jos/pasp/vegf.html]", x);
   bp = crybaby_group(checkbox("[0] Bypass [tooltip: When this is checked, the wah pedal has no effect]"));
   wah = crybaby_group(hslider("[1] Wah parameter [tooltip: wah pedal angle between 0 (rocked back) and 1 (rocked forward)]",0.8,0,1,0.01));
};

//------------ apnl(a1,a2) ---------------
// Passive Nonlinear Allpass:
// switch between allpass coefficient a1 and a2 at signal zero crossings
// REFERENCE:
//  "A Passive Nonlinear Digital Filter Design ..."
//  by John R. Pierce and Scott A. Van Duyne,
//  JASA, vol. 101, no. 2, pp. 1120-1126, 1997
// Written by Romain Michon and JOS based on Pierce switching springs idea:
  apnl(a1,a2,x) = nonLinFilter
  with{
   condition = _>0;
   nonLinFilter = (x - _ <: _*(condition*a1 + (1-condition)*a2),_')~_ :> +;
  };

//------------ piano_dispersion_filter(M,B,f0) ---------------
// Piano dispersion allpass filter in closed form
//
// ARGUMENTS:
//   M = number of first-order allpass sections (compile-time only)
//       Keep below 20. 8 is typical for medium-sized piano strings.
//   B = string inharmonicity coefficient (0.0001 is typical)
//  f0 = fundamental frequency in Hz
//
// INPUT:
//   Signal to be filtered by the allpass chain
//
// OUTPUTS:
//  1. MINUS the estimated delay at f0 of allpass chain in samples,
//     provided in negative form to facilitate subtraction
//     from delay-line length (see USAGE below).
//  2. Output signal from allpass chain
//
// USAGE:
//  piano_dispersion_filter(1,B,f0) : +(totalDelay),_ : fdelay(maxDelay)
//
// REFERENCE:
//   "Dispersion Modeling in Waveguide Piano Synthesis
//    Using Tunable Allpass Filters",
//   by Jukka Rauhala and Vesa Valimaki, DAFX-2006, pp. 71-76
//   URL: http://www.dafx.ca/proceedings/papers/p_071.pdf
//   NOTE: An erratum in Eq. (7) is corrected in Dr. Rauhala's
//    encompassing dissertation (and below).
//   See also: http://www.acoustics.hut.fi/research/asp/piano/
//
piano_dispersion_filter(M,B,f0) = -Df0*M,seq(i,M,tf1(a1,1,a1))
with {
 a1 = (1-D)/(1+D); // By Eq. 3, have D >= 0, hence a1 >= 0 also
 D = exp(Cd - Ikey(f0)*kd);
 trt = pow(2.0,1.0/12.0); // 12th root of 2
 logb(b,x) = log(x) / log(b); // log-base-b of x
 Ikey(f0) = logb(trt,f0*trt/27.5);
 Bc = max(B,0.000001);
 kd = exp(k1*log(Bc)*log(Bc) + k2*log(Bc)+k3);
 Cd = exp((m1*log(M)+m2)*log(Bc)+m3*log(M)+m4);
 k1 = -0.00179;
 k2 = -0.0233;
 k3 = -2.93;
 m1 = 0.0126;
 m2 = 0.0606;
 m3 = -0.00825;
 m4 = 1.97;
 wT = 2*PI*f0/SR;
 polydel(a) = atan(sin(wT)/(a+cos(wT)))/wT;
 Df0 = polydel(a1) - polydel(1.0/a1);
};

//===================== Phasing and Flanging Effects  ====================

//--------------- flanger_mono, flanger_stereo, flanger_demo -------------
// Flanging effect
//
// USAGE:
//     _ : flanger_mono(dmax,curdel,depth,fb,invert) : _;
//   _,_ : flanger_stereo(dmax,curdel1,curdel2,depth,fb,invert) : _,_;
//   _,_ : flanger_demo : _,_;
//
// ARGUMENTS:
//   dmax   = maximum delay-line length (power of 2) - 10 ms typical
//   curdel = current dynamic delay (not to exceed dmax)
//   depth  = effect strength between 0 and 1 (1 typical)
//   fb     = feedback gain between 0 and 1 (0 typical)
//   invert = 0 for normal, 1 to invert sign of flanging sum
//
// REFERENCE:
//    https://ccrma.stanford.edu/~jos/pasp/Flanging.html
//
flanger_mono(dmax,curdel,depth,fb,invert)
  = _ <: _, (-:fdelay(dmax,curdel)) ~ *(fb) : _,
  *(select2(invert,depth,0-depth))
  : + : *(0.5);

flanger_stereo(dmax,curdel1,curdel2,depth,fb,invert)
  =  flanger_mono(dmax,curdel1,depth,fb,invert),
     flanger_mono(dmax,curdel2,depth,fb,invert);

//------------------------- flanger_demo ---------------------------
// USAGE:  _,_ : flanger_demo : _,_;
//
flanger_demo = bypass2(fbp,flanger_stereo_demo) with {
   flanger_group(x) =
    vgroup("FLANGER [tooltip: Reference: https://ccrma.stanford.edu/~jos/pasp/Flanging.html]", x);
   meter_group(x) = flanger_group(hgroup("[0]", x));
   ctl_group(x)  = flanger_group(hgroup("[1]", x));
   del_group(x)  = flanger_group(hgroup("[2] Delay Controls", x));
   lvl_group(x)  = flanger_group(hgroup("[3]", x));

   fbp = meter_group(checkbox(
         "[0] Bypass  [tooltip: When this is checked, the flanger has no effect]"));
   invert = meter_group(checkbox("[1] Invert Flange Sum"));

   // FIXME: This should be an amplitude-response display:
   flangeview = lfor(freq) + lfol(freq) : meter_group(hbargraph(
      "[2] Flange LFO [style: led] [tooltip: Display sum of flange delays]", -1.5,+1.5));

   flanger_stereo_demo(x,y) = attach(x,flangeview),y :
     *(level),*(level) : flanger_stereo(dmax,curdel1,curdel2,depth,fb,invert);

   lfol = component("oscillator.lib").oscrs; // sine for left channel
   lfor = component("oscillator.lib").oscrc; // cosine for right channel
   dmax = 2048;
   dflange = 0.001 * SR *
     del_group(hslider("[1] Flange Delay [unit:ms] [style:knob]", 10, 0, 20, 0.001));
   odflange = 0.001 * SR *
     del_group(hslider("[2] Delay Offset [unit:ms] [style:knob]", 1, 0, 20, 0.001));
   freq   = ctl_group(hslider("[1] Speed [unit:Hz] [style:knob]", 0.5, 0, 10, 0.01));
   depth  = ctl_group(hslider("[2] Depth [style:knob]", 1, 0, 1, 0.001));
   fb     = ctl_group(hslider("[3] Feedback [style:knob]", 0, -0.999, 0.999, 0.001));
   level  = lvl_group(hslider("Flanger Output Level [unit:dB]", 0, -60, 10, 0.1)) : db2linear;
   curdel1 = odflange+dflange*(1 + lfol(freq))/2;
   curdel2 = odflange+dflange*(1 + lfor(freq))/2;
};

//------- phaser2_mono, phaser2_stereo, phaser2_demo -------
// Phasing effect
//
// USAGE:
//   _ : phaser2_mono(Notches,width,frqmin,fratio,frqmax,speed,depth,fb,invert) : _;
// _,_ : phaser2_stereo(") : _,_;
// _,_ : phaser2_demo : _,_;
//
// ARGUMENTS:
//   Notches = number of spectral notches (MACRO ARGUMENT - not a signal)
//   width  = approximate width of spectral notches in Hz
//   frqmin = approximate minimum frequency of first spectral notch in Hz
//   fratio = ratio of adjacent notch frequencies
//   frqmax = approximate maximum frequency of first spectral notch in Hz
//   speed  = LFO frequency in Hz (rate of periodic notch sweep cycles)
//   depth  = effect strength between 0 and 1 (1 typical) (aka "intensity")
//            when depth=2, "vibrato mode" is obtained (pure allpass chain)
//   fb     = feedback gain between -1 and 1 (0 typical)
//   invert = 0 for normal, 1 to invert sign of flanging sum
//
// REFERENCES:
//    https://ccrma.stanford.edu/~jos/pasp/Phasing.html
//    http://www.geofex.com/Article_Folders/phasers/phase.html
//    'An Allpass Approach to Digital Phasing and Flanging', Julius O. Smith III,
//    Proc. Int. Computer Music Conf. (ICMC-84), pp. 103-109, Paris, 1984.
//    CCRMA Tech. Report STAN-M-21: https://ccrma.stanford.edu/STANM/stanms/stanm21/

vibrato2_mono(sections,phase01,fb,width,frqmin,fratio,frqmax,speed) =
 (+ : seq(i,sections,ap2p(R,th(i)))) ~ *(fb)
with {
     tf2 = component("filter.lib").tf2;
     // second-order resonant digital allpass given pole radius and angle:
     ap2p(R,th) = tf2(a2,a1,1,a1,a2) with {
       a2 = R^2;
       a1 = -2*R*cos(th);
     };
     SR = component("music.lib").SR;
     R = exp(-pi*width/SR);
     cososc = component("oscillator.lib").oscrc;
     sinosc = component("oscillator.lib").oscrs;
     osc = cososc(speed) * phase01 + sinosc(speed) * (1-phase01);
     lfo = (1-osc)/2; // in [0,1]
     pi = 4*atan(1);
     thmin = 2*pi*frqmin/SR;
     thmax = 2*pi*frqmax/SR;
     th1 = thmin + (thmax-thmin)*lfo;
     th(i) = (fratio^(i+1))*th1;
};

phaser2_mono(Notches,phase01,width,frqmin,fratio,frqmax,speed,depth,fb,invert) =
      _ <: *(g1) + g2mi*vibrato2_mono(Notches,phase01,fb,width,frqmin,fratio,frqmax,speed)
with {               // depth=0 => direct-signal only
     g1 = 1-depth/2; // depth=1 => phaser mode (equal sum of direct and allpass-chain)
     g2 = depth/2;   // depth=2 => vibrato mode (allpass-chain signal only)
     g2mi = select2(invert,g2,-g2); // inversion negates the allpass-chain signal
};

phaser2_stereo(Notches,width,frqmin,fratio,frqmax,speed,depth,fb,invert)
   = phaser2_mono(Notches,0,width,frqmin,fratio,frqmax,speed,depth,fb,invert),
     phaser2_mono(Notches,1,width,frqmin,fratio,frqmax,speed,depth,fb,invert);

//------------------------- phaser2_demo ---------------------------
// USAGE:  _,_ : phaser2_demo : _,_;
//
phaser2_demo = bypass2(pbp,phaser2_stereo_demo) with {
   phaser2_group(x) =
    vgroup("PHASER2 [tooltip: Reference: https://ccrma.stanford.edu/~jos/pasp/Flanging.html]", x);
   meter_group(x) = phaser2_group(hgroup("[0]", x));
   ctl_group(x)  = phaser2_group(hgroup("[1]", x));
   nch_group(x)  = phaser2_group(hgroup("[2]", x));
   lvl_group(x)  = phaser2_group(hgroup("[3]", x));

   pbp = meter_group(checkbox(
         "[0] Bypass  [tooltip: When this is checked, the phaser has no effect]"));
   invert = meter_group(checkbox("[1] Invert Internal Phaser Sum"));
   vibr   = meter_group(checkbox("[2] Vibrato Mode")); // In this mode you can hear any "Doppler"

   // FIXME: This should be an amplitude-response display:
   //flangeview = phaser2_amp_resp : meter_group(hspectrumview("[2] Phaser Amplitude Response", 0,1));
   //phaser2_stereo_demo(x,y) = attach(x,flangeview),y : ...

   phaser2_stereo_demo = *(level),*(level) :
     phaser2_stereo(Notches,width,frqmin,fratio,frqmax,speed,mdepth,fb,invert);

   Notches = 4; // Compile-time parameter: 2 is typical for analog phaser stomp-boxes

   // FIXME: Add tooltips
   speed  = ctl_group(hslider("[1] Speed [unit:Hz] [style:knob]", 0.5, 0, 10, 0.001));
   depth  = ctl_group(hslider("[2] Notch Depth (Intensity) [style:knob]", 1, 0, 1, 0.001));
   fb     = ctl_group(hslider("[3] Feedback Gain [style:knob]", 0, -0.999, 0.999, 0.001));

   width  = nch_group(hslider("[1] Notch width [unit:Hz] [style:knob]", 1000, 10, 5000, 1));
   frqmin = nch_group(hslider("[2] Min Notch1 Freq [unit:Hz] [style:knob]", 100, 20, 5000, 1));
   frqmax = nch_group(hslider("[3] Max Notch1 Freq [unit:Hz] [style:knob]", 800, 20, 10000, 1)) : max(frqmin);
   fratio = nch_group(hslider("[4] Notch Freq Ratio: NotchFreq(n+1)/NotchFreq(n)  [style:knob]", 1.5, 1.1, 4, 0.001));

   level  = lvl_group(hslider("Phaser Output Level [unit:dB]", 0, -60, 10, 0.1)) : component("music.lib").db2linear;

   mdepth = select2(vibr,depth,2); // Improve "ease of use"
};

//------------------------- stereo_width(w) ---------------------------
// Stereo Width effect using the Blumlein Shuffler technique.
//
// USAGE: "_,_ : stereo_width(w) : _,_", where
//   w = stereo width between 0 and 1
//
// At w=0, the output signal is mono ((left+right)/2 in both channels).
// At w=1, there is no effect (original stereo image).
// Thus, w between 0 and 1 varies stereo width from 0 to "original".
//
// REFERENCE:
// "Applications of Blumlein Shuffling to Stereo Microphone Techniques"
// Michael A. Gerzon, JAES vol. 42, no. 6, June 1994
//
stereo_width(w) = shuffle : *(mgain),*(sgain) : shuffle
with {
     shuffle =  _,_ <: +,-; // normally scaled by 1/sqrt(2) for orthonormality,
     mgain = 1-w/2;  // but we pick up the needed normalization here.
     sgain = w/2;
};

//--------------------------- amp_follower ---------------------------
// Classic analog audio envelope follower with infinitely fast rise and
// exponential decay.  The amplitude envelope instantaneously follows
// the absolute value going up, but then floats down exponentially.
//
// USAGE:
//    _ : amp_follower(rel) : _
//
// where
//  rel = release time = amplitude-envelope time-constant (sec) going down
//
// REFERENCES:
//  Musical Engineer's Handbook, Bernie Hutchins, Ithaca NY, 1975
//  Elecronotes Newsletter, Bernie Hutchins

amp_follower(rel) = abs : env with {
 p = tau2pole(rel);
 env(x) = x * (1.0 - p) : + ~ max(x,_) * p;
};

//--------------------------- amp_follower_ud ---------------------------
// Envelope follower with different up and down time-constants
//
// USAGE:
//    _ : amp_follower_ud(att,rel) : _
//
// where
//  att = attack time = amplitude-envelope time constant (sec) going up
//  rel = release time = amplitude-envelope time constant (sec) going down
//
// For audio, att should be faster (smaller) than rel (e.g., 0.001 and 0.01)

amp_follower_ud(att,rel) = amp_follower(rel) : smooth(tau2pole(att));

//=============== Gates, Limiters, and Dynamic Range Compression ============

//----------------- gate_mono, gate_stereo -------------------
// Mono and stereo signal gates
//
// USAGE:
//    _ : gate_mono(thresh,att,hold,rel)   : _
// or
//  _,_ : gate_stereo(thresh,att,hold,rel) : _,_
//
// where
//  thresh = dB level threshold above which gate opens (e.g., -60 dB)
//  att    = attack time = time constant (sec) for gate to open (e.g., 0.0001 s = 0.1 ms)
//  hold   = hold time = time (sec) gate stays open after signal level < thresh (e.g., 0.1 s)
//  rel    = release time = time constant (sec) for gate to close (e.g., 0.020 s = 20 ms)
//
// REFERENCES:
// - http://en.wikipedia.org/wiki/Noise_gate
// - http://www.soundonsound.com/sos/apr01/articles/advanced.asp
// - http://en.wikipedia.org/wiki/Gating_(sound_engineering)

gate_mono(thresh,att,hold,rel,x) = x * gate_gain_mono(thresh,att,hold,rel,x);

gate_stereo(thresh,att,hold,rel,x,y) = ggm*x, ggm*y with {
  ggm = gate_gain_mono(thresh,att,hold,rel,abs(x)+abs(y));
};

gate_gain_mono(thresh,att,hold,rel,x) = extendedrawgate : amp_follower_ud(att,rel) with {
  extendedrawgate = max(rawgatesig,holdsig);
  rawgatesig = inlevel(x) > db2linear(thresh);
  inlevel(x) = amp_follower_ud(att/2,rel/2,x);
  holdsig = ((max(holdreset & holdsamps,_) ~-(1)) > 0);
  holdreset = rawgatesig > rawgatesig'; // reset hold when raw gate falls
  holdsamps = int(hold*SR);
};

//-------------------- compressor_mono, compressor_stereo ----------------------
// Mono and stereo dynamic range compressor_s
//
// USAGE:
//    _ : compressor_mono(ratio,thresh,att,rel)   : _
// or
//  _,_ : compressor_stereo(ratio,thresh,att,rel) : _,_
//
// where
//  ratio  = compression ratio (1 = no compression, >1 means compression)
//  thresh = dB level threshold above which compression kicks in
//  att    = attack time = time constant (sec) when level & compression going up
//  rel    = release time = time constant (sec) coming out of compression
//
// REFERENCES:
// - http://en.wikipedia.org/wiki/Dynamic_range_compression
// - https://ccrma.stanford.edu/~jos/filters/Nonlinear_Filter_Example_Dynamic.html
// - Albert Graef's <faust2pd>/examples/synth/compressor_.dsp
//

compressor_mono(ratio,thresh,att,rel,x) = x * compression_gain_mono(ratio,thresh,att,rel,x);

compressor_stereo(ratio,thresh,att,rel,x,y) = cgm*x, cgm*y with {
  cgm = compression_gain_mono(ratio,thresh,att,rel,abs(x)+abs(y));
};

compression_gain_mono(ratio,thresh,att,rel) =
  amp_follower_ud(att,rel) : linear2db : outminusindb(ratio,thresh) :
  kneesmooth(att) : db2linear
with {
  // kneesmooth(att) installs a "knee" in the dynamic-range compression,
  // where knee smoothness is set equal to half that of the compression-attack.
  // A general 'knee' parameter could be used instead of tying it to att/2:
  kneesmooth(att)  = smooth(tau2pole(att/2.0));
  // compression gain in dB:
   outminusindb(ratio,thresh,level) = max(level-thresh,0) * (1/float(ratio)-1);
  // Note: "float(ratio)" REQUIRED when ratio is an integer > 1!
};

//---------------------------- gate_demo -------------------------
// USAGE: _,_ : gate_demo : _,_;
//
gate_demo = bypass2(gbp,gate_stereo_demo) with {

   gate_group(x)  = vgroup("GATE  [tooltip: Reference: http://en.wikipedia.org/wiki/Noise_gate]", x);
   meter_group(x) = gate_group(hgroup("[0]", x));
   knob_group(x)  = gate_group(hgroup("[1]", x));

   gbp = meter_group(checkbox("[0] Bypass  [tooltip: When this is checked, the gate has no effect]"));

   gateview = gate_gain_mono(gatethr,gateatt,gatehold,gaterel) : linear2db :
     meter_group(hbargraph("[1] Gate Gain [unit:dB]  [tooltip: Current gain of the gate in dB]",
      -50,+10)); // [style:led]

   gate_stereo_demo(x,y) = attach(x,gateview(abs(x)+abs(y))),y :
     gate_stereo(gatethr,gateatt,gatehold,gaterel);

   gatethr = knob_group(hslider("[1] Threshold [unit:dB] [style:knob]  [tooltip: When the signal level falls below the Threshold (expressed in dB), the signal is muted]",
     -30, -120, 0, 0.1));

   gateatt = knob_group(hslider("[2] Attack [unit:us] [style:knob]  [tooltip: Time constant in MICROseconds (1/e smoothing time) for the gate gain to go (exponentially) from 0 (muted) to 1 (unmuted)]",
     10, 10, 10000, 1)) : *(0.000001) : max(1/SR);

   gatehold = knob_group(hslider("[3] Hold [unit:ms] [style:knob]  [tooltip: Time in ms to keep the gate open (no muting) after the signal level falls below the Threshold]",
     200, 0, 1000, 1)) : *(0.001) : max(1/SR);

   gaterel = knob_group(hslider("[4] Release [unit:ms] [style:knob]  [tooltip: Time constant in ms (1/e smoothing time) for the gain to go (exponentially) from 1 (unmuted) to 0 (muted)]",
     100, 0, 1000, 1)) : *(0.001) : max(1/SR);
};

//---------------------------- compressor_demo -------------------------
// USAGE: _,_ : compressor_demo : _,_;
//
compressor_demo = bypass2(cbp,compressor_stereo_demo) with {

   comp_group(x) = vgroup("COMPRESSOR  [tooltip: Reference: http://en.wikipedia.org/wiki/Dynamic_range_compression]", x);

   meter_group(x)  = comp_group(hgroup("[0]", x));
   knob_group(x)  = comp_group(hgroup("[1]", x));

   cbp = meter_group(checkbox("[0] Bypass  [tooltip: When this is checked, the compressor has no effect]"));

   gainview =
     compression_gain_mono(ratio,threshold,attack,release) : linear2db :
     meter_group(hbargraph("[1] Compressor Gain [unit:dB] [tooltip: Current gain of the compressor in dB]",
      -50,+10));

   displaygain = _,_ <: _,_,(abs,abs:+) : _,_,gainview : _,attach;

   compressor_stereo_demo =
     displaygain(compressor_stereo(ratio,threshold,attack,release)) :
     *(makeupgain), *(makeupgain);

   ctl_group(x)  = knob_group(hgroup("[3] Compression Control", x));

   ratio = ctl_group(hslider("[0] Ratio [style:knob]  [tooltip: A compression Ratio of N means that for each N dB increase in input signal level above Threshold, the output level goes up 1 dB]",
     5, 1, 20, 0.1));

   threshold = ctl_group(hslider("[1] Threshold [unit:dB] [style:knob]  [tooltip: When the signal level exceeds the Threshold (in dB), its level is compressed according to the Ratio]",
     -30, -100, 10, 0.1));

   env_group(x)  = knob_group(hgroup("[4] Compression Response", x));

   attack = env_group(hslider("[1] Attack [unit:ms] [style:knob]  [tooltip: Time constant in ms (1/e smoothing time) for the compression gain to approach (exponentially) a new lower target level (the compression `kicking in')]",
     50, 0, 500, 0.1)) : *(0.001) : max(1/SR);

   release = env_group(hslider("[2] Release [unit:ms] [style: knob]  [tooltip: Time constant in ms (1/e smoothing time) for the compression gain to approach (exponentially) a new higher target level (the compression 'releasing')]",
     500, 0, 1000, 0.1)) : *(0.001) : max(1/SR);

   makeupgain = comp_group(hslider("[5] Makeup Gain [unit:dB]  [tooltip: The compressed-signal output level is increased by this amount (in dB) to make up for the level lost due to compression]",
     40, -96, 96, 0.1)) : db2linear;
};

//------------------------------- limiter_* ------------------------------------
// USAGE:
//     _ : limiter_1176_R4_mono   : _;
//   _,_ : limiter_1176_R4_stereo : _,_;
//
// DESCRIPTION:
//   A limiter guards against hard-clipping.  It can be can be
//   implemented as a compressor having a high threshold (near the
//   clipping level), fast attack and release, and high ratio.  Since
//   the ratio is so high, some knee smoothing is
//   desirable ("soft limiting").  This example is intended
//   to get you started using compressor_* as a limiter, so all
//   parameters are hardwired to nominal values here.
//
// REFERENCE: http://en.wikipedia.org/wiki/1176_Peak_Limiter
//   Ratios: 4 (moderate compression), 8 (severe compression),
//          12 (mild limiting), or 20 to 1 (hard limiting)
//   Att: 20-800 MICROseconds (Note: scaled by ratio in the 1176)
//   Rel: 50-1100 ms (Note: scaled by ratio in the 1176)
//   Mike Shipley likes 4:1 (Grammy-winning mixer for Queen, Tom Petty, etc.)
//     Faster attack gives "more bite" (e.g. on vocals)
//     He hears a bright, clear eq effect as well (not implemented here)
//
limiter_1176_R4_mono = compressor_mono(4,-6,0.0008,0.5);
limiter_1176_R4_stereo = compressor_stereo(4,-6,0.0008,0.5);

//========================== Schroeder Reverberators  ======================

//------------------------------ jcrev,satrev ------------------------------
// USAGE:
//     _ :  jcrev : _,_,_,_
//     _ : satrev : _,_
//
// DESCRIPTION:
//   These artificial reverberators take a mono signal and output stereo
//   (satrev) and quad (jcrev).  They were implemented by John Chowning
//   in the MUS10 computer-music language (descended from Music V by Max
//   Mathews).  They are Schroeder Reverberators, well tuned for their size.
//   Nowadays, the more expensive freeverb is more commonly used (see the
//   Faust examples directory).

// The reverb below was made from a listing of "RV", dated April 14, 1972,
// which was recovered from an old SAIL DART backup tape.
// John Chowning thinks this might be the one that became the
// well known and often copied JCREV:

jcrev = *(0.06) : allpass_chain <: comb_bank :> _ <: mix_mtx with {

  rev1N = component("filter.lib").rev1;

  rev12(len,g) = rev1N(2048,len,g);
  rev14(len,g) = rev1N(4096,len,g);

  allpass_chain =
    rev2(512,347,0.7) :
    rev2(128,113,0.7) :
    rev2( 64, 37,0.7);

  comb_bank =
    rev12(1601,.802),
    rev12(1867,.773),
    rev14(2053,.753),
    rev14(2251,.733);

    mix_mtx = _,_,_,_ <: psum, -psum, asum, -asum : _,_,_,_ with {
    psum = _,_,_,_ :> _;
    asum = *(-1),_,*(-1),_ :> _;
  };
};

// The reverb below was made from a listing of "SATREV", dated May 15, 1971,
// which was recovered from an old SAIL DART backup tape.
// John Chowning thinks this might be the one used on his
// often-heard brass canon sound examples, one of which can be found at
// https://ccrma.stanford.edu/~jos/wav/FM_BrassCanon2.wav

satrev = *(0.2) <: comb_bank :> allpass_chain <: _,*(-1) with {

  rev1N = component("filter.lib").rev1;

  rev11(len,g) = rev1N(1024,len,g);
  rev12(len,g) = rev1N(2048,len,g);

  comb_bank =
    rev11( 778,.827),
    rev11( 901,.805),
    rev11(1011,.783),
    rev12(1123,.764);

  rev2N = component("filter.lib").rev2;

  allpass_chain =
    rev2N(128,125,0.7) :
    rev2N( 64, 42,0.7) :
    rev2N( 16, 12,0.7);
};

//-------------------------------- freeverb --------------------------------
// Freeverb is a widely used, free, open-source Schroeder reverb contributed
// by ``Jezar at Dreampoint.'' See <faust_distribution>/examples/freeverb.dsp

//=============== Feedback Delay Network (FDN) Reverberators  ==============

//-------------------------------- fdnrev0 ---------------------------------
// Pure Feedback Delay Network Reverberator (generalized for easy scaling).
//
// USAGE:
//   <1,2,4,...,N signals> <:
//   fdnrev0(MAXDELAY,delays,BBSO,freqs,durs,loopgainmax,nonl) :>
//   <1,2,4,...,N signals>
//
// WHERE
//   N = 2, 4, 8, ...  (power of 2)
//   MAXDELAY = power of 2 at least as large as longest delay-line length
//   delays = N delay lines, N a power of 2, lengths perferably coprime
//   BBSO = odd positive integer = order of bandsplit desired at freqs
//   freqs  = NB-1 crossover frequencies separating desired frequency bands
//   durs   = NB decay times (t60) desired for the various bands
//   loopgainmax = scalar gain between 0 and 1 used to "squelch" the reverb
//   nonl = nonlinearity (0 to 0.999..., 0 being linear)
//
// REFERENCE:
//   https://ccrma.stanford.edu/~jos/pasp/FDN_Reverberation.html
//
// DEPENDENCIES: filter.lib (filterbank)

fdnrev0(MAXDELAY, delays, BBSO, freqs, durs, loopgainmax, nonl)
  = (bus(2*N) :> bus(N) : delaylines(N)) ~
    (delayfilters(N,freqs,durs) : feedbackmatrix(N))
with {
  N = count(delays);
  NB = count(durs);
//assert(count(freqs)+1==NB);
  delayval(i) = take(i+1,delays);
  dlmax(i) = MAXDELAY; // must hardwire this from argument for now
//dlmax(i) = 2^max(1,nextpow2(delayval(i))) // try when slider min/max is known
//           with { nextpow2(x) = ceil(log(x)/log(2.0)); };
// -1 is for feedback delay:
  delaylines(N) = par(i,N,(delay(dlmax(i),(delayval(i)-1))));
  delayfilters(N,freqs,durs) = par(i,N,filter(i,freqs,durs));
  feedbackmatrix(N) = bhadamard(N);
  vbutterfly(n) = bus(n) <: (bus(n):>bus(n/2)) , ((bus(n/2),(bus(n/2):par(i,n/2,*(-1)))) :> bus(n/2));
  bhadamard(2) = bus(2) <: +,-;
  bhadamard(n) = bus(n) <: (bus(n):>bus(n/2)) , ((bus(n/2),(bus(n/2):par(i,n/2,*(-1)))) :> bus(n/2))
                 : (bhadamard(n/2) , bhadamard(n/2));

  // Experimental nonlinearities:
  // nonlinallpass = apnl(nonl,-nonl);
  // s = nonl*PI;
  // nonlinallpass(x) = allpassnn(3,(s*x,s*x*x,s*x*x*x)); // filter.lib
     nonlinallpass = _; // disabled by default (rather expensive)

  filter(i,freqs,durs) = filterbank(BBSO,freqs) : par(j,NB,*(g(j,i)))
                         :> *(loopgainmax) / sqrt(N) : nonlinallpass
  with {
    dur(j) = take(j+1,durs);
    n60(j) = dur(j)*SR; // decay time in samples
    g(j,i) = exp(-3.0*log(10.0)*delayval(i)/n60(j));
        // ~ 1.0 - 6.91*delayval(i)/(SR*dur(j)); // valid for large dur(j)
  };
};

// ---------- prime_power_delays -----
// Prime Power Delay Line Lengths
//
// USAGE:
//   bus(N) : prime_power_delays(N,pathmin,pathmax) : bus(N);
//
// WHERE
//   N = positive integer up to 16
//       (for higher powers of 2, extend 'primes' array below.)
//   pathmin = minimum acoustic ray length in the reverberator (in meters)
//   pathmax = maximum acoustic ray length (meters) - think "room size"
//
// DEPENDENCIES:
//   math.lib (SR, selector, take)
//   music.lib (db2linear)
//
// REFERENCE:
//   https://ccrma.stanford.edu/~jos/pasp/Prime_Power_Delay_Line.html
//
prime_power_delays(N,pathmin,pathmax) = par(i,N,delayvals(i)) with {
  Np = 16;
  primes = 2,3,5,7,11,13,17,19,23,29,31,37,41,43,47,53;
  prime(n) = primes : selector(n,Np); // math.lib

  // Prime Power Bounds [matlab: floor(log(maxdel)./log(primes(53)))]
  maxdel=8192; // more than 63 meters at 44100 samples/sec & 343 m/s
  ppbs = 13,8,5,4, 3,3,3,3, 2,2,2,2, 2,2,2,2; // 8192 is enough for all
  ppb(i) = take(i+1,ppbs);

  // Approximate desired delay-line lengths using powers of distinct primes:
  c = 343; // soundspeed in m/s at 20 degrees C for dry air
  dmin = SR*pathmin/c;
  dmax = SR*pathmax/c;
  dl(i) = dmin * (dmax/dmin)^(i/float(N-1)); // desired delay in samples
  ppwr(i) = floor(0.5+log(dl(i))/log(prime(i))); // best prime power
  delayvals(i) = prime(i)^ppwr(i); // each delay a power of a distinct prime
};

//--------------------- stereo_reverb_tester --------------------
// Handy test inputs for reverberator demos below.

stereo_reverb_tester(revin_group,x,y) = inx,iny with {
  ck_group(x) = revin_group(vgroup("[1] Input Config",x));
  mutegain = 1 - ck_group(checkbox("[1] Mute Ext Inputs
         [tooltip: When this is checked, the stereo external audio inputs are disabled (good for hearing the impulse response or pink-noise response alone)]"));
  pinkin = ck_group(checkbox("[2] Pink Noise
         [tooltip: Pink Noise (or 1/f noise) is Constant-Q Noise (useful for adjusting the EQ sections)]"));

  impulsify = _ <: _,mem : - : >(0);
  imp_group(x) = revin_group(hgroup("[2] Impulse Selection",x));
  pulseL =  imp_group(button("[1] Left
         [tooltip: Send impulse into LEFT channel]")) : impulsify;
  pulseC =  imp_group(button("[2] Center
         [tooltip: Send impulse into LEFT and RIGHT channels]")) : impulsify;
  pulseR =  imp_group(button("[3] Right
         [tooltip: Send impulse into RIGHT channel]")) : impulsify;

  inx = x*mutegain + (pulseL+pulseC) + pn;
  iny = y*mutegain + (pulseR+pulseC) + pn;
  pn = 0.1*pinkin*component("oscillator.lib").pink_noise;
};

//------------------------- fdnrev0_demo ---------------------------
// USAGE: _,_ : fdnrev0_demo(N,NB,BBSO) : _,_
// WHERE
//    N = Feedback Delay Network (FDN) order
//      = number of delay lines used = order of feedback matrix
//      = 2, 4, 8, or 16 [extend primes array below for 32, 64, ...]
//   NB = number of frequency bands
//      = number of (nearly) independent T60 controls
//      = integer 3 or greater
// BBSO = Butterworth band-split order
//      = order of lowpass/highpass bandsplit used at each crossover freq
//      = odd positive integer

fdnrev0_demo(N,NB,BBSO,x,y) = stereo_reverb_tester(revin_group,x,y)
          <: fdnrev0(MAXDELAY,delays,BBSO,freqs,durs,loopgainmax,nonl)
          :> *(gain),*(gain)
with {
  MAXDELAY = 8192; // sync w delays and prime_power_delays above
  defdurs = (8.4,6.5,5.0,3.8,2.7); // NB default durations (sec)
  deffreqs = (500,1000,2000,4000); // NB-1 default crossover frequencies (Hz)
  deflens = (56.3,63.0); // 2 default min and max path lengths

  fdn_group(x)  = vgroup("FEEDBACK DELAY NETWORK (FDN) REVERBERATOR, ORDER 16
    [tooltip: See Faust's effect.lib for documentation and references]", x);

  freq_group(x)  = fdn_group(vgroup("[1] Band Crossover Frequencies", x));
  t60_group(x)  = fdn_group(hgroup("[2] Band Decay Times (T60)", x));
  path_group(x)  = fdn_group(vgroup("[3] Room Dimensions", x));
  revin_group(x)  = fdn_group(hgroup("[4] Input Controls", x));
  nonl_group(x) = revin_group(vgroup("[4] Nonnlinearity",x));
  quench_group(x) = revin_group(vgroup("[3] Reverb State",x));

  nonl = nonl_group(hslider("[style:knob] [tooltip: nonlinear mode coupling]",
            0, -0.999, 0.999, 0.001));
  loopgainmax = 1.0-0.5*quench_group(button("[1] Quench
         [tooltip: Hold down 'Quench' to clear the reverberator]"));

  pathmin = path_group(hslider("[1] min acoustic ray length [unit:m]
    [tooltip: This length (in meters) determines the shortest delay-line used in the FDN reverberator.
              Think of it as the shortest wall-to-wall separation in the room.]",
            46, 0.1, 63, 0.1));
  pathmax = path_group(hslider("[2] max acoustic ray length [unit:m]
    [tooltip: This length (in meters) determines the longest delay-line used in the FDN reverberator.
              Think of it as the largest wall-to-wall separation in the room.]",
            63, 0.1, 63, 0.1));

  durvals(i) = t60_group(vslider("[%i] %i [unit:s]
    [tooltip: T60 is the 60dB decay-time in seconds. For concert halls, an overall reverberation time (T60) near 1.9 seconds is typical [Beranek 2004]. Here we may set T60 independently in each frequency band.  In real rooms, higher frequency bands generally decay faster due to absorption and scattering.]",
    take(i+1,defdurs), 0.1, 10, 0.1));
  durs = par(i,NB,durvals(NB-1-i));

  freqvals(i) = freq_group(hslider("[%i] Band %i upper edge in Hz [unit:Hz]
    [tooltip: Each delay-line signal is split into frequency-bands for separate decay-time control in each band]",
    take(i+1,deffreqs), 100, 10000, 1));
  freqs = par(i,NB-1,freqvals(i));

  delays = prime_power_delays(N,pathmin,pathmax);

  gain = hslider("[3] Output Level (dB) [unit:dB]
    [tooltip: Output scale factor]", -40, -70, 20, 0.1) : db2linear;
     // (can cause infinite loop:) with { db2linear(x) = pow(10, x/20.0); };
};

//------------------------------- zita_rev_fdn -------------------------------
// Internal 8x8 late-reverberation FDN used in the FOSS Linux reverb zita-rev1
// by Fons Adriaensen <fons@linuxaudio.org>.  This is an FDN reverb with
// allpass comb filters in each feedback delay in addition to the
// damping filters.
//
// USAGE:
//   bus(8) : zita_rev_fdn(f1,f2,t60dc,t60m,fsmax) : bus(8)
//
// WHERE
//   f1    = crossover frequency (Hz) separating dc and midrange frequencies
//   f2    = frequency (Hz) above f1 where T60 = t60m/2 (see below)
//   t60dc = desired decay time (t60) at frequency 0 (sec)
//   t60m  = desired decay time (t60) at midrange frequencies (sec)
//   fsmax = maximum sampling rate to be used (Hz)
//
// REFERENCES:
//   http://www.kokkinizita.net/linuxaudio/zita-rev1-doc/quickguide.html
//   https://ccrma.stanford.edu/~jos/pasp/Zita_Rev1.html
//
// DEPENDENCIES:
//   filter.lib (allpass_comb, lowpass, smooth)
//   math.lib (hadamard, take, etc.)

zita_rev_fdn(f1,f2,t60dc,t60m,fsmax) =
  ((bus(2*N) :> allpass_combs(N) : feedbackmatrix(N)) ~
   (delayfilters(N,freqs,durs) : fbdelaylines(N)))
with {
  N = 8;

  // Delay-line lengths in seconds:
  apdelays = (0.020346, 0.024421, 0.031604, 0.027333, 0.022904,
              0.029291, 0.013458, 0.019123); // feedforward delays in seconds
  tdelays = ( 0.153129, 0.210389, 0.127837, 0.256891, 0.174713,
              0.192303, 0.125000, 0.219991); // total delays in seconds
  tdelay(i) = floor(0.5 + SR*take(i+1,tdelays)); // samples
  apdelay(i) = floor(0.5 + SR*take(i+1,apdelays));
  fbdelay(i) = tdelay(i) - apdelay(i);
  // NOTE: Since SR is not bounded at compile time, we can't use it to
  // allocate delay lines; hence, the fsmax parameter:
  tdelaymaxfs(i) = floor(0.5 + fsmax*take(i+1,tdelays));
  apdelaymaxfs(i) = floor(0.5 + fsmax*take(i+1,apdelays));
  fbdelaymaxfs(i) = tdelaymaxfs(i) - apdelaymaxfs(i);
  nextpow2(x) = ceil(log(x)/log(2.0));
  maxapdelay(i) = int(2.0^max(1.0,nextpow2(apdelaymaxfs(i))));
  maxfbdelay(i) = int(2.0^max(1.0,nextpow2(fbdelaymaxfs(i))));

  apcoeff(i) = select2(i&1,0.6,-0.6);  // allpass comb-filter coefficient
  allpass_combs(N) =
    par(i,N,(allpass_comb(maxapdelay(i),apdelay(i),apcoeff(i)))); // filter.lib
  fbdelaylines(N) = par(i,N,(delay(maxfbdelay(i),(fbdelay(i)))));
  freqs = (f1,f2); durs = (t60dc,t60m);
  delayfilters(N,freqs,durs) = par(i,N,filter(i,freqs,durs));
  feedbackmatrix(N) = hadamard(N); // math.lib

  staynormal = 10.0^(-20); // let signals decay well below LSB, but not to zero

  special_lowpass(g,f) = smooth(p) with {
    // unity-dc-gain lowpass needs gain g at frequency f => quadratic formula:
    p = mbo2 - sqrt(max(0,mbo2*mbo2 - 1.0)); // other solution is unstable
    mbo2 = (1.0 - gs*c)/(1.0 - gs); // NOTE: must ensure |g|<1 (t60m finite)
    gs = g*g;
    c = cos(2.0*PI*f/float(SR));
  };

  filter(i,freqs,durs) = lowshelf_lowpass(i)/sqrt(float(N))+staynormal
  with {
    lowshelf_lowpass(i) = gM*low_shelf1_l(g0/gM,f(1)):special_lowpass(gM,f(2));
    low_shelf1_l(G0,fx,x) = x + (G0-1)*lowpass(1,fx,x); // filter.lib
    g0 = g(0,i);
    gM = g(1,i);
    f(k) = take(k,freqs);
    dur(j) = take(j+1,durs);
    n60(j) = dur(j)*SR; // decay time in samples
    g(j,i) = exp(-3.0*log(10.0)*tdelay(i)/n60(j));
  };
};

// Stereo input delay used by zita_rev1 in both stereo and ambisonics mode:
zita_in_delay(rdel) = zita_delay_mono(rdel), zita_delay_mono(rdel) with {
  zita_delay_mono(rdel) = delay(8192,SR*rdel*0.001) * 0.3;
};

// Stereo input mapping used by zita_rev1 in both stereo and ambisonics mode:
zita_distrib2(N) = _,_ <: fanflip(N) with {
   fanflip(4) = _,_,*(-1),*(-1);
   fanflip(N) = fanflip(N/2),fanflip(N/2);
};

//--------------------------- zita_rev_fdn_demo ------------------------------
// zita_rev_fdn_demo = zita_rev_fdn (above) + basic GUI
//
// USAGE:
//   bus(8) : zita_rev_fdn_demo(f1,f2,t60dc,t60m,fsmax) : bus(8)
//
// WHERE
//   (args and references as for zita_rev_fdn above)

zita_rev_fdn_demo = zita_rev_fdn(f1,f2,t60dc,t60m,fsmax)
with {
  fsmax = 48000.0;
  fdn_group(x) = hgroup(
    "Zita_Rev Internal FDN Reverb [tooltip: ~ Zita_Rev's internal 8x8 Feedback Delay Network (FDN) & Schroeder allpass-comb reverberator.  See Faust's effect.lib for documentation and references]",x);
  t60dc = fdn_group(vslider("[1] Low RT60 [unit:s] [style:knob]
    [style:knob]
    [tooltip: T60 = time (in seconds) to decay 60dB in low-frequency band]",
    3, 1, 8, 0.1));
  f1 = fdn_group(vslider("[2] LF X [unit:Hz] [style:knob]
    [tooltip: Crossover frequency (Hz) separating low and middle frequencies]",
    200, 50, 1000, 1));
  t60m = fdn_group(vslider("[3] Mid RT60 [unit:s] [style:knob]
    [tooltip: T60 = time (in seconds) to decay 60dB in middle band]",
    2, 1, 8, 0.1));
  f2 = fdn_group(vslider("[4] HF Damping [unit:Hz] [style:knob]
    [tooltip: Frequency (Hz) at which the high-frequency T60 is half the middle-band's T60]",
    6000, 1500, 0.49*fsmax, 1));
};

//---------------------------- zita_rev1_stereo ---------------------------
// Extend zita_rev_fdn to include zita_rev1 input/output mapping in stereo mode.
//
// USAGE:
//   _,_ : zita_rev1_stereo(rdel,f1,f2,t60dc,t60m,fsmax) : _,_
//
// WHERE
//   rdel  = delay (in ms) before reverberation begins (e.g., 0 to ~100 ms)
//   (remaining args and refs as for zita_rev_fdn above)

zita_rev1_stereo(rdel,f1,f2,t60dc,t60m,fsmax) =
   zita_in_delay(rdel)
 : zita_distrib2(N)
 : zita_rev_fdn(f1,f2,t60dc,t60m,fsmax)
 : output2(N)
with {
 N = 8;
 output2(N) = outmix(N) : *(t1),*(t1);
 t1 = 0.37; // zita-rev1 linearly ramps from 0 to t1 over one buffer
 outmix(4) = !,butterfly(2),!; // probably the result of some experimenting!
 outmix(N) = outmix(N/2),par(i,N/2,!);
};

//----------------------------- zita_rev1_ambi ---------------------------
// Extend zita_rev_fdn to include zita_rev1 input/output mapping in
// "ambisonics mode", as provided in the Linux C++ version.
//
// USAGE:
//   _,_ : zita_rev1_ambi(rgxyz,rdel,f1,f2,t60dc,t60m,fsmax) : _,_,_,_
//
// WHERE
//   rgxyz = relative gain of lanes 1,4,2 to lane 0 in output (e.g., -9 to 9)
//   (remaining args and references as for zita_rev1_stereo above)

zita_rev1_ambi(rgxyz,rdel,f1,f2,t60dc,t60m,fsmax) =
   zita_in_delay(rdel)
 : zita_distrib2(N)
 : zita_rev_fdn(f1,f2,t60dc,t60m,fsmax)
 : output4(N) // ambisonics mode
with {
  N=8;
  output4(N) = select4 : *(t0),*(t1),*(t1),*(t1);
  select4 = _,_,_,!,_,!,!,! : _,_,cross with { cross(x,y) = y,x; };
  t0 = 1.0/sqrt(2.0);
  t1 = t0 * 10.0^(0.05 * rgxyz);
};

//---------------------------------- zita_rev1 ------------------------------
// Example GUI for zita_rev1_stereo (mostly following the Linux zita-rev1 GUI).
//
// Only the dry/wet and output level parameters are "dezippered" here.  If
// parameters are to be varied in real time, use "smooth(0.999)" or the like
// in the same way.
//
// REFERENCE:
//   http://www.kokkinizita.net/linuxaudio/zita-rev1-doc/quickguide.html
//
// DEPENDENCIES:
//   filter.lib (peak_eq_rm)

zita_rev1(x,y) = zita_rev1_stereo(rdel,f1,f2,t60dc,t60m,fsmax,x,y)
          : out_eq : dry_wet(x,y) : out_level
with {

  fsmax = 48000.0;  // highest sampling rate that will be used

  fdn_group(x) = hgroup(
    "[0] Zita_Rev1 [tooltip: ~ ZITA REV1 FEEDBACK DELAY NETWORK (FDN) & SCHROEDER ALLPASS-COMB REVERBERATOR (8x8). See Faust's effect.lib for documentation and references]", x);

  in_group(x) = fdn_group(hgroup("[1] Input", x));

  rdel = in_group(vslider("[1] In Delay [unit:ms] [style:knob]
                  [tooltip: Delay in ms before reverberation begins]",
                  60,20,100,1));

  freq_group(x) = fdn_group(hgroup("[2] Decay Times in Bands (see tooltips)", x));

  f1 = freq_group(vslider("[1] LF X [unit:Hz] [style:knob]
       [tooltip: Crossover frequency (Hz) separating low and middle frequencies]",
       200, 50, 1000, 1));

  t60dc = freq_group(vslider("[2] Low RT60 [unit:s] [style:knob]
          [style:knob] [tooltip: T60 = time (in seconds) to decay 60dB in low-frequency band]",
          3, 1, 8, 0.1));

  t60m = freq_group(vslider("[3] Mid RT60 [unit:s] [style:knob]
          [tooltip: T60 = time (in seconds) to decay 60dB in middle band]",
          2, 1, 8, 0.1));

  f2 = freq_group(vslider("[4] HF Damping [unit:Hz] [style:knob]
       [tooltip: Frequency (Hz) at which the high-frequency T60 is half the middle-band's T60]",
       6000, 1500, 0.49*fsmax, 1));

  out_eq = pareq_stereo(eq1f,eq1l,eq1q) : pareq_stereo(eq2f,eq2l,eq2q);
// Zolzer style peaking eq (not used in zita-rev1) (filter.lib):
// pareq_stereo(eqf,eql,Q) = peak_eq(eql,eqf,eqf/Q), peak_eq(eql,eqf,eqf/Q);
// Regalia-Mitra peaking eq with "Q" hard-wired near sqrt(g)/2 (filter.lib):
  pareq_stereo(eqf,eql,Q) = peak_eq_rm(eql,eqf,tpbt), peak_eq_rm(eql,eqf,tpbt)
  with {
    tpbt = wcT/sqrt(max(0,g)); // tan(PI*B/SR), B bw in Hz (Q^2 ~ g/4)
    wcT = 2*PI*eqf/SR;  // peak frequency in rad/sample
    g = db2linear(eql); // peak gain
  };

  eq1_group(x) = fdn_group(hgroup("[3] RM Peaking Equalizer 1", x));

  eq1f = eq1_group(vslider("[1] Eq1 Freq [unit:Hz] [style:knob]
       [tooltip: Center-frequency of second-order Regalia-Mitra peaking equalizer section 1]",
       315, 40, 2500, 1));

  eq1l = eq1_group(vslider("[2] Eq1 Level [unit:dB] [style:knob]
       [tooltip: Peak level in dB of second-order Regalia-Mitra peaking equalizer section 1]",
       0, -15, 15, 0.1));

  eq1q = eq1_group(vslider("[3] Eq1 Q [style:knob]
       [tooltip: Q = centerFrequency/bandwidth of second-order peaking equalizer section 1]",
       3, 0.1, 10, 0.1));

  eq2_group(x) = fdn_group(hgroup("[4] RM Peaking Equalizer 2", x));

  eq2f = eq2_group(vslider("[1] Eq2 Freq [unit:Hz] [style:knob]
       [tooltip: Center-frequency of second-order Regalia-Mitra peaking equalizer section 2]",
       315, 40, 2500, 1));

  eq2l = eq2_group(vslider("[2] Eq2 Level [unit:dB] [style:knob]
       [tooltip: Peak level in dB of second-order Regalia-Mitra peaking equalizer section 2]",
       0, -15, 15, 0.1));

  eq2q = eq2_group(vslider("[3] Eq2 Q [style:knob]
       [tooltip: Q = centerFrequency/bandwidth of second-order peaking equalizer section 2]",
       3, 0.1, 10, 0.1));

  out_group(x)  = fdn_group(hgroup("[5] Output", x));

  dry_wet(x,y) = *(wet) + dry*x, *(wet) + dry*y with {
    wet = 0.5*(drywet+1.0);
    dry = 1.0-wet;
  };

  drywet = out_group(vslider("[1] Dry/Wet Mix [style:knob]
       [tooltip: -1 = dry, 1 = wet]",
       0, -1.0, 1.0, 0.01)) : smooth(0.999);

  out_level = *(gain),*(gain);

  gain = out_group(vslider("[2] Level [unit:dB] [style:knob]
    [tooltip: Output scale factor]", -20, -70, 40, 0.1))
    : smooth(0.999) : db2linear;

};

//---------------------------------- mesh_square ------------------------------
// Square Rectangular Digital Waveguide Mesh
//
// USAGE:
//   bus(4*N) : mesh_square(N) : bus(4*N);
//
// WHERE
//   N = number of nodes along each edge - a power of two (1,2,4,8,...)
//
// EXAMPLE: Reflectively terminated mesh impulsed at one corner:
//   mesh_square_test(N,x) = mesh_square(N)~(busi(4*N,x)) // input to corner
//      with { busi(N,x) = bus(N) : par(i,N,*(-1)) : par(i,N-1,_), +(x); };
//   process = 1-1' : mesh_square_test(4); // all modes excited forever
//
// REQUIRES: math.lib.
//
// REFERENCE:
//   https://ccrma.stanford.edu/~jos/pasp/Digital_Waveguide_Mesh.html

// four-port scattering junction:
mesh_square(1)
  = bus(4) <: par(i,4,*(-1)), (bus(4) :> (*(.5)) <: bus(4)) :> bus(4);

// rectangular NxN square waveguide mesh:
mesh_square(N) = bus(4*N) : (route_inputs(N/2) : par(i,4,mesh_square(N/2)))
          ~(prune_feedback(N/2))
          : prune_outputs(N/2) : route_outputs(N/2) : bus(4*N)
with {
  block(N) = par(i,N,!);

  // select block i of N, block size = M:
  s(i,N,M) = par(j, M*N, Sv(i, j))
     with { Sv(i,i) = bus(N); Sv(i,j) = block(N); };

  // prune mesh outputs down to the signals which make it out:
  prune_outputs(N)
    = bus(16*N) :
      block(N), bus(N),   block(N), bus(N),
      block(N), bus(N),   bus(N),   block(N),
      bus(N),   block(N), block(N), bus(N),
      bus(N),   block(N), bus(N),   block(N)
      : bus(8*N);

  // collect mesh outputs into standard order (N,W,E,S):
  route_outputs(N)
    = bus(8*N)
      <: s(4,N,8),s(5,N,8), s(0,N,8),s(2,N,8),
         s(3,N,8),s(7,N,8), s(1,N,8),s(6,N,8)
      : bus(8*N);

  // collect signals used as feedback:
  prune_feedback(N) = bus(16*N) :
      bus(N),   block(N), bus(N),   block(N),
      bus(N),   block(N), block(N), bus(N),
      block(N), bus(N),   bus(N),   block(N),
      block(N), bus(N),   block(N), bus(N) :
      bus(8*N);

  // route mesh inputs (feedback, external inputs):
  route_inputs(N) = bus(8*N), bus(8*N)
  <:s(8,N,16),s(4,N,16), s(12,N,16),s(3,N,16),
    s(9,N,16),s(6,N,16), s(1,N,16),s(14,N,16),
    s(0,N,16),s(10,N,16), s(13,N,16),s(7,N,16),
    s(2,N,16),s(11,N,16), s(5,N,16),s(15,N,16)
    : bus(16*N);
};