1 /*
2 * Copyright (c) 2012-2013 Clément Bœsch
3 * Copyright (c) 2013 Rudolf Polzer <divverent@xonotic.org>
4 * Copyright (c) 2015 Paul B Mahol
5 *
6 * This file is part of FFmpeg.
7 *
8 * FFmpeg is free software; you can redistribute it and/or
9 * modify it under the terms of the GNU Lesser General Public
10 * License as published by the Free Software Foundation; either
11 * version 2.1 of the License, or (at your option) any later version.
12 *
13 * FFmpeg is distributed in the hope that it will be useful,
14 * but WITHOUT ANY WARRANTY; without even the implied warranty of
15 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
16 * Lesser General Public License for more details.
17 *
18 * You should have received a copy of the GNU Lesser General Public
19 * License along with FFmpeg; if not, write to the Free Software
20 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
21 */
22
23 /**
24 * @file
25 * audio to spectrum (video) transmedia filter, based on ffplay rdft showmode
26 * (by Michael Niedermayer) and lavfi/avf_showwaves (by Stefano Sabatini).
27 */
28
29 #include <math.h>
30
31 #include "libavcodec/avfft.h"
32 #include "libavutil/audio_fifo.h"
33 #include "libavutil/avassert.h"
34 #include "libavutil/avstring.h"
35 #include "libavutil/channel_layout.h"
36 #include "libavutil/opt.h"
37 #include "libavutil/parseutils.h"
38 #include "libavutil/xga_font_data.h"
39 #include "audio.h"
40 #include "video.h"
41 #include "avfilter.h"
42 #include "filters.h"
43 #include "internal.h"
44 #include "window_func.h"
45
46 enum DisplayMode { COMBINED, SEPARATE, NB_MODES };
47 enum DataMode { D_MAGNITUDE, D_PHASE, NB_DMODES };
48 enum FrequencyScale { F_LINEAR, F_LOG, NB_FSCALES };
49 enum DisplayScale { LINEAR, SQRT, CBRT, LOG, FOURTHRT, FIFTHRT, NB_SCALES };
50 enum ColorMode { CHANNEL, INTENSITY, RAINBOW, MORELAND, NEBULAE, FIRE, FIERY, FRUIT, COOL, MAGMA, GREEN, VIRIDIS, PLASMA, CIVIDIS, TERRAIN, NB_CLMODES };
51 enum SlideMode { REPLACE, SCROLL, FULLFRAME, RSCROLL, NB_SLIDES };
52 enum Orientation { VERTICAL, HORIZONTAL, NB_ORIENTATIONS };
53
54 typedef struct ShowSpectrumContext {
55 const AVClass *class;
56 int w, h;
57 char *rate_str;
58 AVRational auto_frame_rate;
59 AVRational frame_rate;
60 AVFrame *outpicref;
61 int nb_display_channels;
62 int orientation;
63 int channel_width;
64 int channel_height;
65 int sliding; ///< 1 if sliding mode, 0 otherwise
66 int mode; ///< channel display mode
67 int color_mode; ///< display color scheme
68 int scale;
69 int fscale;
70 float saturation; ///< color saturation multiplier
71 float rotation; ///< color rotation
72 int start, stop; ///< zoom mode
73 int data;
74 int xpos; ///< x position (current column)
75 FFTContext **fft; ///< Fast Fourier Transform context
76 FFTContext **ifft; ///< Inverse Fast Fourier Transform context
77 int fft_bits; ///< number of bits (FFT window size = 1<<fft_bits)
78 FFTComplex **fft_data; ///< bins holder for each (displayed) channels
79 FFTComplex **fft_scratch; ///< scratch buffers
80 float *window_func_lut; ///< Window function LUT
81 float **magnitudes;
82 float **phases;
83 int win_func;
84 int win_size;
85 int buf_size;
86 double win_scale;
87 float overlap;
88 float gain;
89 int consumed;
90 int hop_size;
91 float *combine_buffer; ///< color combining buffer (3 * h items)
92 float **color_buffer; ///< color buffer (3 * h * ch items)
93 AVAudioFifo *fifo;
94 int64_t pts;
95 int64_t old_pts;
96 int old_len;
97 int single_pic;
98 int legend;
99 int start_x, start_y;
100 int (*plot_channel)(AVFilterContext *ctx, void *arg, int jobnr, int nb_jobs);
101 } ShowSpectrumContext;
102
103 #define OFFSET(x) offsetof(ShowSpectrumContext, x)
104 #define FLAGS AV_OPT_FLAG_FILTERING_PARAM|AV_OPT_FLAG_VIDEO_PARAM
105
106 static const AVOption showspectrum_options[] = {
107 { "size", "set video size", OFFSET(w), AV_OPT_TYPE_IMAGE_SIZE, {.str = "640x512"}, 0, 0, FLAGS },
108 { "s", "set video size", OFFSET(w), AV_OPT_TYPE_IMAGE_SIZE, {.str = "640x512"}, 0, 0, FLAGS },
109 { "slide", "set sliding mode", OFFSET(sliding), AV_OPT_TYPE_INT, {.i64 = 0}, 0, NB_SLIDES-1, FLAGS, "slide" },
110 { "replace", "replace old columns with new", 0, AV_OPT_TYPE_CONST, {.i64=REPLACE}, 0, 0, FLAGS, "slide" },
111 { "scroll", "scroll from right to left", 0, AV_OPT_TYPE_CONST, {.i64=SCROLL}, 0, 0, FLAGS, "slide" },
112 { "fullframe", "return full frames", 0, AV_OPT_TYPE_CONST, {.i64=FULLFRAME}, 0, 0, FLAGS, "slide" },
113 { "rscroll", "scroll from left to right", 0, AV_OPT_TYPE_CONST, {.i64=RSCROLL}, 0, 0, FLAGS, "slide" },
114 { "mode", "set channel display mode", OFFSET(mode), AV_OPT_TYPE_INT, {.i64=COMBINED}, COMBINED, NB_MODES-1, FLAGS, "mode" },
115 { "combined", "combined mode", 0, AV_OPT_TYPE_CONST, {.i64=COMBINED}, 0, 0, FLAGS, "mode" },
116 { "separate", "separate mode", 0, AV_OPT_TYPE_CONST, {.i64=SEPARATE}, 0, 0, FLAGS, "mode" },
117 { "color", "set channel coloring", OFFSET(color_mode), AV_OPT_TYPE_INT, {.i64=CHANNEL}, CHANNEL, NB_CLMODES-1, FLAGS, "color" },
118 { "channel", "separate color for each channel", 0, AV_OPT_TYPE_CONST, {.i64=CHANNEL}, 0, 0, FLAGS, "color" },
119 { "intensity", "intensity based coloring", 0, AV_OPT_TYPE_CONST, {.i64=INTENSITY}, 0, 0, FLAGS, "color" },
120 { "rainbow", "rainbow based coloring", 0, AV_OPT_TYPE_CONST, {.i64=RAINBOW}, 0, 0, FLAGS, "color" },
121 { "moreland", "moreland based coloring", 0, AV_OPT_TYPE_CONST, {.i64=MORELAND}, 0, 0, FLAGS, "color" },
122 { "nebulae", "nebulae based coloring", 0, AV_OPT_TYPE_CONST, {.i64=NEBULAE}, 0, 0, FLAGS, "color" },
123 { "fire", "fire based coloring", 0, AV_OPT_TYPE_CONST, {.i64=FIRE}, 0, 0, FLAGS, "color" },
124 { "fiery", "fiery based coloring", 0, AV_OPT_TYPE_CONST, {.i64=FIERY}, 0, 0, FLAGS, "color" },
125 { "fruit", "fruit based coloring", 0, AV_OPT_TYPE_CONST, {.i64=FRUIT}, 0, 0, FLAGS, "color" },
126 { "cool", "cool based coloring", 0, AV_OPT_TYPE_CONST, {.i64=COOL}, 0, 0, FLAGS, "color" },
127 { "magma", "magma based coloring", 0, AV_OPT_TYPE_CONST, {.i64=MAGMA}, 0, 0, FLAGS, "color" },
128 { "green", "green based coloring", 0, AV_OPT_TYPE_CONST, {.i64=GREEN}, 0, 0, FLAGS, "color" },
129 { "viridis", "viridis based coloring", 0, AV_OPT_TYPE_CONST, {.i64=VIRIDIS}, 0, 0, FLAGS, "color" },
130 { "plasma", "plasma based coloring", 0, AV_OPT_TYPE_CONST, {.i64=PLASMA}, 0, 0, FLAGS, "color" },
131 { "cividis", "cividis based coloring", 0, AV_OPT_TYPE_CONST, {.i64=CIVIDIS}, 0, 0, FLAGS, "color" },
132 { "terrain", "terrain based coloring", 0, AV_OPT_TYPE_CONST, {.i64=TERRAIN}, 0, 0, FLAGS, "color" },
133 { "scale", "set display scale", OFFSET(scale), AV_OPT_TYPE_INT, {.i64=SQRT}, LINEAR, NB_SCALES-1, FLAGS, "scale" },
134 { "lin", "linear", 0, AV_OPT_TYPE_CONST, {.i64=LINEAR}, 0, 0, FLAGS, "scale" },
135 { "sqrt", "square root", 0, AV_OPT_TYPE_CONST, {.i64=SQRT}, 0, 0, FLAGS, "scale" },
136 { "cbrt", "cubic root", 0, AV_OPT_TYPE_CONST, {.i64=CBRT}, 0, 0, FLAGS, "scale" },
137 { "log", "logarithmic", 0, AV_OPT_TYPE_CONST, {.i64=LOG}, 0, 0, FLAGS, "scale" },
138 { "4thrt","4th root", 0, AV_OPT_TYPE_CONST, {.i64=FOURTHRT}, 0, 0, FLAGS, "scale" },
139 { "5thrt","5th root", 0, AV_OPT_TYPE_CONST, {.i64=FIFTHRT}, 0, 0, FLAGS, "scale" },
140 { "fscale", "set frequency scale", OFFSET(fscale), AV_OPT_TYPE_INT, {.i64=F_LINEAR}, 0, NB_FSCALES-1, FLAGS, "fscale" },
141 { "lin", "linear", 0, AV_OPT_TYPE_CONST, {.i64=F_LINEAR}, 0, 0, FLAGS, "fscale" },
142 { "log", "logarithmic", 0, AV_OPT_TYPE_CONST, {.i64=F_LOG}, 0, 0, FLAGS, "fscale" },
143 { "saturation", "color saturation multiplier", OFFSET(saturation), AV_OPT_TYPE_FLOAT, {.dbl = 1}, -10, 10, FLAGS },
144 { "win_func", "set window function", OFFSET(win_func), AV_OPT_TYPE_INT, {.i64 = WFUNC_HANNING}, 0, NB_WFUNC-1, FLAGS, "win_func" },
145 { "rect", "Rectangular", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_RECT}, 0, 0, FLAGS, "win_func" },
146 { "bartlett", "Bartlett", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_BARTLETT}, 0, 0, FLAGS, "win_func" },
147 { "hann", "Hann", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_HANNING}, 0, 0, FLAGS, "win_func" },
148 { "hanning", "Hanning", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_HANNING}, 0, 0, FLAGS, "win_func" },
149 { "hamming", "Hamming", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_HAMMING}, 0, 0, FLAGS, "win_func" },
150 { "blackman", "Blackman", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_BLACKMAN}, 0, 0, FLAGS, "win_func" },
151 { "welch", "Welch", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_WELCH}, 0, 0, FLAGS, "win_func" },
152 { "flattop", "Flat-top", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_FLATTOP}, 0, 0, FLAGS, "win_func" },
153 { "bharris", "Blackman-Harris", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_BHARRIS}, 0, 0, FLAGS, "win_func" },
154 { "bnuttall", "Blackman-Nuttall", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_BNUTTALL}, 0, 0, FLAGS, "win_func" },
155 { "bhann", "Bartlett-Hann", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_BHANN}, 0, 0, FLAGS, "win_func" },
156 { "sine", "Sine", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_SINE}, 0, 0, FLAGS, "win_func" },
157 { "nuttall", "Nuttall", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_NUTTALL}, 0, 0, FLAGS, "win_func" },
158 { "lanczos", "Lanczos", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_LANCZOS}, 0, 0, FLAGS, "win_func" },
159 { "gauss", "Gauss", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_GAUSS}, 0, 0, FLAGS, "win_func" },
160 { "tukey", "Tukey", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_TUKEY}, 0, 0, FLAGS, "win_func" },
161 { "dolph", "Dolph-Chebyshev", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_DOLPH}, 0, 0, FLAGS, "win_func" },
162 { "cauchy", "Cauchy", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_CAUCHY}, 0, 0, FLAGS, "win_func" },
163 { "parzen", "Parzen", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_PARZEN}, 0, 0, FLAGS, "win_func" },
164 { "poisson", "Poisson", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_POISSON}, 0, 0, FLAGS, "win_func" },
165 { "bohman", "Bohman", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_BOHMAN}, 0, 0, FLAGS, "win_func" },
166 { "orientation", "set orientation", OFFSET(orientation), AV_OPT_TYPE_INT, {.i64=VERTICAL}, 0, NB_ORIENTATIONS-1, FLAGS, "orientation" },
167 { "vertical", NULL, 0, AV_OPT_TYPE_CONST, {.i64=VERTICAL}, 0, 0, FLAGS, "orientation" },
168 { "horizontal", NULL, 0, AV_OPT_TYPE_CONST, {.i64=HORIZONTAL}, 0, 0, FLAGS, "orientation" },
169 { "overlap", "set window overlap", OFFSET(overlap), AV_OPT_TYPE_FLOAT, {.dbl = 0}, 0, 1, FLAGS },
170 { "gain", "set scale gain", OFFSET(gain), AV_OPT_TYPE_FLOAT, {.dbl = 1}, 0, 128, FLAGS },
171 { "data", "set data mode", OFFSET(data), AV_OPT_TYPE_INT, {.i64 = 0}, 0, NB_DMODES-1, FLAGS, "data" },
172 { "magnitude", NULL, 0, AV_OPT_TYPE_CONST, {.i64=D_MAGNITUDE}, 0, 0, FLAGS, "data" },
173 { "phase", NULL, 0, AV_OPT_TYPE_CONST, {.i64=D_PHASE}, 0, 0, FLAGS, "data" },
174 { "rotation", "color rotation", OFFSET(rotation), AV_OPT_TYPE_FLOAT, {.dbl = 0}, -1, 1, FLAGS },
175 { "start", "start frequency", OFFSET(start), AV_OPT_TYPE_INT, {.i64 = 0}, 0, INT32_MAX, FLAGS },
176 { "stop", "stop frequency", OFFSET(stop), AV_OPT_TYPE_INT, {.i64 = 0}, 0, INT32_MAX, FLAGS },
177 { "fps", "set video rate", OFFSET(rate_str), AV_OPT_TYPE_STRING, {.str = "auto"}, 0, 0, FLAGS },
178 { "legend", "draw legend", OFFSET(legend), AV_OPT_TYPE_BOOL, {.i64 = 0}, 0, 1, FLAGS },
179 { NULL }
180 };
181
182 AVFILTER_DEFINE_CLASS(showspectrum);
183
184 static const struct ColorTable {
185 float a, y, u, v;
186 } color_table[][8] = {
187 [INTENSITY] = {
188 { 0, 0, 0, 0 },
189 { 0.13, .03587126228984074, .1573300977624594, -.02548747583751842 },
190 { 0.30, .18572281794568020, .1772436246393981, .17475554840414750 },
191 { 0.60, .28184980583656130, -.1593064119945782, .47132074554608920 },
192 { 0.73, .65830621175547810, -.3716070802232764, .24352759331252930 },
193 { 0.78, .76318535758242900, -.4307467689263783, .16866496622310430 },
194 { 0.91, .95336363636363640, -.2045454545454546, .03313636363636363 },
195 { 1, 1, 0, 0 }},
196 [RAINBOW] = {
197 { 0, 0, 0, 0 },
198 { 0.13, 44/256., (189-128)/256., (138-128)/256. },
199 { 0.25, 29/256., (186-128)/256., (119-128)/256. },
200 { 0.38, 119/256., (194-128)/256., (53-128)/256. },
201 { 0.60, 111/256., (73-128)/256., (59-128)/256. },
202 { 0.73, 205/256., (19-128)/256., (149-128)/256. },
203 { 0.86, 135/256., (83-128)/256., (200-128)/256. },
204 { 1, 73/256., (95-128)/256., (225-128)/256. }},
205 [MORELAND] = {
206 { 0, 44/256., (181-128)/256., (112-128)/256. },
207 { 0.13, 126/256., (177-128)/256., (106-128)/256. },
208 { 0.25, 164/256., (163-128)/256., (109-128)/256. },
209 { 0.38, 200/256., (140-128)/256., (120-128)/256. },
210 { 0.60, 201/256., (117-128)/256., (141-128)/256. },
211 { 0.73, 177/256., (103-128)/256., (165-128)/256. },
212 { 0.86, 136/256., (100-128)/256., (183-128)/256. },
213 { 1, 68/256., (117-128)/256., (203-128)/256. }},
214 [NEBULAE] = {
215 { 0, 10/256., (134-128)/256., (132-128)/256. },
216 { 0.23, 21/256., (137-128)/256., (130-128)/256. },
217 { 0.45, 35/256., (134-128)/256., (134-128)/256. },
218 { 0.57, 51/256., (130-128)/256., (139-128)/256. },
219 { 0.67, 104/256., (116-128)/256., (162-128)/256. },
220 { 0.77, 120/256., (105-128)/256., (188-128)/256. },
221 { 0.87, 140/256., (105-128)/256., (188-128)/256. },
222 { 1, 1, 0, 0 }},
223 [FIRE] = {
224 { 0, 0, 0, 0 },
225 { 0.23, 44/256., (132-128)/256., (127-128)/256. },
226 { 0.45, 62/256., (116-128)/256., (140-128)/256. },
227 { 0.57, 75/256., (105-128)/256., (152-128)/256. },
228 { 0.67, 95/256., (91-128)/256., (166-128)/256. },
229 { 0.77, 126/256., (74-128)/256., (172-128)/256. },
230 { 0.87, 164/256., (73-128)/256., (162-128)/256. },
231 { 1, 1, 0, 0 }},
232 [FIERY] = {
233 { 0, 0, 0, 0 },
234 { 0.23, 36/256., (116-128)/256., (163-128)/256. },
235 { 0.45, 52/256., (102-128)/256., (200-128)/256. },
236 { 0.57, 116/256., (84-128)/256., (196-128)/256. },
237 { 0.67, 157/256., (67-128)/256., (181-128)/256. },
238 { 0.77, 193/256., (40-128)/256., (155-128)/256. },
239 { 0.87, 221/256., (101-128)/256., (134-128)/256. },
240 { 1, 1, 0, 0 }},
241 [FRUIT] = {
242 { 0, 0, 0, 0 },
243 { 0.20, 29/256., (136-128)/256., (119-128)/256. },
244 { 0.30, 60/256., (119-128)/256., (90-128)/256. },
245 { 0.40, 85/256., (91-128)/256., (85-128)/256. },
246 { 0.50, 116/256., (70-128)/256., (105-128)/256. },
247 { 0.60, 151/256., (50-128)/256., (146-128)/256. },
248 { 0.70, 191/256., (63-128)/256., (178-128)/256. },
249 { 1, 98/256., (80-128)/256., (221-128)/256. }},
250 [COOL] = {
251 { 0, 0, 0, 0 },
252 { .15, 0, .5, -.5 },
253 { 1, 1, -.5, .5 }},
254 [MAGMA] = {
255 { 0, 0, 0, 0 },
256 { 0.10, 23/256., (175-128)/256., (120-128)/256. },
257 { 0.23, 43/256., (158-128)/256., (144-128)/256. },
258 { 0.35, 85/256., (138-128)/256., (179-128)/256. },
259 { 0.48, 96/256., (128-128)/256., (189-128)/256. },
260 { 0.64, 128/256., (103-128)/256., (214-128)/256. },
261 { 0.92, 205/256., (80-128)/256., (152-128)/256. },
262 { 1, 1, 0, 0 }},
263 [GREEN] = {
264 { 0, 0, 0, 0 },
265 { .75, .5, 0, -.5 },
266 { 1, 1, 0, 0 }},
267 [VIRIDIS] = {
268 { 0, 0, 0, 0 },
269 { 0.10, 0x39/255., (0x9D -128)/255., (0x8F -128)/255. },
270 { 0.23, 0x5C/255., (0x9A -128)/255., (0x68 -128)/255. },
271 { 0.35, 0x69/255., (0x93 -128)/255., (0x57 -128)/255. },
272 { 0.48, 0x76/255., (0x88 -128)/255., (0x4B -128)/255. },
273 { 0.64, 0x8A/255., (0x72 -128)/255., (0x4F -128)/255. },
274 { 0.80, 0xA3/255., (0x50 -128)/255., (0x66 -128)/255. },
275 { 1, 0xCC/255., (0x2F -128)/255., (0x87 -128)/255. }},
276 [PLASMA] = {
277 { 0, 0, 0, 0 },
278 { 0.10, 0x27/255., (0xC2 -128)/255., (0x82 -128)/255. },
279 { 0.58, 0x5B/255., (0x9A -128)/255., (0xAE -128)/255. },
280 { 0.70, 0x89/255., (0x44 -128)/255., (0xAB -128)/255. },
281 { 0.80, 0xB4/255., (0x2B -128)/255., (0x9E -128)/255. },
282 { 0.91, 0xD2/255., (0x38 -128)/255., (0x92 -128)/255. },
283 { 1, 1, 0, 0. }},
284 [CIVIDIS] = {
285 { 0, 0, 0, 0 },
286 { 0.20, 0x28/255., (0x98 -128)/255., (0x6F -128)/255. },
287 { 0.50, 0x48/255., (0x95 -128)/255., (0x74 -128)/255. },
288 { 0.63, 0x69/255., (0x84 -128)/255., (0x7F -128)/255. },
289 { 0.76, 0x89/255., (0x75 -128)/255., (0x84 -128)/255. },
290 { 0.90, 0xCE/255., (0x35 -128)/255., (0x95 -128)/255. },
291 { 1, 1, 0, 0. }},
292 [TERRAIN] = {
293 { 0, 0, 0, 0 },
294 { 0.15, 0, .5, 0 },
295 { 0.60, 1, -.5, -.5 },
296 { 0.85, 1, -.5, .5 },
297 { 1, 1, 0, 0 }},
298 };
299
uninit(AVFilterContext * ctx)300 static av_cold void uninit(AVFilterContext *ctx)
301 {
302 ShowSpectrumContext *s = ctx->priv;
303 int i;
304
305 av_freep(&s->combine_buffer);
306 if (s->fft) {
307 for (i = 0; i < s->nb_display_channels; i++)
308 av_fft_end(s->fft[i]);
309 }
310 av_freep(&s->fft);
311 if (s->ifft) {
312 for (i = 0; i < s->nb_display_channels; i++)
313 av_fft_end(s->ifft[i]);
314 }
315 av_freep(&s->ifft);
316 if (s->fft_data) {
317 for (i = 0; i < s->nb_display_channels; i++)
318 av_freep(&s->fft_data[i]);
319 }
320 av_freep(&s->fft_data);
321 if (s->fft_scratch) {
322 for (i = 0; i < s->nb_display_channels; i++)
323 av_freep(&s->fft_scratch[i]);
324 }
325 av_freep(&s->fft_scratch);
326 if (s->color_buffer) {
327 for (i = 0; i < s->nb_display_channels; i++)
328 av_freep(&s->color_buffer[i]);
329 }
330 av_freep(&s->color_buffer);
331 av_freep(&s->window_func_lut);
332 if (s->magnitudes) {
333 for (i = 0; i < s->nb_display_channels; i++)
334 av_freep(&s->magnitudes[i]);
335 }
336 av_freep(&s->magnitudes);
337 av_frame_free(&s->outpicref);
338 av_audio_fifo_free(s->fifo);
339 if (s->phases) {
340 for (i = 0; i < s->nb_display_channels; i++)
341 av_freep(&s->phases[i]);
342 }
343 av_freep(&s->phases);
344 }
345
query_formats(AVFilterContext * ctx)346 static int query_formats(AVFilterContext *ctx)
347 {
348 AVFilterFormats *formats = NULL;
349 AVFilterChannelLayouts *layouts = NULL;
350 AVFilterLink *inlink = ctx->inputs[0];
351 AVFilterLink *outlink = ctx->outputs[0];
352 static const enum AVSampleFormat sample_fmts[] = { AV_SAMPLE_FMT_FLTP, AV_SAMPLE_FMT_NONE };
353 static const enum AVPixelFormat pix_fmts[] = { AV_PIX_FMT_YUV444P, AV_PIX_FMT_YUVJ444P, AV_PIX_FMT_NONE };
354 int ret;
355
356 /* set input audio formats */
357 formats = ff_make_format_list(sample_fmts);
358 if ((ret = ff_formats_ref(formats, &inlink->outcfg.formats)) < 0)
359 return ret;
360
361 layouts = ff_all_channel_layouts();
362 if ((ret = ff_channel_layouts_ref(layouts, &inlink->outcfg.channel_layouts)) < 0)
363 return ret;
364
365 formats = ff_all_samplerates();
366 if ((ret = ff_formats_ref(formats, &inlink->outcfg.samplerates)) < 0)
367 return ret;
368
369 /* set output video format */
370 formats = ff_make_format_list(pix_fmts);
371 if ((ret = ff_formats_ref(formats, &outlink->incfg.formats)) < 0)
372 return ret;
373
374 return 0;
375 }
376
run_channel_fft(AVFilterContext * ctx,void * arg,int jobnr,int nb_jobs)377 static int run_channel_fft(AVFilterContext *ctx, void *arg, int jobnr, int nb_jobs)
378 {
379 ShowSpectrumContext *s = ctx->priv;
380 AVFilterLink *inlink = ctx->inputs[0];
381 const float *window_func_lut = s->window_func_lut;
382 AVFrame *fin = arg;
383 const int ch = jobnr;
384 int n;
385
386 /* fill FFT input with the number of samples available */
387 const float *p = (float *)fin->extended_data[ch];
388
389 for (n = 0; n < s->win_size; n++) {
390 s->fft_data[ch][n].re = p[n] * window_func_lut[n];
391 s->fft_data[ch][n].im = 0;
392 }
393
394 if (s->stop) {
395 float theta, phi, psi, a, b, S, c;
396 FFTComplex *g = s->fft_data[ch];
397 FFTComplex *h = s->fft_scratch[ch];
398 int L = s->buf_size;
399 int N = s->win_size;
400 int M = s->win_size / 2;
401
402 phi = 2.f * M_PI * (s->stop - s->start) / (float)inlink->sample_rate / (M - 1);
403 theta = 2.f * M_PI * s->start / (float)inlink->sample_rate;
404
405 for (int n = 0; n < M; n++) {
406 h[n].re = cosf(n * n / 2.f * phi);
407 h[n].im = sinf(n * n / 2.f * phi);
408 }
409
410 for (int n = M; n < L; n++) {
411 h[n].re = 0.f;
412 h[n].im = 0.f;
413 }
414
415 for (int n = L - N; n < L; n++) {
416 h[n].re = cosf((L - n) * (L - n) / 2.f * phi);
417 h[n].im = sinf((L - n) * (L - n) / 2.f * phi);
418 }
419
420 for (int n = 0; n < N; n++) {
421 g[n].re = s->fft_data[ch][n].re;
422 g[n].im = s->fft_data[ch][n].im;
423 }
424
425 for (int n = N; n < L; n++) {
426 g[n].re = 0.f;
427 g[n].im = 0.f;
428 }
429
430 for (int n = 0; n < N; n++) {
431 psi = n * theta + n * n / 2.f * phi;
432 c = cosf(psi);
433 S = -sinf(psi);
434 a = c * g[n].re - S * g[n].im;
435 b = S * g[n].re + c * g[n].im;
436 g[n].re = a;
437 g[n].im = b;
438 }
439
440 av_fft_permute(s->fft[ch], h);
441 av_fft_calc(s->fft[ch], h);
442
443 av_fft_permute(s->fft[ch], g);
444 av_fft_calc(s->fft[ch], g);
445
446 for (int n = 0; n < L; n++) {
447 c = g[n].re;
448 S = g[n].im;
449 a = c * h[n].re - S * h[n].im;
450 b = S * h[n].re + c * h[n].im;
451
452 g[n].re = a / L;
453 g[n].im = b / L;
454 }
455
456 av_fft_permute(s->ifft[ch], g);
457 av_fft_calc(s->ifft[ch], g);
458
459 for (int k = 0; k < M; k++) {
460 psi = k * k / 2.f * phi;
461 c = cosf(psi);
462 S = -sinf(psi);
463 a = c * g[k].re - S * g[k].im;
464 b = S * g[k].re + c * g[k].im;
465 s->fft_data[ch][k].re = a;
466 s->fft_data[ch][k].im = b;
467 }
468 } else {
469 /* run FFT on each samples set */
470 av_fft_permute(s->fft[ch], s->fft_data[ch]);
471 av_fft_calc(s->fft[ch], s->fft_data[ch]);
472 }
473
474 return 0;
475 }
476
drawtext(AVFrame * pic,int x,int y,const char * txt,int o)477 static void drawtext(AVFrame *pic, int x, int y, const char *txt, int o)
478 {
479 const uint8_t *font;
480 int font_height;
481 int i;
482
483 font = avpriv_cga_font, font_height = 8;
484
485 for (i = 0; txt[i]; i++) {
486 int char_y, mask;
487
488 if (o) {
489 for (char_y = font_height - 1; char_y >= 0; char_y--) {
490 uint8_t *p = pic->data[0] + (y + i * 10) * pic->linesize[0] + x;
491 for (mask = 0x80; mask; mask >>= 1) {
492 if (font[txt[i] * font_height + font_height - 1 - char_y] & mask)
493 p[char_y] = ~p[char_y];
494 p += pic->linesize[0];
495 }
496 }
497 } else {
498 uint8_t *p = pic->data[0] + y*pic->linesize[0] + (x + i*8);
499 for (char_y = 0; char_y < font_height; char_y++) {
500 for (mask = 0x80; mask; mask >>= 1) {
501 if (font[txt[i] * font_height + char_y] & mask)
502 *p = ~(*p);
503 p++;
504 }
505 p += pic->linesize[0] - 8;
506 }
507 }
508 }
509 }
510
color_range(ShowSpectrumContext * s,int ch,float * yf,float * uf,float * vf)511 static void color_range(ShowSpectrumContext *s, int ch,
512 float *yf, float *uf, float *vf)
513 {
514 switch (s->mode) {
515 case COMBINED:
516 // reduce range by channel count
517 *yf = 256.0f / s->nb_display_channels;
518 switch (s->color_mode) {
519 case RAINBOW:
520 case MORELAND:
521 case NEBULAE:
522 case FIRE:
523 case FIERY:
524 case FRUIT:
525 case COOL:
526 case GREEN:
527 case VIRIDIS:
528 case PLASMA:
529 case CIVIDIS:
530 case TERRAIN:
531 case MAGMA:
532 case INTENSITY:
533 *uf = *yf;
534 *vf = *yf;
535 break;
536 case CHANNEL:
537 /* adjust saturation for mixed UV coloring */
538 /* this factor is correct for infinite channels, an approximation otherwise */
539 *uf = *yf * M_PI;
540 *vf = *yf * M_PI;
541 break;
542 default:
543 av_assert0(0);
544 }
545 break;
546 case SEPARATE:
547 // full range
548 *yf = 256.0f;
549 *uf = 256.0f;
550 *vf = 256.0f;
551 break;
552 default:
553 av_assert0(0);
554 }
555
556 if (s->color_mode == CHANNEL) {
557 if (s->nb_display_channels > 1) {
558 *uf *= 0.5f * sinf((2 * M_PI * ch) / s->nb_display_channels + M_PI * s->rotation);
559 *vf *= 0.5f * cosf((2 * M_PI * ch) / s->nb_display_channels + M_PI * s->rotation);
560 } else {
561 *uf *= 0.5f * sinf(M_PI * s->rotation);
562 *vf *= 0.5f * cosf(M_PI * s->rotation + M_PI_2);
563 }
564 } else {
565 *uf += *uf * sinf(M_PI * s->rotation);
566 *vf += *vf * cosf(M_PI * s->rotation + M_PI_2);
567 }
568
569 *uf *= s->saturation;
570 *vf *= s->saturation;
571 }
572
pick_color(ShowSpectrumContext * s,float yf,float uf,float vf,float a,float * out)573 static void pick_color(ShowSpectrumContext *s,
574 float yf, float uf, float vf,
575 float a, float *out)
576 {
577 if (s->color_mode > CHANNEL) {
578 const int cm = s->color_mode;
579 float y, u, v;
580 int i;
581
582 for (i = 1; i < FF_ARRAY_ELEMS(color_table[cm]) - 1; i++)
583 if (color_table[cm][i].a >= a)
584 break;
585 // i now is the first item >= the color
586 // now we know to interpolate between item i - 1 and i
587 if (a <= color_table[cm][i - 1].a) {
588 y = color_table[cm][i - 1].y;
589 u = color_table[cm][i - 1].u;
590 v = color_table[cm][i - 1].v;
591 } else if (a >= color_table[cm][i].a) {
592 y = color_table[cm][i].y;
593 u = color_table[cm][i].u;
594 v = color_table[cm][i].v;
595 } else {
596 float start = color_table[cm][i - 1].a;
597 float end = color_table[cm][i].a;
598 float lerpfrac = (a - start) / (end - start);
599 y = color_table[cm][i - 1].y * (1.0f - lerpfrac)
600 + color_table[cm][i].y * lerpfrac;
601 u = color_table[cm][i - 1].u * (1.0f - lerpfrac)
602 + color_table[cm][i].u * lerpfrac;
603 v = color_table[cm][i - 1].v * (1.0f - lerpfrac)
604 + color_table[cm][i].v * lerpfrac;
605 }
606
607 out[0] = y * yf;
608 out[1] = u * uf;
609 out[2] = v * vf;
610 } else {
611 out[0] = a * yf;
612 out[1] = a * uf;
613 out[2] = a * vf;
614 }
615 }
616
get_time(AVFilterContext * ctx,float seconds,int x)617 static char *get_time(AVFilterContext *ctx, float seconds, int x)
618 {
619 char *units;
620
621 if (x == 0)
622 units = av_asprintf("0");
623 else if (log10(seconds) > 6)
624 units = av_asprintf("%.2fh", seconds / (60 * 60));
625 else if (log10(seconds) > 3)
626 units = av_asprintf("%.2fm", seconds / 60);
627 else
628 units = av_asprintf("%.2fs", seconds);
629 return units;
630 }
631
log_scale(const float value,const float min,const float max)632 static float log_scale(const float value, const float min, const float max)
633 {
634 if (value < min)
635 return min;
636 if (value > max)
637 return max;
638
639 {
640 const float b = logf(max / min) / (max - min);
641 const float a = max / expf(max * b);
642
643 return expf(value * b) * a;
644 }
645 }
646
get_log_hz(const int bin,const int num_bins,const float sample_rate)647 static float get_log_hz(const int bin, const int num_bins, const float sample_rate)
648 {
649 const float max_freq = sample_rate / 2;
650 const float hz_per_bin = max_freq / num_bins;
651 const float freq = hz_per_bin * bin;
652 const float scaled_freq = log_scale(freq + 1, 21, max_freq) - 1;
653
654 return num_bins * scaled_freq / max_freq;
655 }
656
inv_log_scale(const float value,const float min,const float max)657 static float inv_log_scale(const float value, const float min, const float max)
658 {
659 if (value < min)
660 return min;
661 if (value > max)
662 return max;
663
664 {
665 const float b = logf(max / min) / (max - min);
666 const float a = max / expf(max * b);
667
668 return logf(value / a) / b;
669 }
670 }
671
bin_pos(const int bin,const int num_bins,const float sample_rate)672 static float bin_pos(const int bin, const int num_bins, const float sample_rate)
673 {
674 const float max_freq = sample_rate / 2;
675 const float hz_per_bin = max_freq / num_bins;
676 const float freq = hz_per_bin * bin;
677 const float scaled_freq = inv_log_scale(freq + 1, 21, max_freq) - 1;
678
679 return num_bins * scaled_freq / max_freq;
680 }
681
draw_legend(AVFilterContext * ctx,int samples)682 static int draw_legend(AVFilterContext *ctx, int samples)
683 {
684 ShowSpectrumContext *s = ctx->priv;
685 AVFilterLink *inlink = ctx->inputs[0];
686 AVFilterLink *outlink = ctx->outputs[0];
687 int ch, y, x = 0, sz = s->orientation == VERTICAL ? s->w : s->h;
688 int multi = (s->mode == SEPARATE && s->color_mode == CHANNEL);
689 float spp = samples / (float)sz;
690 char *text;
691 uint8_t *dst;
692 char chlayout_str[128];
693
694 av_get_channel_layout_string(chlayout_str, sizeof(chlayout_str), inlink->channels,
695 inlink->channel_layout);
696
697 text = av_asprintf("%d Hz | %s", inlink->sample_rate, chlayout_str);
698 if (!text)
699 return AVERROR(ENOMEM);
700
701 drawtext(s->outpicref, 2, outlink->h - 10, "CREATED BY LIBAVFILTER", 0);
702 drawtext(s->outpicref, outlink->w - 2 - strlen(text) * 10, outlink->h - 10, text, 0);
703 av_freep(&text);
704 if (s->stop) {
705 text = av_asprintf("Zoom: %d Hz - %d Hz", s->start, s->stop);
706 if (!text)
707 return AVERROR(ENOMEM);
708 drawtext(s->outpicref, outlink->w - 2 - strlen(text) * 10, 3, text, 0);
709 av_freep(&text);
710 }
711
712 dst = s->outpicref->data[0] + (s->start_y - 1) * s->outpicref->linesize[0] + s->start_x - 1;
713 for (x = 0; x < s->w + 1; x++)
714 dst[x] = 200;
715 dst = s->outpicref->data[0] + (s->start_y + s->h) * s->outpicref->linesize[0] + s->start_x - 1;
716 for (x = 0; x < s->w + 1; x++)
717 dst[x] = 200;
718 for (y = 0; y < s->h + 2; y++) {
719 dst = s->outpicref->data[0] + (y + s->start_y - 1) * s->outpicref->linesize[0];
720 dst[s->start_x - 1] = 200;
721 dst[s->start_x + s->w] = 200;
722 }
723 if (s->orientation == VERTICAL) {
724 int h = s->mode == SEPARATE ? s->h / s->nb_display_channels : s->h;
725 int hh = s->mode == SEPARATE ? -(s->h % s->nb_display_channels) + 1 : 1;
726 for (ch = 0; ch < (s->mode == SEPARATE ? s->nb_display_channels : 1); ch++) {
727 for (y = 0; y < h; y += 20) {
728 dst = s->outpicref->data[0] + (s->start_y + h * (ch + 1) - y - hh) * s->outpicref->linesize[0];
729 dst[s->start_x - 2] = 200;
730 dst[s->start_x + s->w + 1] = 200;
731 }
732 for (y = 0; y < h; y += 40) {
733 dst = s->outpicref->data[0] + (s->start_y + h * (ch + 1) - y - hh) * s->outpicref->linesize[0];
734 dst[s->start_x - 3] = 200;
735 dst[s->start_x + s->w + 2] = 200;
736 }
737 dst = s->outpicref->data[0] + (s->start_y - 2) * s->outpicref->linesize[0] + s->start_x;
738 for (x = 0; x < s->w; x+=40)
739 dst[x] = 200;
740 dst = s->outpicref->data[0] + (s->start_y - 3) * s->outpicref->linesize[0] + s->start_x;
741 for (x = 0; x < s->w; x+=80)
742 dst[x] = 200;
743 dst = s->outpicref->data[0] + (s->h + s->start_y + 1) * s->outpicref->linesize[0] + s->start_x;
744 for (x = 0; x < s->w; x+=40) {
745 dst[x] = 200;
746 }
747 dst = s->outpicref->data[0] + (s->h + s->start_y + 2) * s->outpicref->linesize[0] + s->start_x;
748 for (x = 0; x < s->w; x+=80) {
749 dst[x] = 200;
750 }
751 for (y = 0; y < h; y += 40) {
752 float range = s->stop ? s->stop - s->start : inlink->sample_rate / 2;
753 float bin = s->fscale == F_LINEAR ? y : get_log_hz(y, h, inlink->sample_rate);
754 float hertz = s->start + bin * range / (float)(1 << (int)ceil(log2(h)));
755 char *units;
756
757 if (hertz == 0)
758 units = av_asprintf("DC");
759 else
760 units = av_asprintf("%.2f", hertz);
761 if (!units)
762 return AVERROR(ENOMEM);
763
764 drawtext(s->outpicref, s->start_x - 8 * strlen(units) - 4, h * (ch + 1) + s->start_y - y - 4 - hh, units, 0);
765 av_free(units);
766 }
767 }
768
769 for (x = 0; x < s->w && s->single_pic; x+=80) {
770 float seconds = x * spp / inlink->sample_rate;
771 char *units = get_time(ctx, seconds, x);
772 if (!units)
773 return AVERROR(ENOMEM);
774
775 drawtext(s->outpicref, s->start_x + x - 4 * strlen(units), s->h + s->start_y + 6, units, 0);
776 drawtext(s->outpicref, s->start_x + x - 4 * strlen(units), s->start_y - 12, units, 0);
777 av_free(units);
778 }
779
780 drawtext(s->outpicref, outlink->w / 2 - 4 * 4, outlink->h - s->start_y / 2, "TIME", 0);
781 drawtext(s->outpicref, s->start_x / 7, outlink->h / 2 - 14 * 4, "FREQUENCY (Hz)", 1);
782 } else {
783 int w = s->mode == SEPARATE ? s->w / s->nb_display_channels : s->w;
784 for (y = 0; y < s->h; y += 20) {
785 dst = s->outpicref->data[0] + (s->start_y + y) * s->outpicref->linesize[0];
786 dst[s->start_x - 2] = 200;
787 dst[s->start_x + s->w + 1] = 200;
788 }
789 for (y = 0; y < s->h; y += 40) {
790 dst = s->outpicref->data[0] + (s->start_y + y) * s->outpicref->linesize[0];
791 dst[s->start_x - 3] = 200;
792 dst[s->start_x + s->w + 2] = 200;
793 }
794 for (ch = 0; ch < (s->mode == SEPARATE ? s->nb_display_channels : 1); ch++) {
795 dst = s->outpicref->data[0] + (s->start_y - 2) * s->outpicref->linesize[0] + s->start_x + w * ch;
796 for (x = 0; x < w; x+=40)
797 dst[x] = 200;
798 dst = s->outpicref->data[0] + (s->start_y - 3) * s->outpicref->linesize[0] + s->start_x + w * ch;
799 for (x = 0; x < w; x+=80)
800 dst[x] = 200;
801 dst = s->outpicref->data[0] + (s->h + s->start_y + 1) * s->outpicref->linesize[0] + s->start_x + w * ch;
802 for (x = 0; x < w; x+=40) {
803 dst[x] = 200;
804 }
805 dst = s->outpicref->data[0] + (s->h + s->start_y + 2) * s->outpicref->linesize[0] + s->start_x + w * ch;
806 for (x = 0; x < w; x+=80) {
807 dst[x] = 200;
808 }
809 for (x = 0; x < w - 79; x += 80) {
810 float range = s->stop ? s->stop - s->start : inlink->sample_rate / 2;
811 float bin = s->fscale == F_LINEAR ? x : get_log_hz(x, w, inlink->sample_rate);
812 float hertz = s->start + bin * range / (float)(1 << (int)ceil(log2(w)));
813 char *units;
814
815 if (hertz == 0)
816 units = av_asprintf("DC");
817 else
818 units = av_asprintf("%.2f", hertz);
819 if (!units)
820 return AVERROR(ENOMEM);
821
822 drawtext(s->outpicref, s->start_x - 4 * strlen(units) + x + w * ch, s->start_y - 12, units, 0);
823 drawtext(s->outpicref, s->start_x - 4 * strlen(units) + x + w * ch, s->h + s->start_y + 6, units, 0);
824 av_free(units);
825 }
826 }
827 for (y = 0; y < s->h && s->single_pic; y+=40) {
828 float seconds = y * spp / inlink->sample_rate;
829 char *units = get_time(ctx, seconds, x);
830 if (!units)
831 return AVERROR(ENOMEM);
832
833 drawtext(s->outpicref, s->start_x - 8 * strlen(units) - 4, s->start_y + y - 4, units, 0);
834 av_free(units);
835 }
836 drawtext(s->outpicref, s->start_x / 7, outlink->h / 2 - 4 * 4, "TIME", 1);
837 drawtext(s->outpicref, outlink->w / 2 - 14 * 4, outlink->h - s->start_y / 2, "FREQUENCY (Hz)", 0);
838 }
839
840 for (ch = 0; ch < (multi ? s->nb_display_channels : 1); ch++) {
841 int h = multi ? s->h / s->nb_display_channels : s->h;
842
843 for (y = 0; y < h; y++) {
844 float out[3] = { 0., 127.5, 127.5};
845 int chn;
846
847 for (chn = 0; chn < (s->mode == SEPARATE ? 1 : s->nb_display_channels); chn++) {
848 float yf, uf, vf;
849 int channel = (multi) ? s->nb_display_channels - ch - 1 : chn;
850 float lout[3];
851
852 color_range(s, channel, &yf, &uf, &vf);
853 pick_color(s, yf, uf, vf, y / (float)h, lout);
854 out[0] += lout[0];
855 out[1] += lout[1];
856 out[2] += lout[2];
857 }
858 memset(s->outpicref->data[0]+(s->start_y + h * (ch + 1) - y - 1) * s->outpicref->linesize[0] + s->w + s->start_x + 20, av_clip_uint8(out[0]), 10);
859 memset(s->outpicref->data[1]+(s->start_y + h * (ch + 1) - y - 1) * s->outpicref->linesize[1] + s->w + s->start_x + 20, av_clip_uint8(out[1]), 10);
860 memset(s->outpicref->data[2]+(s->start_y + h * (ch + 1) - y - 1) * s->outpicref->linesize[2] + s->w + s->start_x + 20, av_clip_uint8(out[2]), 10);
861 }
862
863 for (y = 0; ch == 0 && y < h; y += h / 10) {
864 float value = 120.f * log10f(1.f - y / (float)h);
865 char *text;
866
867 if (value < -120)
868 break;
869 text = av_asprintf("%.0f dB", value);
870 if (!text)
871 continue;
872 drawtext(s->outpicref, s->w + s->start_x + 35, s->start_y + y - 5, text, 0);
873 av_free(text);
874 }
875 }
876
877 return 0;
878 }
879
get_value(AVFilterContext * ctx,int ch,int y)880 static float get_value(AVFilterContext *ctx, int ch, int y)
881 {
882 ShowSpectrumContext *s = ctx->priv;
883 float *magnitudes = s->magnitudes[ch];
884 float *phases = s->phases[ch];
885 float a;
886
887 switch (s->data) {
888 case D_MAGNITUDE:
889 /* get magnitude */
890 a = magnitudes[y];
891 break;
892 case D_PHASE:
893 /* get phase */
894 a = phases[y];
895 break;
896 default:
897 av_assert0(0);
898 }
899
900 /* apply scale */
901 switch (s->scale) {
902 case LINEAR:
903 a = av_clipf(a, 0, 1);
904 break;
905 case SQRT:
906 a = av_clipf(sqrtf(a), 0, 1);
907 break;
908 case CBRT:
909 a = av_clipf(cbrtf(a), 0, 1);
910 break;
911 case FOURTHRT:
912 a = av_clipf(sqrtf(sqrtf(a)), 0, 1);
913 break;
914 case FIFTHRT:
915 a = av_clipf(powf(a, 0.20), 0, 1);
916 break;
917 case LOG:
918 a = 1.f + log10f(av_clipf(a, 1e-6, 1)) / 6.f; // zero = -120dBFS
919 break;
920 default:
921 av_assert0(0);
922 }
923
924 return a;
925 }
926
plot_channel_lin(AVFilterContext * ctx,void * arg,int jobnr,int nb_jobs)927 static int plot_channel_lin(AVFilterContext *ctx, void *arg, int jobnr, int nb_jobs)
928 {
929 ShowSpectrumContext *s = ctx->priv;
930 const int h = s->orientation == VERTICAL ? s->channel_height : s->channel_width;
931 const int ch = jobnr;
932 float yf, uf, vf;
933 int y;
934
935 /* decide color range */
936 color_range(s, ch, &yf, &uf, &vf);
937
938 /* draw the channel */
939 for (y = 0; y < h; y++) {
940 int row = (s->mode == COMBINED) ? y : ch * h + y;
941 float *out = &s->color_buffer[ch][3 * row];
942 float a = get_value(ctx, ch, y);
943
944 pick_color(s, yf, uf, vf, a, out);
945 }
946
947 return 0;
948 }
949
plot_channel_log(AVFilterContext * ctx,void * arg,int jobnr,int nb_jobs)950 static int plot_channel_log(AVFilterContext *ctx, void *arg, int jobnr, int nb_jobs)
951 {
952 ShowSpectrumContext *s = ctx->priv;
953 AVFilterLink *inlink = ctx->inputs[0];
954 const int h = s->orientation == VERTICAL ? s->channel_height : s->channel_width;
955 const int ch = jobnr;
956 float y, yf, uf, vf;
957 int yy = 0;
958
959 /* decide color range */
960 color_range(s, ch, &yf, &uf, &vf);
961
962 /* draw the channel */
963 for (y = 0; y < h && yy < h; yy++) {
964 float pos0 = bin_pos(yy+0, h, inlink->sample_rate);
965 float pos1 = bin_pos(yy+1, h, inlink->sample_rate);
966 float delta = pos1 - pos0;
967 float a0, a1;
968
969 a0 = get_value(ctx, ch, yy+0);
970 a1 = get_value(ctx, ch, FFMIN(yy+1, h-1));
971 for (float j = pos0; j < pos1 && y + j - pos0 < h; j++) {
972 float row = (s->mode == COMBINED) ? y + j - pos0 : ch * h + y + j - pos0;
973 float *out = &s->color_buffer[ch][3 * FFMIN(lrintf(row), h-1)];
974 float lerpfrac = (j - pos0) / delta;
975
976 pick_color(s, yf, uf, vf, lerpfrac * a1 + (1.f-lerpfrac) * a0, out);
977 }
978 y += delta;
979 }
980
981 return 0;
982 }
983
config_output(AVFilterLink * outlink)984 static int config_output(AVFilterLink *outlink)
985 {
986 AVFilterContext *ctx = outlink->src;
987 AVFilterLink *inlink = ctx->inputs[0];
988 ShowSpectrumContext *s = ctx->priv;
989 int i, fft_bits, h, w;
990 float overlap;
991
992 switch (s->fscale) {
993 case F_LINEAR: s->plot_channel = plot_channel_lin; break;
994 case F_LOG: s->plot_channel = plot_channel_log; break;
995 default: return AVERROR_BUG;
996 }
997
998 s->stop = FFMIN(s->stop, inlink->sample_rate / 2);
999 if (s->stop && s->stop <= s->start) {
1000 av_log(ctx, AV_LOG_ERROR, "Stop frequency should be greater than start.\n");
1001 return AVERROR(EINVAL);
1002 }
1003
1004 if (!strcmp(ctx->filter->name, "showspectrumpic"))
1005 s->single_pic = 1;
1006
1007 outlink->w = s->w;
1008 outlink->h = s->h;
1009 outlink->sample_aspect_ratio = (AVRational){1,1};
1010
1011 if (s->legend) {
1012 s->start_x = (log10(inlink->sample_rate) + 1) * 25;
1013 s->start_y = 64;
1014 outlink->w += s->start_x * 2;
1015 outlink->h += s->start_y * 2;
1016 }
1017
1018 h = (s->mode == COMBINED || s->orientation == HORIZONTAL) ? s->h : s->h / inlink->channels;
1019 w = (s->mode == COMBINED || s->orientation == VERTICAL) ? s->w : s->w / inlink->channels;
1020 s->channel_height = h;
1021 s->channel_width = w;
1022
1023 if (s->orientation == VERTICAL) {
1024 /* FFT window size (precision) according to the requested output frame height */
1025 for (fft_bits = 1; 1 << fft_bits < 2 * h; fft_bits++);
1026 } else {
1027 /* FFT window size (precision) according to the requested output frame width */
1028 for (fft_bits = 1; 1 << fft_bits < 2 * w; fft_bits++);
1029 }
1030
1031 s->win_size = 1 << fft_bits;
1032 s->buf_size = s->win_size << !!s->stop;
1033
1034 if (!s->fft) {
1035 s->fft = av_calloc(inlink->channels, sizeof(*s->fft));
1036 if (!s->fft)
1037 return AVERROR(ENOMEM);
1038 }
1039
1040 if (s->stop) {
1041 if (!s->ifft) {
1042 s->ifft = av_calloc(inlink->channels, sizeof(*s->ifft));
1043 if (!s->ifft)
1044 return AVERROR(ENOMEM);
1045 }
1046 }
1047
1048 /* (re-)configuration if the video output changed (or first init) */
1049 if (fft_bits != s->fft_bits) {
1050 AVFrame *outpicref;
1051
1052 s->fft_bits = fft_bits;
1053
1054 /* FFT buffers: x2 for each (display) channel buffer.
1055 * Note: we use free and malloc instead of a realloc-like function to
1056 * make sure the buffer is aligned in memory for the FFT functions. */
1057 for (i = 0; i < s->nb_display_channels; i++) {
1058 if (s->stop) {
1059 av_fft_end(s->ifft[i]);
1060 av_freep(&s->fft_scratch[i]);
1061 }
1062 av_fft_end(s->fft[i]);
1063 av_freep(&s->fft_data[i]);
1064 }
1065 av_freep(&s->fft_data);
1066
1067 s->nb_display_channels = inlink->channels;
1068 for (i = 0; i < s->nb_display_channels; i++) {
1069 s->fft[i] = av_fft_init(fft_bits + !!s->stop, 0);
1070 if (s->stop) {
1071 s->ifft[i] = av_fft_init(fft_bits + !!s->stop, 1);
1072 if (!s->ifft[i]) {
1073 av_log(ctx, AV_LOG_ERROR, "Unable to create Inverse FFT context. "
1074 "The window size might be too high.\n");
1075 return AVERROR(EINVAL);
1076 }
1077 }
1078 if (!s->fft[i]) {
1079 av_log(ctx, AV_LOG_ERROR, "Unable to create FFT context. "
1080 "The window size might be too high.\n");
1081 return AVERROR(EINVAL);
1082 }
1083 }
1084
1085 s->magnitudes = av_calloc(s->nb_display_channels, sizeof(*s->magnitudes));
1086 if (!s->magnitudes)
1087 return AVERROR(ENOMEM);
1088 for (i = 0; i < s->nb_display_channels; i++) {
1089 s->magnitudes[i] = av_calloc(s->orientation == VERTICAL ? s->h : s->w, sizeof(**s->magnitudes));
1090 if (!s->magnitudes[i])
1091 return AVERROR(ENOMEM);
1092 }
1093
1094 s->phases = av_calloc(s->nb_display_channels, sizeof(*s->phases));
1095 if (!s->phases)
1096 return AVERROR(ENOMEM);
1097 for (i = 0; i < s->nb_display_channels; i++) {
1098 s->phases[i] = av_calloc(s->orientation == VERTICAL ? s->h : s->w, sizeof(**s->phases));
1099 if (!s->phases[i])
1100 return AVERROR(ENOMEM);
1101 }
1102
1103 av_freep(&s->color_buffer);
1104 s->color_buffer = av_calloc(s->nb_display_channels, sizeof(*s->color_buffer));
1105 if (!s->color_buffer)
1106 return AVERROR(ENOMEM);
1107 for (i = 0; i < s->nb_display_channels; i++) {
1108 s->color_buffer[i] = av_calloc(s->orientation == VERTICAL ? s->h * 3 : s->w * 3, sizeof(**s->color_buffer));
1109 if (!s->color_buffer[i])
1110 return AVERROR(ENOMEM);
1111 }
1112
1113 s->fft_data = av_calloc(s->nb_display_channels, sizeof(*s->fft_data));
1114 if (!s->fft_data)
1115 return AVERROR(ENOMEM);
1116 s->fft_scratch = av_calloc(s->nb_display_channels, sizeof(*s->fft_scratch));
1117 if (!s->fft_scratch)
1118 return AVERROR(ENOMEM);
1119 for (i = 0; i < s->nb_display_channels; i++) {
1120 s->fft_data[i] = av_calloc(s->buf_size, sizeof(**s->fft_data));
1121 if (!s->fft_data[i])
1122 return AVERROR(ENOMEM);
1123
1124 s->fft_scratch[i] = av_calloc(s->buf_size, sizeof(**s->fft_scratch));
1125 if (!s->fft_scratch[i])
1126 return AVERROR(ENOMEM);
1127 }
1128
1129 /* pre-calc windowing function */
1130 s->window_func_lut =
1131 av_realloc_f(s->window_func_lut, s->win_size,
1132 sizeof(*s->window_func_lut));
1133 if (!s->window_func_lut)
1134 return AVERROR(ENOMEM);
1135 generate_window_func(s->window_func_lut, s->win_size, s->win_func, &overlap);
1136 if (s->overlap == 1)
1137 s->overlap = overlap;
1138 s->hop_size = (1.f - s->overlap) * s->win_size;
1139 if (s->hop_size < 1) {
1140 av_log(ctx, AV_LOG_ERROR, "overlap %f too big\n", s->overlap);
1141 return AVERROR(EINVAL);
1142 }
1143
1144 for (s->win_scale = 0, i = 0; i < s->win_size; i++) {
1145 s->win_scale += s->window_func_lut[i] * s->window_func_lut[i];
1146 }
1147 s->win_scale = 1.f / sqrtf(s->win_scale);
1148
1149 /* prepare the initial picref buffer (black frame) */
1150 av_frame_free(&s->outpicref);
1151 s->outpicref = outpicref =
1152 ff_get_video_buffer(outlink, outlink->w, outlink->h);
1153 if (!outpicref)
1154 return AVERROR(ENOMEM);
1155 outpicref->sample_aspect_ratio = (AVRational){1,1};
1156 for (i = 0; i < outlink->h; i++) {
1157 memset(outpicref->data[0] + i * outpicref->linesize[0], 0, outlink->w);
1158 memset(outpicref->data[1] + i * outpicref->linesize[1], 128, outlink->w);
1159 memset(outpicref->data[2] + i * outpicref->linesize[2], 128, outlink->w);
1160 }
1161 outpicref->color_range = AVCOL_RANGE_JPEG;
1162
1163 if (!s->single_pic && s->legend)
1164 draw_legend(ctx, 0);
1165 }
1166
1167 if ((s->orientation == VERTICAL && s->xpos >= s->w) ||
1168 (s->orientation == HORIZONTAL && s->xpos >= s->h))
1169 s->xpos = 0;
1170
1171 s->auto_frame_rate = av_make_q(inlink->sample_rate, s->hop_size);
1172 if (s->orientation == VERTICAL && s->sliding == FULLFRAME)
1173 s->auto_frame_rate.den *= s->w;
1174 if (s->orientation == HORIZONTAL && s->sliding == FULLFRAME)
1175 s->auto_frame_rate.den *= s->h;
1176 if (!s->single_pic && strcmp(s->rate_str, "auto")) {
1177 int ret = av_parse_video_rate(&s->frame_rate, s->rate_str);
1178 if (ret < 0)
1179 return ret;
1180 } else {
1181 s->frame_rate = s->auto_frame_rate;
1182 }
1183 outlink->frame_rate = s->frame_rate;
1184 outlink->time_base = av_inv_q(outlink->frame_rate);
1185
1186 if (s->orientation == VERTICAL) {
1187 s->combine_buffer =
1188 av_realloc_f(s->combine_buffer, s->h * 3,
1189 sizeof(*s->combine_buffer));
1190 } else {
1191 s->combine_buffer =
1192 av_realloc_f(s->combine_buffer, s->w * 3,
1193 sizeof(*s->combine_buffer));
1194 }
1195
1196 av_log(ctx, AV_LOG_VERBOSE, "s:%dx%d FFT window size:%d\n",
1197 s->w, s->h, s->win_size);
1198
1199 av_audio_fifo_free(s->fifo);
1200 s->fifo = av_audio_fifo_alloc(inlink->format, inlink->channels, s->win_size);
1201 if (!s->fifo)
1202 return AVERROR(ENOMEM);
1203 return 0;
1204 }
1205
1206 #define RE(y, ch) s->fft_data[ch][y].re
1207 #define IM(y, ch) s->fft_data[ch][y].im
1208 #define MAGNITUDE(y, ch) hypotf(RE(y, ch), IM(y, ch))
1209 #define PHASE(y, ch) atan2f(IM(y, ch), RE(y, ch))
1210
calc_channel_magnitudes(AVFilterContext * ctx,void * arg,int jobnr,int nb_jobs)1211 static int calc_channel_magnitudes(AVFilterContext *ctx, void *arg, int jobnr, int nb_jobs)
1212 {
1213 ShowSpectrumContext *s = ctx->priv;
1214 const double w = s->win_scale * (s->scale == LOG ? s->win_scale : 1);
1215 int y, h = s->orientation == VERTICAL ? s->h : s->w;
1216 const float f = s->gain * w;
1217 const int ch = jobnr;
1218 float *magnitudes = s->magnitudes[ch];
1219
1220 for (y = 0; y < h; y++)
1221 magnitudes[y] = MAGNITUDE(y, ch) * f;
1222
1223 return 0;
1224 }
1225
calc_channel_phases(AVFilterContext * ctx,void * arg,int jobnr,int nb_jobs)1226 static int calc_channel_phases(AVFilterContext *ctx, void *arg, int jobnr, int nb_jobs)
1227 {
1228 ShowSpectrumContext *s = ctx->priv;
1229 const int h = s->orientation == VERTICAL ? s->h : s->w;
1230 const int ch = jobnr;
1231 float *phases = s->phases[ch];
1232 int y;
1233
1234 for (y = 0; y < h; y++)
1235 phases[y] = (PHASE(y, ch) / M_PI + 1) / 2;
1236
1237 return 0;
1238 }
1239
acalc_magnitudes(ShowSpectrumContext * s)1240 static void acalc_magnitudes(ShowSpectrumContext *s)
1241 {
1242 const double w = s->win_scale * (s->scale == LOG ? s->win_scale : 1);
1243 int ch, y, h = s->orientation == VERTICAL ? s->h : s->w;
1244 const float f = s->gain * w;
1245
1246 for (ch = 0; ch < s->nb_display_channels; ch++) {
1247 float *magnitudes = s->magnitudes[ch];
1248
1249 for (y = 0; y < h; y++)
1250 magnitudes[y] += MAGNITUDE(y, ch) * f;
1251 }
1252 }
1253
scale_magnitudes(ShowSpectrumContext * s,float scale)1254 static void scale_magnitudes(ShowSpectrumContext *s, float scale)
1255 {
1256 int ch, y, h = s->orientation == VERTICAL ? s->h : s->w;
1257
1258 for (ch = 0; ch < s->nb_display_channels; ch++) {
1259 float *magnitudes = s->magnitudes[ch];
1260
1261 for (y = 0; y < h; y++)
1262 magnitudes[y] *= scale;
1263 }
1264 }
1265
clear_combine_buffer(ShowSpectrumContext * s,int size)1266 static void clear_combine_buffer(ShowSpectrumContext *s, int size)
1267 {
1268 int y;
1269
1270 for (y = 0; y < size; y++) {
1271 s->combine_buffer[3 * y ] = 0;
1272 s->combine_buffer[3 * y + 1] = 127.5;
1273 s->combine_buffer[3 * y + 2] = 127.5;
1274 }
1275 }
1276
plot_spectrum_column(AVFilterLink * inlink,AVFrame * insamples)1277 static int plot_spectrum_column(AVFilterLink *inlink, AVFrame *insamples)
1278 {
1279 AVFilterContext *ctx = inlink->dst;
1280 AVFilterLink *outlink = ctx->outputs[0];
1281 ShowSpectrumContext *s = ctx->priv;
1282 AVFrame *outpicref = s->outpicref;
1283 int ret, plane, x, y, z = s->orientation == VERTICAL ? s->h : s->w;
1284
1285 /* fill a new spectrum column */
1286 /* initialize buffer for combining to black */
1287 clear_combine_buffer(s, z);
1288
1289 ctx->internal->execute(ctx, s->plot_channel, NULL, NULL, s->nb_display_channels);
1290
1291 for (y = 0; y < z * 3; y++) {
1292 for (x = 0; x < s->nb_display_channels; x++) {
1293 s->combine_buffer[y] += s->color_buffer[x][y];
1294 }
1295 }
1296
1297 av_frame_make_writable(s->outpicref);
1298 /* copy to output */
1299 if (s->orientation == VERTICAL) {
1300 if (s->sliding == SCROLL) {
1301 for (plane = 0; plane < 3; plane++) {
1302 for (y = 0; y < s->h; y++) {
1303 uint8_t *p = outpicref->data[plane] + s->start_x +
1304 (y + s->start_y) * outpicref->linesize[plane];
1305 memmove(p, p + 1, s->w - 1);
1306 }
1307 }
1308 s->xpos = s->w - 1;
1309 } else if (s->sliding == RSCROLL) {
1310 for (plane = 0; plane < 3; plane++) {
1311 for (y = 0; y < s->h; y++) {
1312 uint8_t *p = outpicref->data[plane] + s->start_x +
1313 (y + s->start_y) * outpicref->linesize[plane];
1314 memmove(p + 1, p, s->w - 1);
1315 }
1316 }
1317 s->xpos = 0;
1318 }
1319 for (plane = 0; plane < 3; plane++) {
1320 uint8_t *p = outpicref->data[plane] + s->start_x +
1321 (outlink->h - 1 - s->start_y) * outpicref->linesize[plane] +
1322 s->xpos;
1323 for (y = 0; y < s->h; y++) {
1324 *p = lrintf(av_clipf(s->combine_buffer[3 * y + plane], 0, 255));
1325 p -= outpicref->linesize[plane];
1326 }
1327 }
1328 } else {
1329 if (s->sliding == SCROLL) {
1330 for (plane = 0; plane < 3; plane++) {
1331 for (y = 1; y < s->h; y++) {
1332 memmove(outpicref->data[plane] + (y-1 + s->start_y) * outpicref->linesize[plane] + s->start_x,
1333 outpicref->data[plane] + (y + s->start_y) * outpicref->linesize[plane] + s->start_x,
1334 s->w);
1335 }
1336 }
1337 s->xpos = s->h - 1;
1338 } else if (s->sliding == RSCROLL) {
1339 for (plane = 0; plane < 3; plane++) {
1340 for (y = s->h - 1; y >= 1; y--) {
1341 memmove(outpicref->data[plane] + (y + s->start_y) * outpicref->linesize[plane] + s->start_x,
1342 outpicref->data[plane] + (y-1 + s->start_y) * outpicref->linesize[plane] + s->start_x,
1343 s->w);
1344 }
1345 }
1346 s->xpos = 0;
1347 }
1348 for (plane = 0; plane < 3; plane++) {
1349 uint8_t *p = outpicref->data[plane] + s->start_x +
1350 (s->xpos + s->start_y) * outpicref->linesize[plane];
1351 for (x = 0; x < s->w; x++) {
1352 *p = lrintf(av_clipf(s->combine_buffer[3 * x + plane], 0, 255));
1353 p++;
1354 }
1355 }
1356 }
1357
1358 if (s->sliding != FULLFRAME || s->xpos == 0)
1359 outpicref->pts = av_rescale_q(insamples->pts, inlink->time_base, outlink->time_base);
1360
1361 s->xpos++;
1362 if (s->orientation == VERTICAL && s->xpos >= s->w)
1363 s->xpos = 0;
1364 if (s->orientation == HORIZONTAL && s->xpos >= s->h)
1365 s->xpos = 0;
1366 if (!s->single_pic && (s->sliding != FULLFRAME || s->xpos == 0)) {
1367 if (s->old_pts < outpicref->pts) {
1368 AVFrame *clone;
1369
1370 if (s->legend) {
1371 char *units = get_time(ctx, insamples->pts /(float)inlink->sample_rate, x);
1372 if (!units)
1373 return AVERROR(ENOMEM);
1374
1375 if (s->orientation == VERTICAL) {
1376 for (y = 0; y < 10; y++) {
1377 memset(s->outpicref->data[0] + outlink->w / 2 - 4 * s->old_len +
1378 (outlink->h - s->start_y / 2 - 20 + y) * s->outpicref->linesize[0], 0, 10 * s->old_len);
1379 }
1380 drawtext(s->outpicref,
1381 outlink->w / 2 - 4 * strlen(units),
1382 outlink->h - s->start_y / 2 - 20,
1383 units, 0);
1384 } else {
1385 for (y = 0; y < 10 * s->old_len; y++) {
1386 memset(s->outpicref->data[0] + s->start_x / 7 + 20 +
1387 (outlink->h / 2 - 4 * s->old_len + y) * s->outpicref->linesize[0], 0, 10);
1388 }
1389 drawtext(s->outpicref,
1390 s->start_x / 7 + 20,
1391 outlink->h / 2 - 4 * strlen(units),
1392 units, 1);
1393 }
1394 s->old_len = strlen(units);
1395 av_free(units);
1396 }
1397 s->old_pts = outpicref->pts;
1398 clone = av_frame_clone(s->outpicref);
1399 if (!clone)
1400 return AVERROR(ENOMEM);
1401 ret = ff_filter_frame(outlink, clone);
1402 if (ret < 0)
1403 return ret;
1404 return 0;
1405 }
1406 }
1407
1408 return 1;
1409 }
1410
1411 #if CONFIG_SHOWSPECTRUM_FILTER
1412
activate(AVFilterContext * ctx)1413 static int activate(AVFilterContext *ctx)
1414 {
1415 AVFilterLink *inlink = ctx->inputs[0];
1416 AVFilterLink *outlink = ctx->outputs[0];
1417 ShowSpectrumContext *s = ctx->priv;
1418 int ret;
1419
1420 FF_FILTER_FORWARD_STATUS_BACK(outlink, inlink);
1421
1422 if (av_audio_fifo_size(s->fifo) < s->win_size) {
1423 AVFrame *frame = NULL;
1424
1425 ret = ff_inlink_consume_frame(inlink, &frame);
1426 if (ret < 0)
1427 return ret;
1428 if (ret > 0) {
1429 s->pts = frame->pts;
1430 s->consumed = 0;
1431
1432 av_audio_fifo_write(s->fifo, (void **)frame->extended_data, frame->nb_samples);
1433 av_frame_free(&frame);
1434 }
1435 }
1436
1437 if (s->outpicref && (av_audio_fifo_size(s->fifo) >= s->win_size ||
1438 ff_outlink_get_status(inlink))) {
1439 AVFrame *fin = ff_get_audio_buffer(inlink, s->win_size);
1440 if (!fin)
1441 return AVERROR(ENOMEM);
1442
1443 fin->pts = s->pts + s->consumed;
1444 s->consumed += s->hop_size;
1445 ret = av_audio_fifo_peek(s->fifo, (void **)fin->extended_data,
1446 FFMIN(s->win_size, av_audio_fifo_size(s->fifo)));
1447 if (ret < 0) {
1448 av_frame_free(&fin);
1449 return ret;
1450 }
1451
1452 av_assert0(fin->nb_samples == s->win_size);
1453
1454 ctx->internal->execute(ctx, run_channel_fft, fin, NULL, s->nb_display_channels);
1455
1456 if (s->data == D_MAGNITUDE)
1457 ctx->internal->execute(ctx, calc_channel_magnitudes, NULL, NULL, s->nb_display_channels);
1458
1459 if (s->data == D_PHASE)
1460 ctx->internal->execute(ctx, calc_channel_phases, NULL, NULL, s->nb_display_channels);
1461
1462 ret = plot_spectrum_column(inlink, fin);
1463
1464 av_frame_free(&fin);
1465 av_audio_fifo_drain(s->fifo, s->hop_size);
1466 if (ret <= 0 && !ff_outlink_get_status(inlink))
1467 return ret;
1468 }
1469
1470 if (ff_outlink_get_status(inlink) == AVERROR_EOF &&
1471 s->sliding == FULLFRAME &&
1472 s->xpos > 0 && s->outpicref) {
1473 int64_t pts;
1474
1475 if (s->orientation == VERTICAL) {
1476 for (int i = 0; i < outlink->h; i++) {
1477 memset(s->outpicref->data[0] + i * s->outpicref->linesize[0] + s->xpos, 0, outlink->w - s->xpos);
1478 memset(s->outpicref->data[1] + i * s->outpicref->linesize[1] + s->xpos, 128, outlink->w - s->xpos);
1479 memset(s->outpicref->data[2] + i * s->outpicref->linesize[2] + s->xpos, 128, outlink->w - s->xpos);
1480 }
1481 } else {
1482 for (int i = s->xpos; i < outlink->h; i++) {
1483 memset(s->outpicref->data[0] + i * s->outpicref->linesize[0], 0, outlink->w);
1484 memset(s->outpicref->data[1] + i * s->outpicref->linesize[1], 128, outlink->w);
1485 memset(s->outpicref->data[2] + i * s->outpicref->linesize[2], 128, outlink->w);
1486 }
1487 }
1488 s->outpicref->pts += s->consumed;
1489 pts = s->outpicref->pts;
1490 ret = ff_filter_frame(outlink, s->outpicref);
1491 s->outpicref = NULL;
1492 ff_outlink_set_status(outlink, AVERROR_EOF, pts);
1493 return 0;
1494 }
1495
1496 FF_FILTER_FORWARD_STATUS(inlink, outlink);
1497 if (av_audio_fifo_size(s->fifo) >= s->win_size ||
1498 ff_outlink_get_status(inlink) == AVERROR_EOF) {
1499 ff_filter_set_ready(ctx, 10);
1500 return 0;
1501 }
1502
1503 if (ff_outlink_frame_wanted(outlink) && av_audio_fifo_size(s->fifo) < s->win_size &&
1504 ff_outlink_get_status(inlink) != AVERROR_EOF) {
1505 ff_inlink_request_frame(inlink);
1506 return 0;
1507 }
1508
1509 return FFERROR_NOT_READY;
1510 }
1511
1512 static const AVFilterPad showspectrum_inputs[] = {
1513 {
1514 .name = "default",
1515 .type = AVMEDIA_TYPE_AUDIO,
1516 },
1517 { NULL }
1518 };
1519
1520 static const AVFilterPad showspectrum_outputs[] = {
1521 {
1522 .name = "default",
1523 .type = AVMEDIA_TYPE_VIDEO,
1524 .config_props = config_output,
1525 },
1526 { NULL }
1527 };
1528
1529 AVFilter ff_avf_showspectrum = {
1530 .name = "showspectrum",
1531 .description = NULL_IF_CONFIG_SMALL("Convert input audio to a spectrum video output."),
1532 .uninit = uninit,
1533 .query_formats = query_formats,
1534 .priv_size = sizeof(ShowSpectrumContext),
1535 .inputs = showspectrum_inputs,
1536 .outputs = showspectrum_outputs,
1537 .activate = activate,
1538 .priv_class = &showspectrum_class,
1539 .flags = AVFILTER_FLAG_SLICE_THREADS,
1540 };
1541 #endif // CONFIG_SHOWSPECTRUM_FILTER
1542
1543 #if CONFIG_SHOWSPECTRUMPIC_FILTER
1544
1545 static const AVOption showspectrumpic_options[] = {
1546 { "size", "set video size", OFFSET(w), AV_OPT_TYPE_IMAGE_SIZE, {.str = "4096x2048"}, 0, 0, FLAGS },
1547 { "s", "set video size", OFFSET(w), AV_OPT_TYPE_IMAGE_SIZE, {.str = "4096x2048"}, 0, 0, FLAGS },
1548 { "mode", "set channel display mode", OFFSET(mode), AV_OPT_TYPE_INT, {.i64=COMBINED}, 0, NB_MODES-1, FLAGS, "mode" },
1549 { "combined", "combined mode", 0, AV_OPT_TYPE_CONST, {.i64=COMBINED}, 0, 0, FLAGS, "mode" },
1550 { "separate", "separate mode", 0, AV_OPT_TYPE_CONST, {.i64=SEPARATE}, 0, 0, FLAGS, "mode" },
1551 { "color", "set channel coloring", OFFSET(color_mode), AV_OPT_TYPE_INT, {.i64=INTENSITY}, 0, NB_CLMODES-1, FLAGS, "color" },
1552 { "channel", "separate color for each channel", 0, AV_OPT_TYPE_CONST, {.i64=CHANNEL}, 0, 0, FLAGS, "color" },
1553 { "intensity", "intensity based coloring", 0, AV_OPT_TYPE_CONST, {.i64=INTENSITY}, 0, 0, FLAGS, "color" },
1554 { "rainbow", "rainbow based coloring", 0, AV_OPT_TYPE_CONST, {.i64=RAINBOW}, 0, 0, FLAGS, "color" },
1555 { "moreland", "moreland based coloring", 0, AV_OPT_TYPE_CONST, {.i64=MORELAND}, 0, 0, FLAGS, "color" },
1556 { "nebulae", "nebulae based coloring", 0, AV_OPT_TYPE_CONST, {.i64=NEBULAE}, 0, 0, FLAGS, "color" },
1557 { "fire", "fire based coloring", 0, AV_OPT_TYPE_CONST, {.i64=FIRE}, 0, 0, FLAGS, "color" },
1558 { "fiery", "fiery based coloring", 0, AV_OPT_TYPE_CONST, {.i64=FIERY}, 0, 0, FLAGS, "color" },
1559 { "fruit", "fruit based coloring", 0, AV_OPT_TYPE_CONST, {.i64=FRUIT}, 0, 0, FLAGS, "color" },
1560 { "cool", "cool based coloring", 0, AV_OPT_TYPE_CONST, {.i64=COOL}, 0, 0, FLAGS, "color" },
1561 { "magma", "magma based coloring", 0, AV_OPT_TYPE_CONST, {.i64=MAGMA}, 0, 0, FLAGS, "color" },
1562 { "green", "green based coloring", 0, AV_OPT_TYPE_CONST, {.i64=GREEN}, 0, 0, FLAGS, "color" },
1563 { "viridis", "viridis based coloring", 0, AV_OPT_TYPE_CONST, {.i64=VIRIDIS}, 0, 0, FLAGS, "color" },
1564 { "plasma", "plasma based coloring", 0, AV_OPT_TYPE_CONST, {.i64=PLASMA}, 0, 0, FLAGS, "color" },
1565 { "cividis", "cividis based coloring", 0, AV_OPT_TYPE_CONST, {.i64=CIVIDIS}, 0, 0, FLAGS, "color" },
1566 { "terrain", "terrain based coloring", 0, AV_OPT_TYPE_CONST, {.i64=TERRAIN}, 0, 0, FLAGS, "color" },
1567 { "scale", "set display scale", OFFSET(scale), AV_OPT_TYPE_INT, {.i64=LOG}, 0, NB_SCALES-1, FLAGS, "scale" },
1568 { "lin", "linear", 0, AV_OPT_TYPE_CONST, {.i64=LINEAR}, 0, 0, FLAGS, "scale" },
1569 { "sqrt", "square root", 0, AV_OPT_TYPE_CONST, {.i64=SQRT}, 0, 0, FLAGS, "scale" },
1570 { "cbrt", "cubic root", 0, AV_OPT_TYPE_CONST, {.i64=CBRT}, 0, 0, FLAGS, "scale" },
1571 { "log", "logarithmic", 0, AV_OPT_TYPE_CONST, {.i64=LOG}, 0, 0, FLAGS, "scale" },
1572 { "4thrt","4th root", 0, AV_OPT_TYPE_CONST, {.i64=FOURTHRT}, 0, 0, FLAGS, "scale" },
1573 { "5thrt","5th root", 0, AV_OPT_TYPE_CONST, {.i64=FIFTHRT}, 0, 0, FLAGS, "scale" },
1574 { "fscale", "set frequency scale", OFFSET(fscale), AV_OPT_TYPE_INT, {.i64=F_LINEAR}, 0, NB_FSCALES-1, FLAGS, "fscale" },
1575 { "lin", "linear", 0, AV_OPT_TYPE_CONST, {.i64=F_LINEAR}, 0, 0, FLAGS, "fscale" },
1576 { "log", "logarithmic", 0, AV_OPT_TYPE_CONST, {.i64=F_LOG}, 0, 0, FLAGS, "fscale" },
1577 { "saturation", "color saturation multiplier", OFFSET(saturation), AV_OPT_TYPE_FLOAT, {.dbl = 1}, -10, 10, FLAGS },
1578 { "win_func", "set window function", OFFSET(win_func), AV_OPT_TYPE_INT, {.i64 = WFUNC_HANNING}, 0, NB_WFUNC-1, FLAGS, "win_func" },
1579 { "rect", "Rectangular", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_RECT}, 0, 0, FLAGS, "win_func" },
1580 { "bartlett", "Bartlett", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_BARTLETT}, 0, 0, FLAGS, "win_func" },
1581 { "hann", "Hann", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_HANNING}, 0, 0, FLAGS, "win_func" },
1582 { "hanning", "Hanning", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_HANNING}, 0, 0, FLAGS, "win_func" },
1583 { "hamming", "Hamming", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_HAMMING}, 0, 0, FLAGS, "win_func" },
1584 { "blackman", "Blackman", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_BLACKMAN}, 0, 0, FLAGS, "win_func" },
1585 { "welch", "Welch", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_WELCH}, 0, 0, FLAGS, "win_func" },
1586 { "flattop", "Flat-top", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_FLATTOP}, 0, 0, FLAGS, "win_func" },
1587 { "bharris", "Blackman-Harris", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_BHARRIS}, 0, 0, FLAGS, "win_func" },
1588 { "bnuttall", "Blackman-Nuttall", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_BNUTTALL}, 0, 0, FLAGS, "win_func" },
1589 { "bhann", "Bartlett-Hann", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_BHANN}, 0, 0, FLAGS, "win_func" },
1590 { "sine", "Sine", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_SINE}, 0, 0, FLAGS, "win_func" },
1591 { "nuttall", "Nuttall", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_NUTTALL}, 0, 0, FLAGS, "win_func" },
1592 { "lanczos", "Lanczos", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_LANCZOS}, 0, 0, FLAGS, "win_func" },
1593 { "gauss", "Gauss", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_GAUSS}, 0, 0, FLAGS, "win_func" },
1594 { "tukey", "Tukey", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_TUKEY}, 0, 0, FLAGS, "win_func" },
1595 { "dolph", "Dolph-Chebyshev", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_DOLPH}, 0, 0, FLAGS, "win_func" },
1596 { "cauchy", "Cauchy", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_CAUCHY}, 0, 0, FLAGS, "win_func" },
1597 { "parzen", "Parzen", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_PARZEN}, 0, 0, FLAGS, "win_func" },
1598 { "poisson", "Poisson", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_POISSON}, 0, 0, FLAGS, "win_func" },
1599 { "bohman", "Bohman", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_BOHMAN}, 0, 0, FLAGS, "win_func" },
1600 { "orientation", "set orientation", OFFSET(orientation), AV_OPT_TYPE_INT, {.i64=VERTICAL}, 0, NB_ORIENTATIONS-1, FLAGS, "orientation" },
1601 { "vertical", NULL, 0, AV_OPT_TYPE_CONST, {.i64=VERTICAL}, 0, 0, FLAGS, "orientation" },
1602 { "horizontal", NULL, 0, AV_OPT_TYPE_CONST, {.i64=HORIZONTAL}, 0, 0, FLAGS, "orientation" },
1603 { "gain", "set scale gain", OFFSET(gain), AV_OPT_TYPE_FLOAT, {.dbl = 1}, 0, 128, FLAGS },
1604 { "legend", "draw legend", OFFSET(legend), AV_OPT_TYPE_BOOL, {.i64 = 1}, 0, 1, FLAGS },
1605 { "rotation", "color rotation", OFFSET(rotation), AV_OPT_TYPE_FLOAT, {.dbl = 0}, -1, 1, FLAGS },
1606 { "start", "start frequency", OFFSET(start), AV_OPT_TYPE_INT, {.i64 = 0}, 0, INT32_MAX, FLAGS },
1607 { "stop", "stop frequency", OFFSET(stop), AV_OPT_TYPE_INT, {.i64 = 0}, 0, INT32_MAX, FLAGS },
1608 { NULL }
1609 };
1610
1611 AVFILTER_DEFINE_CLASS(showspectrumpic);
1612
showspectrumpic_request_frame(AVFilterLink * outlink)1613 static int showspectrumpic_request_frame(AVFilterLink *outlink)
1614 {
1615 AVFilterContext *ctx = outlink->src;
1616 ShowSpectrumContext *s = ctx->priv;
1617 AVFilterLink *inlink = ctx->inputs[0];
1618 int ret, samples;
1619
1620 ret = ff_request_frame(inlink);
1621 samples = av_audio_fifo_size(s->fifo);
1622 if (ret == AVERROR_EOF && s->outpicref && samples > 0) {
1623 int consumed = 0;
1624 int x = 0, sz = s->orientation == VERTICAL ? s->w : s->h;
1625 int ch, spf, spb;
1626 AVFrame *fin;
1627
1628 spf = s->win_size * (samples / ((s->win_size * sz) * ceil(samples / (float)(s->win_size * sz))));
1629 spf = FFMAX(1, spf);
1630
1631 spb = (samples / (spf * sz)) * spf;
1632
1633 fin = ff_get_audio_buffer(inlink, s->win_size);
1634 if (!fin)
1635 return AVERROR(ENOMEM);
1636
1637 while (x < sz) {
1638 ret = av_audio_fifo_peek(s->fifo, (void **)fin->extended_data, s->win_size);
1639 if (ret < 0) {
1640 av_frame_free(&fin);
1641 return ret;
1642 }
1643
1644 av_audio_fifo_drain(s->fifo, spf);
1645
1646 if (ret < s->win_size) {
1647 for (ch = 0; ch < s->nb_display_channels; ch++) {
1648 memset(fin->extended_data[ch] + ret * sizeof(float), 0,
1649 (s->win_size - ret) * sizeof(float));
1650 }
1651 }
1652
1653 ctx->internal->execute(ctx, run_channel_fft, fin, NULL, s->nb_display_channels);
1654 acalc_magnitudes(s);
1655
1656 consumed += spf;
1657 if (consumed >= spb) {
1658 int h = s->orientation == VERTICAL ? s->h : s->w;
1659
1660 scale_magnitudes(s, 1.f / (consumed / spf));
1661 plot_spectrum_column(inlink, fin);
1662 consumed = 0;
1663 x++;
1664 for (ch = 0; ch < s->nb_display_channels; ch++)
1665 memset(s->magnitudes[ch], 0, h * sizeof(float));
1666 }
1667 }
1668
1669 av_frame_free(&fin);
1670 s->outpicref->pts = 0;
1671
1672 if (s->legend)
1673 draw_legend(ctx, samples);
1674
1675 ret = ff_filter_frame(outlink, s->outpicref);
1676 s->outpicref = NULL;
1677 }
1678
1679 return ret;
1680 }
1681
showspectrumpic_filter_frame(AVFilterLink * inlink,AVFrame * insamples)1682 static int showspectrumpic_filter_frame(AVFilterLink *inlink, AVFrame *insamples)
1683 {
1684 AVFilterContext *ctx = inlink->dst;
1685 ShowSpectrumContext *s = ctx->priv;
1686 int ret;
1687
1688 ret = av_audio_fifo_write(s->fifo, (void **)insamples->extended_data, insamples->nb_samples);
1689 av_frame_free(&insamples);
1690 return ret;
1691 }
1692
1693 static const AVFilterPad showspectrumpic_inputs[] = {
1694 {
1695 .name = "default",
1696 .type = AVMEDIA_TYPE_AUDIO,
1697 .filter_frame = showspectrumpic_filter_frame,
1698 },
1699 { NULL }
1700 };
1701
1702 static const AVFilterPad showspectrumpic_outputs[] = {
1703 {
1704 .name = "default",
1705 .type = AVMEDIA_TYPE_VIDEO,
1706 .config_props = config_output,
1707 .request_frame = showspectrumpic_request_frame,
1708 },
1709 { NULL }
1710 };
1711
1712 AVFilter ff_avf_showspectrumpic = {
1713 .name = "showspectrumpic",
1714 .description = NULL_IF_CONFIG_SMALL("Convert input audio to a spectrum video output single picture."),
1715 .uninit = uninit,
1716 .query_formats = query_formats,
1717 .priv_size = sizeof(ShowSpectrumContext),
1718 .inputs = showspectrumpic_inputs,
1719 .outputs = showspectrumpic_outputs,
1720 .priv_class = &showspectrumpic_class,
1721 .flags = AVFILTER_FLAG_SLICE_THREADS,
1722 };
1723
1724 #endif // CONFIG_SHOWSPECTRUMPIC_FILTER
1725