summaryrefslogtreecommitdiff
path: root/libavfilter/af_afade.c (plain)
blob: 9acadc51c50353e1ece23d1b7f05f0728b0769a0
1/*
2 * Copyright (c) 2013-2015 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 * fade audio filter
24 */
25
26#include "libavutil/audio_fifo.h"
27#include "libavutil/opt.h"
28#include "audio.h"
29#include "avfilter.h"
30#include "internal.h"
31
32typedef struct {
33 const AVClass *class;
34 int type;
35 int curve, curve2;
36 int nb_samples;
37 int64_t start_sample;
38 int64_t duration;
39 int64_t start_time;
40 int overlap;
41 int cf0_eof;
42 int crossfade_is_over;
43 AVAudioFifo *fifo[2];
44 int64_t pts;
45
46 void (*fade_samples)(uint8_t **dst, uint8_t * const *src,
47 int nb_samples, int channels, int direction,
48 int64_t start, int range, int curve);
49 void (*crossfade_samples)(uint8_t **dst, uint8_t * const *cf0,
50 uint8_t * const *cf1,
51 int nb_samples, int channels,
52 int curve0, int curve1);
53} AudioFadeContext;
54
55enum CurveType { TRI, QSIN, ESIN, HSIN, LOG, IPAR, QUA, CUB, SQU, CBR, PAR, EXP, IQSIN, IHSIN, DESE, DESI, NB_CURVES };
56
57#define OFFSET(x) offsetof(AudioFadeContext, x)
58#define FLAGS AV_OPT_FLAG_AUDIO_PARAM|AV_OPT_FLAG_FILTERING_PARAM
59
60static int query_formats(AVFilterContext *ctx)
61{
62 AVFilterFormats *formats;
63 AVFilterChannelLayouts *layouts;
64 static const enum AVSampleFormat sample_fmts[] = {
65 AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_S16P,
66 AV_SAMPLE_FMT_S32, AV_SAMPLE_FMT_S32P,
67 AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_FLTP,
68 AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_DBLP,
69 AV_SAMPLE_FMT_NONE
70 };
71 int ret;
72
73 layouts = ff_all_channel_counts();
74 if (!layouts)
75 return AVERROR(ENOMEM);
76 ret = ff_set_common_channel_layouts(ctx, layouts);
77 if (ret < 0)
78 return ret;
79
80 formats = ff_make_format_list(sample_fmts);
81 if (!formats)
82 return AVERROR(ENOMEM);
83 ret = ff_set_common_formats(ctx, formats);
84 if (ret < 0)
85 return ret;
86
87 formats = ff_all_samplerates();
88 if (!formats)
89 return AVERROR(ENOMEM);
90 return ff_set_common_samplerates(ctx, formats);
91}
92
93static double fade_gain(int curve, int64_t index, int range)
94{
95#define CUBE(a) ((a)*(a)*(a))
96 double gain;
97
98 gain = av_clipd(1.0 * index / range, 0, 1.0);
99
100 switch (curve) {
101 case QSIN:
102 gain = sin(gain * M_PI / 2.0);
103 break;
104 case IQSIN:
105 /* 0.6... = 2 / M_PI */
106 gain = 0.6366197723675814 * asin(gain);
107 break;
108 case ESIN:
109 gain = 1.0 - cos(M_PI / 4.0 * (CUBE(2.0*gain - 1) + 1));
110 break;
111 case HSIN:
112 gain = (1.0 - cos(gain * M_PI)) / 2.0;
113 break;
114 case IHSIN:
115 /* 0.3... = 1 / M_PI */
116 gain = 0.3183098861837907 * acos(1 - 2 * gain);
117 break;
118 case EXP:
119 /* -11.5... = 5*ln(0.1) */
120 gain = exp(-11.512925464970227 * (1 - gain));
121 break;
122 case LOG:
123 gain = av_clipd(1 + 0.2 * log10(gain), 0, 1.0);
124 break;
125 case PAR:
126 gain = 1 - sqrt(1 - gain);
127 break;
128 case IPAR:
129 gain = (1 - (1 - gain) * (1 - gain));
130 break;
131 case QUA:
132 gain *= gain;
133 break;
134 case CUB:
135 gain = CUBE(gain);
136 break;
137 case SQU:
138 gain = sqrt(gain);
139 break;
140 case CBR:
141 gain = cbrt(gain);
142 break;
143 case DESE:
144 gain = gain <= 0.5 ? cbrt(2 * gain) / 2: 1 - cbrt(2 * (1 - gain)) / 2;
145 break;
146 case DESI:
147 gain = gain <= 0.5 ? CUBE(2 * gain) / 2: 1 - CUBE(2 * (1 - gain)) / 2;
148 break;
149 }
150
151 return gain;
152}
153
154#define FADE_PLANAR(name, type) \
155static void fade_samples_## name ##p(uint8_t **dst, uint8_t * const *src, \
156 int nb_samples, int channels, int dir, \
157 int64_t start, int range, int curve) \
158{ \
159 int i, c; \
160 \
161 for (i = 0; i < nb_samples; i++) { \
162 double gain = fade_gain(curve, start + i * dir, range); \
163 for (c = 0; c < channels; c++) { \
164 type *d = (type *)dst[c]; \
165 const type *s = (type *)src[c]; \
166 \
167 d[i] = s[i] * gain; \
168 } \
169 } \
170}
171
172#define FADE(name, type) \
173static void fade_samples_## name (uint8_t **dst, uint8_t * const *src, \
174 int nb_samples, int channels, int dir, \
175 int64_t start, int range, int curve) \
176{ \
177 type *d = (type *)dst[0]; \
178 const type *s = (type *)src[0]; \
179 int i, c, k = 0; \
180 \
181 for (i = 0; i < nb_samples; i++) { \
182 double gain = fade_gain(curve, start + i * dir, range); \
183 for (c = 0; c < channels; c++, k++) \
184 d[k] = s[k] * gain; \
185 } \
186}
187
188FADE_PLANAR(dbl, double)
189FADE_PLANAR(flt, float)
190FADE_PLANAR(s16, int16_t)
191FADE_PLANAR(s32, int32_t)
192
193FADE(dbl, double)
194FADE(flt, float)
195FADE(s16, int16_t)
196FADE(s32, int32_t)
197
198static int config_output(AVFilterLink *outlink)
199{
200 AVFilterContext *ctx = outlink->src;
201 AudioFadeContext *s = ctx->priv;
202
203 switch (outlink->format) {
204 case AV_SAMPLE_FMT_DBL: s->fade_samples = fade_samples_dbl; break;
205 case AV_SAMPLE_FMT_DBLP: s->fade_samples = fade_samples_dblp; break;
206 case AV_SAMPLE_FMT_FLT: s->fade_samples = fade_samples_flt; break;
207 case AV_SAMPLE_FMT_FLTP: s->fade_samples = fade_samples_fltp; break;
208 case AV_SAMPLE_FMT_S16: s->fade_samples = fade_samples_s16; break;
209 case AV_SAMPLE_FMT_S16P: s->fade_samples = fade_samples_s16p; break;
210 case AV_SAMPLE_FMT_S32: s->fade_samples = fade_samples_s32; break;
211 case AV_SAMPLE_FMT_S32P: s->fade_samples = fade_samples_s32p; break;
212 }
213
214 if (s->duration)
215 s->nb_samples = av_rescale(s->duration, outlink->sample_rate, AV_TIME_BASE);
216 if (s->start_time)
217 s->start_sample = av_rescale(s->start_time, outlink->sample_rate, AV_TIME_BASE);
218
219 return 0;
220}
221
222#if CONFIG_AFADE_FILTER
223
224static const AVOption afade_options[] = {
225 { "type", "set the fade direction", OFFSET(type), AV_OPT_TYPE_INT, {.i64 = 0 }, 0, 1, FLAGS, "type" },
226 { "t", "set the fade direction", OFFSET(type), AV_OPT_TYPE_INT, {.i64 = 0 }, 0, 1, FLAGS, "type" },
227 { "in", "fade-in", 0, AV_OPT_TYPE_CONST, {.i64 = 0 }, 0, 0, FLAGS, "type" },
228 { "out", "fade-out", 0, AV_OPT_TYPE_CONST, {.i64 = 1 }, 0, 0, FLAGS, "type" },
229 { "start_sample", "set number of first sample to start fading", OFFSET(start_sample), AV_OPT_TYPE_INT64, {.i64 = 0 }, 0, INT64_MAX, FLAGS },
230 { "ss", "set number of first sample to start fading", OFFSET(start_sample), AV_OPT_TYPE_INT64, {.i64 = 0 }, 0, INT64_MAX, FLAGS },
231 { "nb_samples", "set number of samples for fade duration", OFFSET(nb_samples), AV_OPT_TYPE_INT, {.i64 = 44100}, 1, INT32_MAX, FLAGS },
232 { "ns", "set number of samples for fade duration", OFFSET(nb_samples), AV_OPT_TYPE_INT, {.i64 = 44100}, 1, INT32_MAX, FLAGS },
233 { "start_time", "set time to start fading", OFFSET(start_time), AV_OPT_TYPE_DURATION, {.i64 = 0. }, 0, INT32_MAX, FLAGS },
234 { "st", "set time to start fading", OFFSET(start_time), AV_OPT_TYPE_DURATION, {.i64 = 0. }, 0, INT32_MAX, FLAGS },
235 { "duration", "set fade duration", OFFSET(duration), AV_OPT_TYPE_DURATION, {.i64 = 0. }, 0, INT32_MAX, FLAGS },
236 { "d", "set fade duration", OFFSET(duration), AV_OPT_TYPE_DURATION, {.i64 = 0. }, 0, INT32_MAX, FLAGS },
237 { "curve", "set fade curve type", OFFSET(curve), AV_OPT_TYPE_INT, {.i64 = TRI }, 0, NB_CURVES - 1, FLAGS, "curve" },
238 { "c", "set fade curve type", OFFSET(curve), AV_OPT_TYPE_INT, {.i64 = TRI }, 0, NB_CURVES - 1, FLAGS, "curve" },
239 { "tri", "linear slope", 0, AV_OPT_TYPE_CONST, {.i64 = TRI }, 0, 0, FLAGS, "curve" },
240 { "qsin", "quarter of sine wave", 0, AV_OPT_TYPE_CONST, {.i64 = QSIN }, 0, 0, FLAGS, "curve" },
241 { "esin", "exponential sine wave", 0, AV_OPT_TYPE_CONST, {.i64 = ESIN }, 0, 0, FLAGS, "curve" },
242 { "hsin", "half of sine wave", 0, AV_OPT_TYPE_CONST, {.i64 = HSIN }, 0, 0, FLAGS, "curve" },
243 { "log", "logarithmic", 0, AV_OPT_TYPE_CONST, {.i64 = LOG }, 0, 0, FLAGS, "curve" },
244 { "ipar", "inverted parabola", 0, AV_OPT_TYPE_CONST, {.i64 = IPAR }, 0, 0, FLAGS, "curve" },
245 { "qua", "quadratic", 0, AV_OPT_TYPE_CONST, {.i64 = QUA }, 0, 0, FLAGS, "curve" },
246 { "cub", "cubic", 0, AV_OPT_TYPE_CONST, {.i64 = CUB }, 0, 0, FLAGS, "curve" },
247 { "squ", "square root", 0, AV_OPT_TYPE_CONST, {.i64 = SQU }, 0, 0, FLAGS, "curve" },
248 { "cbr", "cubic root", 0, AV_OPT_TYPE_CONST, {.i64 = CBR }, 0, 0, FLAGS, "curve" },
249 { "par", "parabola", 0, AV_OPT_TYPE_CONST, {.i64 = PAR }, 0, 0, FLAGS, "curve" },
250 { "exp", "exponential", 0, AV_OPT_TYPE_CONST, {.i64 = EXP }, 0, 0, FLAGS, "curve" },
251 { "iqsin", "inverted quarter of sine wave", 0, AV_OPT_TYPE_CONST, {.i64 = IQSIN}, 0, 0, FLAGS, "curve" },
252 { "ihsin", "inverted half of sine wave", 0, AV_OPT_TYPE_CONST, {.i64 = IHSIN}, 0, 0, FLAGS, "curve" },
253 { "dese", "double-exponential seat", 0, AV_OPT_TYPE_CONST, {.i64 = DESE }, 0, 0, FLAGS, "curve" },
254 { "desi", "double-exponential sigmoid", 0, AV_OPT_TYPE_CONST, {.i64 = DESI }, 0, 0, FLAGS, "curve" },
255 { NULL }
256};
257
258AVFILTER_DEFINE_CLASS(afade);
259
260static av_cold int init(AVFilterContext *ctx)
261{
262 AudioFadeContext *s = ctx->priv;
263
264 if (INT64_MAX - s->nb_samples < s->start_sample)
265 return AVERROR(EINVAL);
266
267 return 0;
268}
269
270static int filter_frame(AVFilterLink *inlink, AVFrame *buf)
271{
272 AudioFadeContext *s = inlink->dst->priv;
273 AVFilterLink *outlink = inlink->dst->outputs[0];
274 int nb_samples = buf->nb_samples;
275 AVFrame *out_buf;
276 int64_t cur_sample = av_rescale_q(buf->pts, inlink->time_base, (AVRational){1, inlink->sample_rate});
277
278 if ((!s->type && (s->start_sample + s->nb_samples < cur_sample)) ||
279 ( s->type && (cur_sample + nb_samples < s->start_sample)))
280 return ff_filter_frame(outlink, buf);
281
282 if (av_frame_is_writable(buf)) {
283 out_buf = buf;
284 } else {
285 out_buf = ff_get_audio_buffer(inlink, nb_samples);
286 if (!out_buf)
287 return AVERROR(ENOMEM);
288 av_frame_copy_props(out_buf, buf);
289 }
290
291 if ((!s->type && (cur_sample + nb_samples < s->start_sample)) ||
292 ( s->type && (s->start_sample + s->nb_samples < cur_sample))) {
293 av_samples_set_silence(out_buf->extended_data, 0, nb_samples,
294 av_frame_get_channels(out_buf), out_buf->format);
295 } else {
296 int64_t start;
297
298 if (!s->type)
299 start = cur_sample - s->start_sample;
300 else
301 start = s->start_sample + s->nb_samples - cur_sample;
302
303 s->fade_samples(out_buf->extended_data, buf->extended_data,
304 nb_samples, av_frame_get_channels(buf),
305 s->type ? -1 : 1, start,
306 s->nb_samples, s->curve);
307 }
308
309 if (buf != out_buf)
310 av_frame_free(&buf);
311
312 return ff_filter_frame(outlink, out_buf);
313}
314
315static const AVFilterPad avfilter_af_afade_inputs[] = {
316 {
317 .name = "default",
318 .type = AVMEDIA_TYPE_AUDIO,
319 .filter_frame = filter_frame,
320 },
321 { NULL }
322};
323
324static const AVFilterPad avfilter_af_afade_outputs[] = {
325 {
326 .name = "default",
327 .type = AVMEDIA_TYPE_AUDIO,
328 .config_props = config_output,
329 },
330 { NULL }
331};
332
333AVFilter ff_af_afade = {
334 .name = "afade",
335 .description = NULL_IF_CONFIG_SMALL("Fade in/out input audio."),
336 .query_formats = query_formats,
337 .priv_size = sizeof(AudioFadeContext),
338 .init = init,
339 .inputs = avfilter_af_afade_inputs,
340 .outputs = avfilter_af_afade_outputs,
341 .priv_class = &afade_class,
342 .flags = AVFILTER_FLAG_SUPPORT_TIMELINE_GENERIC,
343};
344
345#endif /* CONFIG_AFADE_FILTER */
346
347#if CONFIG_ACROSSFADE_FILTER
348
349static const AVOption acrossfade_options[] = {
350 { "nb_samples", "set number of samples for cross fade duration", OFFSET(nb_samples), AV_OPT_TYPE_INT, {.i64 = 44100}, 1, INT32_MAX/10, FLAGS },
351 { "ns", "set number of samples for cross fade duration", OFFSET(nb_samples), AV_OPT_TYPE_INT, {.i64 = 44100}, 1, INT32_MAX/10, FLAGS },
352 { "duration", "set cross fade duration", OFFSET(duration), AV_OPT_TYPE_DURATION, {.i64 = 0. }, 0, 60, FLAGS },
353 { "d", "set cross fade duration", OFFSET(duration), AV_OPT_TYPE_DURATION, {.i64 = 0. }, 0, 60, FLAGS },
354 { "overlap", "overlap 1st stream end with 2nd stream start", OFFSET(overlap), AV_OPT_TYPE_BOOL, {.i64 = 1 }, 0, 1, FLAGS },
355 { "o", "overlap 1st stream end with 2nd stream start", OFFSET(overlap), AV_OPT_TYPE_BOOL, {.i64 = 1 }, 0, 1, FLAGS },
356 { "curve1", "set fade curve type for 1st stream", OFFSET(curve), AV_OPT_TYPE_INT, {.i64 = TRI }, 0, NB_CURVES - 1, FLAGS, "curve" },
357 { "c1", "set fade curve type for 1st stream", OFFSET(curve), AV_OPT_TYPE_INT, {.i64 = TRI }, 0, NB_CURVES - 1, FLAGS, "curve" },
358 { "tri", "linear slope", 0, AV_OPT_TYPE_CONST, {.i64 = TRI }, 0, 0, FLAGS, "curve" },
359 { "qsin", "quarter of sine wave", 0, AV_OPT_TYPE_CONST, {.i64 = QSIN }, 0, 0, FLAGS, "curve" },
360 { "esin", "exponential sine wave", 0, AV_OPT_TYPE_CONST, {.i64 = ESIN }, 0, 0, FLAGS, "curve" },
361 { "hsin", "half of sine wave", 0, AV_OPT_TYPE_CONST, {.i64 = HSIN }, 0, 0, FLAGS, "curve" },
362 { "log", "logarithmic", 0, AV_OPT_TYPE_CONST, {.i64 = LOG }, 0, 0, FLAGS, "curve" },
363 { "ipar", "inverted parabola", 0, AV_OPT_TYPE_CONST, {.i64 = IPAR }, 0, 0, FLAGS, "curve" },
364 { "qua", "quadratic", 0, AV_OPT_TYPE_CONST, {.i64 = QUA }, 0, 0, FLAGS, "curve" },
365 { "cub", "cubic", 0, AV_OPT_TYPE_CONST, {.i64 = CUB }, 0, 0, FLAGS, "curve" },
366 { "squ", "square root", 0, AV_OPT_TYPE_CONST, {.i64 = SQU }, 0, 0, FLAGS, "curve" },
367 { "cbr", "cubic root", 0, AV_OPT_TYPE_CONST, {.i64 = CBR }, 0, 0, FLAGS, "curve" },
368 { "par", "parabola", 0, AV_OPT_TYPE_CONST, {.i64 = PAR }, 0, 0, FLAGS, "curve" },
369 { "exp", "exponential", 0, AV_OPT_TYPE_CONST, {.i64 = EXP }, 0, 0, FLAGS, "curve" },
370 { "iqsin", "inverted quarter of sine wave", 0, AV_OPT_TYPE_CONST, {.i64 = IQSIN}, 0, 0, FLAGS, "curve" },
371 { "ihsin", "inverted half of sine wave", 0, AV_OPT_TYPE_CONST, {.i64 = IHSIN}, 0, 0, FLAGS, "curve" },
372 { "dese", "double-exponential seat", 0, AV_OPT_TYPE_CONST, {.i64 = DESE }, 0, 0, FLAGS, "curve" },
373 { "desi", "double-exponential sigmoid", 0, AV_OPT_TYPE_CONST, {.i64 = DESI }, 0, 0, FLAGS, "curve" },
374 { "curve2", "set fade curve type for 2nd stream", OFFSET(curve2), AV_OPT_TYPE_INT, {.i64 = TRI }, 0, NB_CURVES - 1, FLAGS, "curve" },
375 { "c2", "set fade curve type for 2nd stream", OFFSET(curve2), AV_OPT_TYPE_INT, {.i64 = TRI }, 0, NB_CURVES - 1, FLAGS, "curve" },
376 { NULL }
377};
378
379AVFILTER_DEFINE_CLASS(acrossfade);
380
381#define CROSSFADE_PLANAR(name, type) \
382static void crossfade_samples_## name ##p(uint8_t **dst, uint8_t * const *cf0, \
383 uint8_t * const *cf1, \
384 int nb_samples, int channels, \
385 int curve0, int curve1) \
386{ \
387 int i, c; \
388 \
389 for (i = 0; i < nb_samples; i++) { \
390 double gain0 = fade_gain(curve0, nb_samples - 1 - i, nb_samples); \
391 double gain1 = fade_gain(curve1, i, nb_samples); \
392 for (c = 0; c < channels; c++) { \
393 type *d = (type *)dst[c]; \
394 const type *s0 = (type *)cf0[c]; \
395 const type *s1 = (type *)cf1[c]; \
396 \
397 d[i] = s0[i] * gain0 + s1[i] * gain1; \
398 } \
399 } \
400}
401
402#define CROSSFADE(name, type) \
403static void crossfade_samples_## name (uint8_t **dst, uint8_t * const *cf0, \
404 uint8_t * const *cf1, \
405 int nb_samples, int channels, \
406 int curve0, int curve1) \
407{ \
408 type *d = (type *)dst[0]; \
409 const type *s0 = (type *)cf0[0]; \
410 const type *s1 = (type *)cf1[0]; \
411 int i, c, k = 0; \
412 \
413 for (i = 0; i < nb_samples; i++) { \
414 double gain0 = fade_gain(curve0, nb_samples - 1 - i, nb_samples); \
415 double gain1 = fade_gain(curve1, i, nb_samples); \
416 for (c = 0; c < channels; c++, k++) \
417 d[k] = s0[k] * gain0 + s1[k] * gain1; \
418 } \
419}
420
421CROSSFADE_PLANAR(dbl, double)
422CROSSFADE_PLANAR(flt, float)
423CROSSFADE_PLANAR(s16, int16_t)
424CROSSFADE_PLANAR(s32, int32_t)
425
426CROSSFADE(dbl, double)
427CROSSFADE(flt, float)
428CROSSFADE(s16, int16_t)
429CROSSFADE(s32, int32_t)
430
431static int acrossfade_filter_frame(AVFilterLink *inlink, AVFrame *in)
432{
433 AVFilterContext *ctx = inlink->dst;
434 AudioFadeContext *s = ctx->priv;
435 AVFilterLink *outlink = ctx->outputs[0];
436 AVFrame *out, *cf[2] = { NULL };
437 int ret = 0, nb_samples;
438
439 if (s->crossfade_is_over) {
440 in->pts = s->pts;
441 s->pts += av_rescale_q(in->nb_samples,
442 (AVRational){ 1, outlink->sample_rate }, outlink->time_base);
443 return ff_filter_frame(outlink, in);
444 } else if (inlink == ctx->inputs[0]) {
445 av_audio_fifo_write(s->fifo[0], (void **)in->extended_data, in->nb_samples);
446
447 nb_samples = av_audio_fifo_size(s->fifo[0]) - s->nb_samples;
448 if (nb_samples > 0) {
449 out = ff_get_audio_buffer(outlink, nb_samples);
450 if (!out) {
451 ret = AVERROR(ENOMEM);
452 goto fail;
453 }
454 av_audio_fifo_read(s->fifo[0], (void **)out->extended_data, nb_samples);
455 out->pts = s->pts;
456 s->pts += av_rescale_q(nb_samples,
457 (AVRational){ 1, outlink->sample_rate }, outlink->time_base);
458 ret = ff_filter_frame(outlink, out);
459 }
460 } else if (av_audio_fifo_size(s->fifo[1]) < s->nb_samples) {
461 if (!s->overlap && av_audio_fifo_size(s->fifo[0]) > 0) {
462 nb_samples = av_audio_fifo_size(s->fifo[0]);
463
464 cf[0] = ff_get_audio_buffer(outlink, nb_samples);
465 out = ff_get_audio_buffer(outlink, nb_samples);
466 if (!out || !cf[0]) {
467 ret = AVERROR(ENOMEM);
468 goto fail;
469 }
470 av_audio_fifo_read(s->fifo[0], (void **)cf[0]->extended_data, nb_samples);
471
472 s->fade_samples(out->extended_data, cf[0]->extended_data, nb_samples,
473 outlink->channels, -1, nb_samples - 1, nb_samples, s->curve);
474 out->pts = s->pts;
475 s->pts += av_rescale_q(nb_samples,
476 (AVRational){ 1, outlink->sample_rate }, outlink->time_base);
477 ret = ff_filter_frame(outlink, out);
478 if (ret < 0)
479 goto fail;
480 }
481
482 av_audio_fifo_write(s->fifo[1], (void **)in->extended_data, in->nb_samples);
483 } else if (av_audio_fifo_size(s->fifo[1]) >= s->nb_samples) {
484 av_audio_fifo_write(s->fifo[1], (void **)in->extended_data, in->nb_samples);
485
486 if (s->overlap) {
487 cf[0] = ff_get_audio_buffer(outlink, s->nb_samples);
488 cf[1] = ff_get_audio_buffer(outlink, s->nb_samples);
489 out = ff_get_audio_buffer(outlink, s->nb_samples);
490 if (!out || !cf[0] || !cf[1]) {
491 av_frame_free(&out);
492 ret = AVERROR(ENOMEM);
493 goto fail;
494 }
495
496 av_audio_fifo_read(s->fifo[0], (void **)cf[0]->extended_data, s->nb_samples);
497 av_audio_fifo_read(s->fifo[1], (void **)cf[1]->extended_data, s->nb_samples);
498
499 s->crossfade_samples(out->extended_data, cf[0]->extended_data,
500 cf[1]->extended_data,
501 s->nb_samples, av_frame_get_channels(in),
502 s->curve, s->curve2);
503 out->pts = s->pts;
504 s->pts += av_rescale_q(s->nb_samples,
505 (AVRational){ 1, outlink->sample_rate }, outlink->time_base);
506 ret = ff_filter_frame(outlink, out);
507 if (ret < 0)
508 goto fail;
509 } else {
510 out = ff_get_audio_buffer(outlink, s->nb_samples);
511 cf[1] = ff_get_audio_buffer(outlink, s->nb_samples);
512 if (!out || !cf[1]) {
513 ret = AVERROR(ENOMEM);
514 av_frame_free(&out);
515 goto fail;
516 }
517
518 av_audio_fifo_read(s->fifo[1], (void **)cf[1]->extended_data, s->nb_samples);
519
520 s->fade_samples(out->extended_data, cf[1]->extended_data, s->nb_samples,
521 outlink->channels, 1, 0, s->nb_samples, s->curve2);
522 out->pts = s->pts;
523 s->pts += av_rescale_q(s->nb_samples,
524 (AVRational){ 1, outlink->sample_rate }, outlink->time_base);
525 ret = ff_filter_frame(outlink, out);
526 if (ret < 0)
527 goto fail;
528 }
529
530 nb_samples = av_audio_fifo_size(s->fifo[1]);
531 if (nb_samples > 0) {
532 out = ff_get_audio_buffer(outlink, nb_samples);
533 if (!out) {
534 ret = AVERROR(ENOMEM);
535 goto fail;
536 }
537
538 av_audio_fifo_read(s->fifo[1], (void **)out->extended_data, nb_samples);
539 out->pts = s->pts;
540 s->pts += av_rescale_q(nb_samples,
541 (AVRational){ 1, outlink->sample_rate }, outlink->time_base);
542 ret = ff_filter_frame(outlink, out);
543 }
544 s->crossfade_is_over = 1;
545 }
546
547fail:
548 av_frame_free(&in);
549 av_frame_free(&cf[0]);
550 av_frame_free(&cf[1]);
551 return ret;
552}
553
554static int acrossfade_request_frame(AVFilterLink *outlink)
555{
556 AVFilterContext *ctx = outlink->src;
557 AudioFadeContext *s = ctx->priv;
558 int ret = 0;
559
560 if (!s->cf0_eof) {
561 AVFilterLink *cf0 = ctx->inputs[0];
562 ret = ff_request_frame(cf0);
563 if (ret < 0 && ret != AVERROR_EOF)
564 return ret;
565 if (ret == AVERROR_EOF) {
566 s->cf0_eof = 1;
567 ret = 0;
568 }
569 } else {
570 AVFilterLink *cf1 = ctx->inputs[1];
571 int nb_samples = av_audio_fifo_size(s->fifo[1]);
572
573 ret = ff_request_frame(cf1);
574 if (ret == AVERROR_EOF && nb_samples > 0) {
575 AVFrame *out = ff_get_audio_buffer(outlink, nb_samples);
576 if (!out)
577 return AVERROR(ENOMEM);
578
579 av_audio_fifo_read(s->fifo[1], (void **)out->extended_data, nb_samples);
580 ret = ff_filter_frame(outlink, out);
581 }
582 }
583
584 return ret;
585}
586
587static int acrossfade_config_output(AVFilterLink *outlink)
588{
589 AVFilterContext *ctx = outlink->src;
590 AudioFadeContext *s = ctx->priv;
591
592 if (ctx->inputs[0]->sample_rate != ctx->inputs[1]->sample_rate) {
593 av_log(ctx, AV_LOG_ERROR,
594 "Inputs must have the same sample rate "
595 "%d for in0 vs %d for in1\n",
596 ctx->inputs[0]->sample_rate, ctx->inputs[1]->sample_rate);
597 return AVERROR(EINVAL);
598 }
599
600 outlink->sample_rate = ctx->inputs[0]->sample_rate;
601 outlink->time_base = ctx->inputs[0]->time_base;
602 outlink->channel_layout = ctx->inputs[0]->channel_layout;
603 outlink->channels = ctx->inputs[0]->channels;
604
605 switch (outlink->format) {
606 case AV_SAMPLE_FMT_DBL: s->crossfade_samples = crossfade_samples_dbl; break;
607 case AV_SAMPLE_FMT_DBLP: s->crossfade_samples = crossfade_samples_dblp; break;
608 case AV_SAMPLE_FMT_FLT: s->crossfade_samples = crossfade_samples_flt; break;
609 case AV_SAMPLE_FMT_FLTP: s->crossfade_samples = crossfade_samples_fltp; break;
610 case AV_SAMPLE_FMT_S16: s->crossfade_samples = crossfade_samples_s16; break;
611 case AV_SAMPLE_FMT_S16P: s->crossfade_samples = crossfade_samples_s16p; break;
612 case AV_SAMPLE_FMT_S32: s->crossfade_samples = crossfade_samples_s32; break;
613 case AV_SAMPLE_FMT_S32P: s->crossfade_samples = crossfade_samples_s32p; break;
614 }
615
616 config_output(outlink);
617
618 s->fifo[0] = av_audio_fifo_alloc(outlink->format, outlink->channels, s->nb_samples);
619 s->fifo[1] = av_audio_fifo_alloc(outlink->format, outlink->channels, s->nb_samples);
620 if (!s->fifo[0] || !s->fifo[1])
621 return AVERROR(ENOMEM);
622
623 return 0;
624}
625
626static av_cold void uninit(AVFilterContext *ctx)
627{
628 AudioFadeContext *s = ctx->priv;
629
630 av_audio_fifo_free(s->fifo[0]);
631 av_audio_fifo_free(s->fifo[1]);
632}
633
634static const AVFilterPad avfilter_af_acrossfade_inputs[] = {
635 {
636 .name = "crossfade0",
637 .type = AVMEDIA_TYPE_AUDIO,
638 .filter_frame = acrossfade_filter_frame,
639 },
640 {
641 .name = "crossfade1",
642 .type = AVMEDIA_TYPE_AUDIO,
643 .filter_frame = acrossfade_filter_frame,
644 },
645 { NULL }
646};
647
648static const AVFilterPad avfilter_af_acrossfade_outputs[] = {
649 {
650 .name = "default",
651 .type = AVMEDIA_TYPE_AUDIO,
652 .request_frame = acrossfade_request_frame,
653 .config_props = acrossfade_config_output,
654 },
655 { NULL }
656};
657
658AVFilter ff_af_acrossfade = {
659 .name = "acrossfade",
660 .description = NULL_IF_CONFIG_SMALL("Cross fade two input audio streams."),
661 .query_formats = query_formats,
662 .priv_size = sizeof(AudioFadeContext),
663 .uninit = uninit,
664 .priv_class = &acrossfade_class,
665 .inputs = avfilter_af_acrossfade_inputs,
666 .outputs = avfilter_af_acrossfade_outputs,
667};
668
669#endif /* CONFIG_ACROSSFADE_FILTER */
670
generated by cgit at 2024-05-17 03:08:03 (GMT)