surroundize

Tool/PipeWire filter to convert stereo audio to surround
git clone https://git.neptards.moe/u3shit/surroundize.git
Log | Files | Refs | README | LICENSE

freesurround_decoder.cpp (12096B)


      1 /*
      2 Copyright (C) 2007-2010 Christian Kothe
      3 
      4 This program is free software; you can redistribute it and/or
      5 modify it under the terms of the GNU General Public License
      6 as published by the Free Software Foundation; either version 2
      7 of the License, or (at your option) any later version.
      8 
      9 This program is distributed in the hope that it will be useful,
     10 but WITHOUT ANY WARRANTY; without even the implied warranty of
     11 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
     12 GNU General Public License for more details.
     13 
     14 You should have received a copy of the GNU General Public License
     15 along with this program; if not, write to the Free Software
     16 Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301, USA.
     17 */
     18 
     19 #include <cmath>
     20 #include <vector>
     21 #include <complex>
     22 #include "kiss_fftr.h"
     23 #include "freesurround_decoder.h"
     24 #include "channelmaps.h"
     25 
     26 typedef std::complex<double> cplx;
     27 const double pi = M_PI;
     28 const double epsilon = 0.000001;
     29 using namespace std;
     30 
     31 #undef min
     32 #undef max
     33 
     34 // FreeSurround implementation
     35 class decoder_impl {
     36 public:
     37 	// instantiate the decoder with a given channel setup and processing block size (in samples)
     38 	decoder_impl(channel_setup setup, unsigned N)
     39       : N(N), C((unsigned)chn_alloc[setup].size()), setup(setup),
     40         lt(N), rt(N), dst(N), lf(N/2+1), rf(N/2+1),
     41         forward(kiss_fftr_alloc(N,0,0,0)), inverse(kiss_fftr_alloc(N,1,0,0)),
     42         buffer_empty(true), inbuf(3*N), wnd(N)
     43 	{
     44 		// allocate per-channel buffers
     45 		outbuf.resize((N+N/2)*C);
     46 		signal.resize(C,vector<cplx>(N));
     47 
     48 		// init the window function
     49 		for (unsigned k=0;k<N;k++)
     50 			wnd[k] = sqrt(0.5*(1-cos(2*pi*k/N))/N);
     51 
     52 		// set default parameters
     53 		set_circular_wrap(90);
     54 		set_shift(0);
     55 		set_depth(1);
     56 		set_focus(0);
     57 		set_front_separation(1);
     58 		set_rear_separation(1);
     59 		set_low_cutoff(40.0/22050);
     60 		set_high_cutoff(90.0/22050);
     61 		set_bass_redirection(false);
     62 	}
     63 
     64 	~decoder_impl()
     65     {
     66       // kiss_fftr_alloc does new char[] inside...
     67       delete[] reinterpret_cast<char*>(forward);
     68       delete[] reinterpret_cast<char*>(inverse);
     69     }
     70 
     71 	// decode a stereo chunk, produces a multichannel chunk of the same size (lagged)
     72 	const float* decode(const float *input) {
     73 		// append incoming data to the end of the input buffer
     74 		memcpy(&inbuf[N], &input[0], 8*N);
     75 		// process first and second half, overlapped
     76 		buffered_decode(&inbuf[0]);
     77 		buffered_decode(&inbuf[N]);
     78 		// shift last half of the input to the beginning (for overlapping with a future block)
     79 		memcpy(&inbuf[0], &inbuf[2*N], 4*N);
     80 		buffer_empty = false;
     81 		return &outbuf[0];
     82 	}
     83 
     84 	// flush the internal buffers
     85 	void flush() {
     86 		memset(&outbuf[0],0,outbuf.size()*4);
     87 		memset(&inbuf[0],0,inbuf.size()*4);
     88 		buffer_empty = true;
     89 	}
     90 
     91 	// number of samples currently held in the buffer
     92 	unsigned buffered() { return buffer_empty ? 0 : N/2; }
     93 
     94 	// set soundfield & rendering parameters
     95 	void set_circular_wrap(float v) { circular_wrap = v; }
     96 	void set_shift(float v) { shift = v; }
     97 	void set_depth(float v) { depth = v; }
     98 	void set_focus(float v) { focus = v; }
     99 	void set_front_separation(float v) { front_separation = v; }
    100 	void set_rear_separation(float v) { rear_separation = v; }
    101 	void set_low_cutoff(float v) { lo_cut = v*(N/2); }
    102 	void set_high_cutoff(float v) { hi_cut = v*(N/2); }
    103 	void set_bass_redirection(bool v) { use_lfe = v; }
    104 
    105 private:
    106 	// helper functions
    107 	static inline double sqr(double x) { return x*x; }
    108 	static inline double amplitude(const cplx &x) { return sqrt(sqr(x.real()) + sqr(x.imag())); }
    109 	static inline double phase(const cplx &x) { return atan2(x.imag(),x.real()); }
    110 	static inline cplx polar(double a, double p) { return cplx(a*cos(p),a*sin(p)); }
    111 	static inline double min(double a, double b) { return a<b?a:b; }
    112 	static inline double max(double a, double b) { return a>b?a:b; }
    113 	static inline double clamp(double x) { return max(-1,min(1,x)); }
    114 	static inline double sign(double x) { return x<0?-1:(x>0?1:0); }
    115 	// get the distance of the soundfield edge, along a given angle
    116 	static inline double edgedistance(double a) { return min(sqrt(1+sqr(tan(a))),sqrt(1+sqr(1/tan(a)))); }
    117 	// get the index (and fractional offset!) in a piecewise-linear channel allocation grid
    118 	int map_to_grid(double &x) { double gp=((x+1)*0.5)*(grid_res-1), i=min(grid_res-2,floor(gp)); x = gp-i; return i; }
    119 
    120 	// decode a block of data and overlap-add it into outbuf
    121 	void buffered_decode(float *input) {
    122 		// demultiplex and apply window function
    123 		for (unsigned k=0;k<N;k++) {
    124 			lt[k] = wnd[k]*input[k*2+0];
    125 			rt[k] = wnd[k]*input[k*2+1];
    126 		}
    127 
    128 		// map into spectral domain
    129 		kiss_fftr(forward,&lt[0],(kiss_fft_cpx*)&lf[0]);
    130 		kiss_fftr(forward,&rt[0],(kiss_fft_cpx*)&rf[0]);
    131 
    132 		// compute multichannel output signal in the spectral domain
    133 		for (unsigned f=1;f<N/2;f++) {
    134 			// get Lt/Rt amplitudes & phases
    135 			double ampL = amplitude(lf[f]), ampR = amplitude(rf[f]);
    136 			double phaseL = phase(lf[f]), phaseR = phase(rf[f]);
    137 			// calculate the amplitude & phase differences
    138 			double ampDiff = clamp((ampL+ampR < epsilon) ? 0 : (ampR-ampL) / (ampR+ampL));
    139 			double phaseDiff = abs(phaseL - phaseR);
    140 			if (phaseDiff > pi) phaseDiff = 2*pi - phaseDiff;
    141 
    142 			// decode into x/y soundfield position
    143 			double x,y; transform_decode(ampDiff,phaseDiff,x,y);
    144 			// add wrap control
    145 			transform_circular_wrap(x,y,circular_wrap);
    146 			// add shift control
    147 			y = clamp(y - shift);
    148 			// add depth control
    149 			y = clamp(1 - (1-y)*depth);
    150 			// add focus control
    151 			transform_focus(x,y,focus);
    152 			// add crossfeed control
    153 			x = clamp(x * (front_separation*(1+y)/2 + rear_separation*(1-y)/2));
    154 
    155 			// get total signal amplitude
    156 			double amp_total = sqrt(ampL*ampL + ampR*ampR);
    157 			// and total L/C/R signal phases
    158 			double phase_of[] = {phaseL,atan2(lf[f].imag()+rf[f].imag(),lf[f].real()+rf[f].real()),phaseR};
    159 			// compute 2d channel map indexes p/q and update x/y to fractional offsets in the map grid
    160 			int p=map_to_grid(x), q=map_to_grid(y);
    161 			// map position to channel volumes
    162 			for (unsigned c=0;c<C-1;c++) {
    163 				// look up channel map at respective position (with bilinear interpolation) and build the signal
    164 				vector<float*> &a = chn_alloc[setup][c];
    165 				signal[c][f] = polar(amp_total*((1-x)*(1-y)*a[q][p] + x*(1-y)*a[q][p+1] + (1-x)*y*a[q+1][p] + x*y*a[q+1][p+1]),
    166 					phase_of[1+(int)sign(chn_xsf[setup][c])]);
    167 			}
    168 
    169 			// optionally redirect bass
    170 			if (use_lfe && f < hi_cut) {
    171 				// level of LFE channel according to normalized frequency
    172 				double lfe_level = f < lo_cut ? 1 : 0.5*(1+cos(pi*(f-lo_cut)/(hi_cut-lo_cut)));
    173 				// assign LFE channel
    174 				signal[C-1][f] = lfe_level * polar(amp_total,phase_of[1]);
    175 				// subtract the signal from the other channels
    176 				for (unsigned c=0;c<C-1;c++)
    177 					signal[c][f] *= (1-lfe_level);
    178 			}
    179 		}
    180 
    181 		// shift the last 2/3 to the first 2/3 of the output buffer
    182 		memmove(&outbuf[0], &outbuf[C*N/2], N*C*4);
    183 		// and clear the rest
    184 		memset(&outbuf[C*N], 0, C*4*N/2);
    185 		// backtransform each channel and overlap-add
    186 		for (unsigned c=0;c<C;c++) {
    187 			// back-transform into time domain
    188 			kiss_fftri(inverse,(kiss_fft_cpx*)&signal[c][0],&dst[0]);
    189 			// add the result to the last 2/3 of the output buffer, windowed (and remultiplex)
    190 			for (unsigned k=0;k<N;k++)
    191 				outbuf[C*(k+N/2)+c] += wnd[k]*dst[k];
    192 		}
    193 	}
    194 
    195 	// transform amp/phase difference space into x/y soundfield space
    196 	void transform_decode(double a, double p, double &x, double &y) {
    197 		x = clamp(1.0047*a + 0.46804*a*p*p*p - 0.2042*a*p*p*p*p + 0.0080586*a*p*p*p*p*p*p*p - 0.0001526*a*p*p*p*p*p*p*p*p*p*p
    198 			- 0.073512*a*a*a*p - 0.2499*a*a*a*p*p*p*p + 0.016932*a*a*a*p*p*p*p*p*p*p - 0.00027707*a*a*a*p*p*p*p*p*p*p*p*p*p
    199 			+ 0.048105*a*a*a*a*a*p*p*p*p*p*p*p - 0.0065947*a*a*a*a*a*p*p*p*p*p*p*p*p*p*p + 0.0016006*a*a*a*a*a*p*p*p*p*p*p*p*p*p*p*p
    200 			- 0.0071132*a*a*a*a*a*a*a*p*p*p*p*p*p*p*p*p + 0.0022336*a*a*a*a*a*a*a*p*p*p*p*p*p*p*p*p*p*p
    201 			- 0.0004804*a*a*a*a*a*a*a*p*p*p*p*p*p*p*p*p*p*p*p);
    202 		y = clamp(0.98592 - 0.62237*p + 0.077875*p*p - 0.0026929*p*p*p*p*p + 0.4971*a*a*p - 0.00032124*a*a*p*p*p*p*p*p
    203 			+ 9.2491e-006*a*a*a*a*p*p*p*p*p*p*p*p*p*p + 0.051549*a*a*a*a*a*a*a*a + 1.0727e-014*a*a*a*a*a*a*a*a*a*a);
    204 	}
    205 
    206 	// apply a circular_wrap transformation to some position
    207 	void transform_circular_wrap(double &x, double &y, double refangle) {
    208 		if (refangle == 90)
    209 			return;
    210 		refangle = refangle*pi/180;
    211 		double baseangle = 90*pi/180;
    212 		// translate into edge-normalized polar coordinates
    213 		double ang = atan2(x,y), len = sqrt(x*x+y*y);
    214 		len = len / edgedistance(ang);
    215 		// apply circular_wrap transform
    216 		if (abs(ang) < baseangle/2)
    217 			// angle falls within the front region (to be enlarged)
    218 			ang *= refangle / baseangle;
    219 		else
    220 			// angle falls within the rear region (to be shrunken)
    221 			ang = pi - (-(((refangle - 2*pi)*(pi - abs(ang))*sign(ang))/(2*pi - baseangle)));
    222 		// translate back into soundfield position
    223 		len = len * edgedistance(ang);
    224 		x = clamp(sin(ang)*len);
    225 		y = clamp(cos(ang)*len);
    226 	}
    227 
    228 	// apply a focus transformation to some position
    229 	void transform_focus(double &x, double &y, double focus) {
    230 		if (focus == 0)
    231 			return;
    232 		// translate into edge-normalized polar coordinates
    233 		double ang = atan2(x,y), len = clamp(sqrt(x*x+y*y)/edgedistance(ang));
    234 		// apply focus
    235 		len = focus > 0 ? 1-pow(1-len,1+focus*20) : pow(len,1-focus*20);
    236 		// back-transform into euclidian soundfield position
    237 		len = len * edgedistance(ang);
    238 		x = clamp(sin(ang)*len);
    239 		y = clamp(cos(ang)*len);
    240 	}
    241 
    242 	// constants
    243 	unsigned N,C;					// number of samples per input/output block, number of output channels
    244 	channel_setup setup;			// the channel setup
    245 
    246 	// parameters
    247 	float circular_wrap;			// angle of the front soundstage around the listener (90°=default)
    248 	float shift;					// forward/backward offset of the soundstage
    249 	float depth;					// backward extension of the soundstage
    250 	float focus;					// localization of the sound events
    251 	float front_separation;			// front stereo separation
    252 	float rear_separation;			// rear stereo separation
    253 	float lo_cut, hi_cut;			// LFE cutoff frequencies
    254 	bool use_lfe;					// whether to use the LFE channel
    255 
    256 	// FFT data structures
    257 	vector<double> lt,rt,dst;		// left total, right total (source arrays), time-domain destination buffer array
    258 	vector<cplx> lf,rf;				// left total / right total in frequency domain
    259 	kiss_fftr_cfg forward,inverse;	// FFT buffers
    260 
    261 	// buffers
    262 	bool buffer_empty;				// whether the buffer is currently empty or dirty
    263 	vector<float> inbuf;			// stereo input buffer (multiplexed)
    264 	vector<float> outbuf;			// multichannel output buffer (multiplexed)
    265 	vector<double> wnd;				// the window function, precomputed
    266 	vector<vector<cplx> > signal;	// the signal to be constructed in every channel, in the frequency domain
    267 };
    268 
    269 
    270 // implementation of the shell class
    271 freesurround_decoder::freesurround_decoder(channel_setup setup, unsigned blocksize): impl(new decoder_impl(setup,blocksize)) { }
    272 freesurround_decoder::~freesurround_decoder() { delete impl; }
    273 const float* freesurround_decoder::decode(const float* input) { return impl->decode(input); }
    274 void freesurround_decoder::flush() { impl->flush(); }
    275 void freesurround_decoder::circular_wrap(float v) { impl->set_circular_wrap(v); }
    276 void freesurround_decoder::shift(float v) { impl->set_shift(v); }
    277 void freesurround_decoder::depth(float v) { impl->set_depth(v); }
    278 void freesurround_decoder::focus(float v) { impl->set_focus(v); }
    279 void freesurround_decoder::front_separation(float v) { impl->set_front_separation(v); }
    280 void freesurround_decoder::rear_separation(float v) { impl->set_rear_separation(v); }
    281 void freesurround_decoder::low_cutoff(float v) { impl->set_low_cutoff(v); }
    282 void freesurround_decoder::high_cutoff(float v) { impl->set_high_cutoff(v); }
    283 void freesurround_decoder::bass_redirection(bool v) { impl->set_bass_redirection(v); }
    284 unsigned freesurround_decoder::buffered() { return impl->buffered(); }
    285 unsigned freesurround_decoder::num_channels(channel_setup s) { return chn_id[s].size(); }
    286 channel_id freesurround_decoder::channel_at(channel_setup s, unsigned i) { return i < chn_id[s].size() ? chn_id[s][i] : ci_none; }