1 /*
2 * Copyright (c) 2015 Paul B Mahol
3 *
4 * This file is part of FFmpeg.
5 *
6 * FFmpeg is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU Lesser General Public
8 * License as published by the Free Software Foundation; either
9 * version 2.1 of the License, or (at your option) any later version.
10 *
11 * FFmpeg is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 * Lesser General Public License for more details.
15 *
16 * You should have received a copy of the GNU Lesser General Public
17 * License along with FFmpeg; if not, write to the Free Software
18 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
19 */
20
21 /**
22 * @file
23 * audio to video multimedia aphasemeter filter
24 */
25
26 #include "libavutil/avassert.h"
27 #include "libavutil/channel_layout.h"
28 #include "libavutil/intreadwrite.h"
29 #include "libavutil/opt.h"
30 #include "libavutil/parseutils.h"
31 #include "libavutil/timestamp.h"
32 #include "avfilter.h"
33 #include "formats.h"
34 #include "audio.h"
35 #include "video.h"
36 #include "internal.h"
37 #include "float.h"
38
39 typedef struct AudioPhaseMeterContext {
40 const AVClass *class;
41 AVFrame *out;
42 int do_video;
43 int do_phasing_detection;
44 int w, h;
45 AVRational frame_rate;
46 int contrast[4];
47 uint8_t *mpc_str;
48 uint8_t mpc[4];
49 int draw_median_phase;
50 int is_mono;
51 int is_out_phase;
52 int start_mono_presence;
53 int start_out_phase_presence;
54 float tolerance;
55 float angle;
56 float phase;
57 AVRational time_base;
58 int64_t duration;
59 int64_t frame_end;
60 int64_t mono_idx[2];
61 int64_t out_phase_idx[2];
62 } AudioPhaseMeterContext;
63
64 #define MAX_DURATION (24*60*60*1000000LL)
65 #define OFFSET(x) offsetof(AudioPhaseMeterContext, x)
66 #define FLAGS AV_OPT_FLAG_FILTERING_PARAM|AV_OPT_FLAG_VIDEO_PARAM
67 #define get_duration(index) (index[1] - index[0])
68
69 static const AVOption aphasemeter_options[] = {
70 { "rate", "set video rate", OFFSET(frame_rate), AV_OPT_TYPE_VIDEO_RATE, {.str="25"}, 0, INT_MAX, FLAGS },
71 { "r", "set video rate", OFFSET(frame_rate), AV_OPT_TYPE_VIDEO_RATE, {.str="25"}, 0, INT_MAX, FLAGS },
72 { "size", "set video size", OFFSET(w), AV_OPT_TYPE_IMAGE_SIZE, {.str="800x400"}, 0, 0, FLAGS },
73 { "s", "set video size", OFFSET(w), AV_OPT_TYPE_IMAGE_SIZE, {.str="800x400"}, 0, 0, FLAGS },
74 { "rc", "set red contrast", OFFSET(contrast[0]), AV_OPT_TYPE_INT, {.i64=2}, 0, 255, FLAGS },
75 { "gc", "set green contrast", OFFSET(contrast[1]), AV_OPT_TYPE_INT, {.i64=7}, 0, 255, FLAGS },
76 { "bc", "set blue contrast", OFFSET(contrast[2]), AV_OPT_TYPE_INT, {.i64=1}, 0, 255, FLAGS },
77 { "mpc", "set median phase color", OFFSET(mpc_str), AV_OPT_TYPE_STRING, {.str = "none"}, 0, 0, FLAGS },
78 { "video", "set video output", OFFSET(do_video), AV_OPT_TYPE_BOOL, {.i64 = 1}, 0, 1, FLAGS },
79 { "phasing", "set mono and out-of-phase detection output", OFFSET(do_phasing_detection), AV_OPT_TYPE_BOOL, {.i64 = 0}, 0, 1, FLAGS },
80 { "tolerance", "set phase tolerance for mono detection", OFFSET(tolerance), AV_OPT_TYPE_FLOAT, {.dbl = 0.}, 0, 1, FLAGS },
81 { "t", "set phase tolerance for mono detection", OFFSET(tolerance), AV_OPT_TYPE_FLOAT, {.dbl = 0.}, 0, 1, FLAGS },
82 { "angle", "set angle threshold for out-of-phase detection", OFFSET(angle), AV_OPT_TYPE_FLOAT, {.dbl = 170.}, 90, 180, FLAGS },
83 { "a", "set angle threshold for out-of-phase detection", OFFSET(angle), AV_OPT_TYPE_FLOAT, {.dbl = 170.}, 90, 180, FLAGS },
84 { "duration", "set minimum mono or out-of-phase duration in seconds", OFFSET(duration), AV_OPT_TYPE_DURATION, {.i64=2000000}, 0, MAX_DURATION, FLAGS },
85 { "d", "set minimum mono or out-of-phase duration in seconds", OFFSET(duration), AV_OPT_TYPE_DURATION, {.i64=2000000}, 0, MAX_DURATION, FLAGS },
86 { NULL }
87 };
88
89 AVFILTER_DEFINE_CLASS(aphasemeter);
90
query_formats(AVFilterContext * ctx)91 static int query_formats(AVFilterContext *ctx)
92 {
93 AudioPhaseMeterContext *s = ctx->priv;
94 AVFilterFormats *formats = NULL;
95 AVFilterChannelLayouts *layout = NULL;
96 AVFilterLink *inlink = ctx->inputs[0];
97 AVFilterLink *outlink = ctx->outputs[0];
98 static const enum AVSampleFormat sample_fmts[] = { AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_NONE };
99 static const enum AVPixelFormat pix_fmts[] = { AV_PIX_FMT_RGBA, AV_PIX_FMT_NONE };
100 int ret;
101
102 formats = ff_make_format_list(sample_fmts);
103 if ((ret = ff_formats_ref (formats, &inlink->outcfg.formats )) < 0 ||
104 (ret = ff_formats_ref (formats, &outlink->incfg.formats )) < 0 ||
105 (ret = ff_add_channel_layout (&layout, AV_CH_LAYOUT_STEREO )) < 0 ||
106 (ret = ff_channel_layouts_ref (layout , &inlink->outcfg.channel_layouts)) < 0 ||
107 (ret = ff_channel_layouts_ref (layout , &outlink->incfg.channel_layouts)) < 0)
108 return ret;
109
110 formats = ff_all_samplerates();
111 if ((ret = ff_formats_ref(formats, &inlink->outcfg.samplerates)) < 0 ||
112 (ret = ff_formats_ref(formats, &outlink->incfg.samplerates)) < 0)
113 return ret;
114
115 if (s->do_video) {
116 AVFilterLink *outlink = ctx->outputs[1];
117
118 formats = ff_make_format_list(pix_fmts);
119 if ((ret = ff_formats_ref(formats, &outlink->incfg.formats)) < 0)
120 return ret;
121 }
122
123 return 0;
124 }
125
config_input(AVFilterLink * inlink)126 static int config_input(AVFilterLink *inlink)
127 {
128 AVFilterContext *ctx = inlink->dst;
129 AudioPhaseMeterContext *s = ctx->priv;
130 int nb_samples;
131 s->duration = av_rescale(s->duration, inlink->sample_rate, AV_TIME_BASE);
132
133 if (s->do_video) {
134 nb_samples = FFMAX(1, av_rescale(inlink->sample_rate, s->frame_rate.den, s->frame_rate.num));
135 inlink->partial_buf_size =
136 inlink->min_samples =
137 inlink->max_samples = nb_samples;
138 }
139
140 return 0;
141 }
142
config_video_output(AVFilterLink * outlink)143 static int config_video_output(AVFilterLink *outlink)
144 {
145 AVFilterContext *ctx = outlink->src;
146 AudioPhaseMeterContext *s = ctx->priv;
147
148 outlink->w = s->w;
149 outlink->h = s->h;
150 outlink->sample_aspect_ratio = (AVRational){1,1};
151 outlink->frame_rate = s->frame_rate;
152
153 if (!strcmp(s->mpc_str, "none"))
154 s->draw_median_phase = 0;
155 else if (av_parse_color(s->mpc, s->mpc_str, -1, ctx) >= 0)
156 s->draw_median_phase = 1;
157 else
158 return AVERROR(EINVAL);
159
160 return 0;
161 }
162
get_x(float phase,int w)163 static inline int get_x(float phase, int w)
164 {
165 return (phase + 1.) / 2. * (w - 1);
166 }
167
add_metadata(AVFrame * insamples,const char * key,char * value)168 static inline void add_metadata(AVFrame *insamples, const char *key, char *value)
169 {
170 char buf[128];
171
172 snprintf(buf, sizeof(buf), "lavfi.aphasemeter.%s", key);
173 av_dict_set(&insamples->metadata, buf, value, 0);
174 }
175
update_mono_detection(AudioPhaseMeterContext * s,AVFrame * insamples,int mono_measurement)176 static inline void update_mono_detection(AudioPhaseMeterContext *s, AVFrame *insamples, int mono_measurement)
177 {
178 int64_t mono_duration;
179 if (!s->is_mono && mono_measurement) {
180 s->is_mono = 1;
181 s->start_mono_presence = 1;
182 s->mono_idx[0] = insamples->pts;
183 }
184 if (s->is_mono && mono_measurement && s->start_mono_presence) {
185 s->mono_idx[1] = s->frame_end;
186 mono_duration = get_duration(s->mono_idx);
187 if (mono_duration >= s->duration) {
188 add_metadata(insamples, "mono_start", av_ts2timestr(s->mono_idx[0], &s->time_base));
189 av_log(s, AV_LOG_INFO, "mono_start: %s\n", av_ts2timestr(s->mono_idx[0], &s->time_base));
190 s->start_mono_presence = 0;
191 }
192 }
193 if (s->is_mono && !mono_measurement) {
194 s->mono_idx[1] = insamples ? insamples->pts : s->frame_end;
195 mono_duration = get_duration(s->mono_idx);
196 if (mono_duration >= s->duration) {
197 if (insamples) {
198 add_metadata(insamples, "mono_end", av_ts2timestr(s->mono_idx[1], &s->time_base));
199 add_metadata(insamples, "mono_duration", av_ts2timestr(mono_duration, &s->time_base));
200 }
201 av_log(s, AV_LOG_INFO, "mono_end: %s | mono_duration: %s\n", av_ts2timestr(s->mono_idx[1], &s->time_base), av_ts2timestr(mono_duration, &s->time_base));
202 }
203 s->is_mono = 0;
204 }
205 }
206
update_out_phase_detection(AudioPhaseMeterContext * s,AVFrame * insamples,int out_phase_measurement)207 static inline void update_out_phase_detection(AudioPhaseMeterContext *s, AVFrame *insamples, int out_phase_measurement)
208 {
209 int64_t out_phase_duration;
210 if (!s->is_out_phase && out_phase_measurement) {
211 s->is_out_phase = 1;
212 s->start_out_phase_presence = 1;
213 s->out_phase_idx[0] = insamples->pts;
214 }
215 if (s->is_out_phase && out_phase_measurement && s->start_out_phase_presence) {
216 s->out_phase_idx[1] = s->frame_end;
217 out_phase_duration = get_duration(s->out_phase_idx);
218 if (out_phase_duration >= s->duration) {
219 add_metadata(insamples, "out_phase_start", av_ts2timestr(s->out_phase_idx[0], &s->time_base));
220 av_log(s, AV_LOG_INFO, "out_phase_start: %s\n", av_ts2timestr(s->out_phase_idx[0], &s->time_base));
221 s->start_out_phase_presence = 0;
222 }
223 }
224 if (s->is_out_phase && !out_phase_measurement) {
225 s->out_phase_idx[1] = insamples ? insamples->pts : s->frame_end;
226 out_phase_duration = get_duration(s->out_phase_idx);
227 if (out_phase_duration >= s->duration) {
228 if (insamples) {
229 add_metadata(insamples, "out_phase_end", av_ts2timestr(s->out_phase_idx[1], &s->time_base));
230 add_metadata(insamples, "out_phase_duration", av_ts2timestr(out_phase_duration, &s->time_base));
231 }
232 av_log(s, AV_LOG_INFO, "out_phase_end: %s | out_phase_duration: %s\n", av_ts2timestr(s->out_phase_idx[1], &s->time_base), av_ts2timestr(out_phase_duration, &s->time_base));
233 }
234 s->is_out_phase = 0;
235 }
236 }
237
filter_frame(AVFilterLink * inlink,AVFrame * in)238 static int filter_frame(AVFilterLink *inlink, AVFrame *in)
239 {
240 AVFilterContext *ctx = inlink->dst;
241 AudioPhaseMeterContext *s = ctx->priv;
242 AVFilterLink *outlink = s->do_video ? ctx->outputs[1] : NULL;
243 AVFilterLink *aoutlink = ctx->outputs[0];
244 AVDictionary **metadata;
245 const int rc = s->contrast[0];
246 const int gc = s->contrast[1];
247 const int bc = s->contrast[2];
248 float fphase = 0;
249 AVFrame *out;
250 uint8_t *dst;
251 int i;
252 int mono_measurement;
253 int out_phase_measurement;
254 float tolerance = 1.0f - s->tolerance;
255 float angle = cosf(s->angle/180.0f*M_PI);
256
257 if (s->do_video && (!s->out || s->out->width != outlink->w ||
258 s->out->height != outlink->h)) {
259 av_frame_free(&s->out);
260 s->out = ff_get_video_buffer(outlink, outlink->w, outlink->h);
261 if (!s->out) {
262 av_frame_free(&in);
263 return AVERROR(ENOMEM);
264 }
265
266 out = s->out;
267 for (i = 0; i < outlink->h; i++)
268 memset(out->data[0] + i * out->linesize[0], 0, outlink->w * 4);
269 } else if (s->do_video) {
270 out = s->out;
271 for (i = outlink->h - 1; i >= 10; i--)
272 memmove(out->data[0] + (i ) * out->linesize[0],
273 out->data[0] + (i-1) * out->linesize[0],
274 outlink->w * 4);
275 for (i = 0; i < outlink->w; i++)
276 AV_WL32(out->data[0] + i * 4, 0);
277 }
278
279 for (i = 0; i < in->nb_samples; i++) {
280 const float *src = (float *)in->data[0] + i * 2;
281 const float f = src[0] * src[1] / (src[0]*src[0] + src[1] * src[1]) * 2;
282 const float phase = isnan(f) ? 1 : f;
283 const int x = get_x(phase, s->w);
284
285 if (s->do_video) {
286 dst = out->data[0] + x * 4;
287 dst[0] = FFMIN(255, dst[0] + rc);
288 dst[1] = FFMIN(255, dst[1] + gc);
289 dst[2] = FFMIN(255, dst[2] + bc);
290 dst[3] = 255;
291 }
292 fphase += phase;
293 }
294 fphase /= in->nb_samples;
295 s->phase = fphase;
296
297 if (s->do_video) {
298 if (s->draw_median_phase) {
299 dst = out->data[0] + get_x(fphase, s->w) * 4;
300 AV_WL32(dst, AV_RL32(s->mpc));
301 }
302
303 for (i = 1; i < 10 && i < outlink->h; i++)
304 memcpy(out->data[0] + i * out->linesize[0], out->data[0], outlink->w * 4);
305 }
306
307 metadata = &in->metadata;
308 if (metadata) {
309 uint8_t value[128];
310
311 snprintf(value, sizeof(value), "%f", fphase);
312 add_metadata(in, "phase", value);
313 }
314
315 if (s->do_phasing_detection) {
316 s->time_base = inlink->time_base;
317 s->frame_end = in->pts + av_rescale_q(in->nb_samples,
318 (AVRational){ 1, in->sample_rate }, inlink->time_base);
319
320 mono_measurement = (tolerance - fphase) < FLT_EPSILON;
321 out_phase_measurement = (angle - fphase) > FLT_EPSILON;
322
323 update_mono_detection(s, in, mono_measurement);
324 update_out_phase_detection(s, in, out_phase_measurement);
325 }
326
327 if (s->do_video) {
328 AVFrame *clone;
329
330 s->out->pts = in->pts;
331 clone = av_frame_clone(s->out);
332 if (!clone)
333 return AVERROR(ENOMEM);
334 ff_filter_frame(outlink, clone);
335 }
336 return ff_filter_frame(aoutlink, in);
337 }
338
uninit(AVFilterContext * ctx)339 static av_cold void uninit(AVFilterContext *ctx)
340 {
341 AudioPhaseMeterContext *s = ctx->priv;
342
343 if (s->do_phasing_detection) {
344 update_mono_detection(s, NULL, 0);
345 update_out_phase_detection(s, NULL, 0);
346 }
347 av_frame_free(&s->out);
348 }
349
init(AVFilterContext * ctx)350 static av_cold int init(AVFilterContext *ctx)
351 {
352 AudioPhaseMeterContext *s = ctx->priv;
353 AVFilterPad pad;
354 int ret;
355
356 pad = (AVFilterPad){
357 .name = "out0",
358 .type = AVMEDIA_TYPE_AUDIO,
359 };
360 ret = ff_insert_outpad(ctx, 0, &pad);
361 if (ret < 0)
362 return ret;
363
364 if (s->do_video) {
365 pad = (AVFilterPad){
366 .name = "out1",
367 .type = AVMEDIA_TYPE_VIDEO,
368 .config_props = config_video_output,
369 };
370 ret = ff_insert_outpad(ctx, 1, &pad);
371 if (ret < 0)
372 return ret;
373 }
374
375 return 0;
376 }
377
378 static const AVFilterPad inputs[] = {
379 {
380 .name = "default",
381 .type = AVMEDIA_TYPE_AUDIO,
382 .config_props = config_input,
383 .filter_frame = filter_frame,
384 },
385 { NULL }
386 };
387
388 AVFilter ff_avf_aphasemeter = {
389 .name = "aphasemeter",
390 .description = NULL_IF_CONFIG_SMALL("Convert input audio to phase meter video output."),
391 .init = init,
392 .uninit = uninit,
393 .query_formats = query_formats,
394 .priv_size = sizeof(AudioPhaseMeterContext),
395 .inputs = inputs,
396 .outputs = NULL,
397 .priv_class = &aphasemeter_class,
398 .flags = AVFILTER_FLAG_DYNAMIC_OUTPUTS,
399 };
400