| 1 | /* |
| 2 | * Copyright (c) 2013 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/opt.h" |
| 27 | #include "audio.h" |
| 28 | #include "avfilter.h" |
| 29 | #include "internal.h" |
| 30 | |
| 31 | typedef struct { |
| 32 | const AVClass *class; |
| 33 | int type; |
| 34 | int curve; |
| 35 | int nb_samples; |
| 36 | int64_t start_sample; |
| 37 | int64_t duration; |
| 38 | int64_t start_time; |
| 39 | |
| 40 | void (*fade_samples)(uint8_t **dst, uint8_t * const *src, |
| 41 | int nb_samples, int channels, int direction, |
| 42 | int64_t start, int range, int curve); |
| 43 | } AudioFadeContext; |
| 44 | |
| 45 | enum CurveType { TRI, QSIN, ESIN, HSIN, LOG, PAR, QUA, CUB, SQU, CBR }; |
| 46 | |
| 47 | #define OFFSET(x) offsetof(AudioFadeContext, x) |
| 48 | #define FLAGS AV_OPT_FLAG_AUDIO_PARAM|AV_OPT_FLAG_FILTERING_PARAM |
| 49 | |
| 50 | static const AVOption afade_options[] = { |
| 51 | { "type", "set the fade direction", OFFSET(type), AV_OPT_TYPE_INT, {.i64 = 0 }, 0, 1, FLAGS, "type" }, |
| 52 | { "t", "set the fade direction", OFFSET(type), AV_OPT_TYPE_INT, {.i64 = 0 }, 0, 1, FLAGS, "type" }, |
| 53 | { "in", "fade-in", 0, AV_OPT_TYPE_CONST, {.i64 = 0 }, 0, 0, FLAGS, "type" }, |
| 54 | { "out", "fade-out", 0, AV_OPT_TYPE_CONST, {.i64 = 1 }, 0, 0, FLAGS, "type" }, |
| 55 | { "start_sample", "set number of first sample to start fading", OFFSET(start_sample), AV_OPT_TYPE_INT64, {.i64 = 0 }, 0, INT64_MAX, FLAGS }, |
| 56 | { "ss", "set number of first sample to start fading", OFFSET(start_sample), AV_OPT_TYPE_INT64, {.i64 = 0 }, 0, INT64_MAX, FLAGS }, |
| 57 | { "nb_samples", "set number of samples for fade duration", OFFSET(nb_samples), AV_OPT_TYPE_INT, {.i64 = 44100}, 1, INT32_MAX, FLAGS }, |
| 58 | { "ns", "set number of samples for fade duration", OFFSET(nb_samples), AV_OPT_TYPE_INT, {.i64 = 44100}, 1, INT32_MAX, FLAGS }, |
| 59 | { "start_time", "set time to start fading", OFFSET(start_time), AV_OPT_TYPE_DURATION, {.i64 = 0. }, 0, INT32_MAX, FLAGS }, |
| 60 | { "st", "set time to start fading", OFFSET(start_time), AV_OPT_TYPE_DURATION, {.i64 = 0. }, 0, INT32_MAX, FLAGS }, |
| 61 | { "duration", "set fade duration", OFFSET(duration), AV_OPT_TYPE_DURATION, {.i64 = 0. }, 0, INT32_MAX, FLAGS }, |
| 62 | { "d", "set fade duration", OFFSET(duration), AV_OPT_TYPE_DURATION, {.i64 = 0. }, 0, INT32_MAX, FLAGS }, |
| 63 | { "curve", "set fade curve type", OFFSET(curve), AV_OPT_TYPE_INT, {.i64 = TRI }, TRI, CBR, FLAGS, "curve" }, |
| 64 | { "c", "set fade curve type", OFFSET(curve), AV_OPT_TYPE_INT, {.i64 = TRI }, TRI, CBR, FLAGS, "curve" }, |
| 65 | { "tri", "linear slope", 0, AV_OPT_TYPE_CONST, {.i64 = TRI }, 0, 0, FLAGS, "curve" }, |
| 66 | { "qsin", "quarter of sine wave", 0, AV_OPT_TYPE_CONST, {.i64 = QSIN }, 0, 0, FLAGS, "curve" }, |
| 67 | { "esin", "exponential sine wave", 0, AV_OPT_TYPE_CONST, {.i64 = ESIN }, 0, 0, FLAGS, "curve" }, |
| 68 | { "hsin", "half of sine wave", 0, AV_OPT_TYPE_CONST, {.i64 = HSIN }, 0, 0, FLAGS, "curve" }, |
| 69 | { "log", "logarithmic", 0, AV_OPT_TYPE_CONST, {.i64 = LOG }, 0, 0, FLAGS, "curve" }, |
| 70 | { "par", "inverted parabola", 0, AV_OPT_TYPE_CONST, {.i64 = PAR }, 0, 0, FLAGS, "curve" }, |
| 71 | { "qua", "quadratic", 0, AV_OPT_TYPE_CONST, {.i64 = QUA }, 0, 0, FLAGS, "curve" }, |
| 72 | { "cub", "cubic", 0, AV_OPT_TYPE_CONST, {.i64 = CUB }, 0, 0, FLAGS, "curve" }, |
| 73 | { "squ", "square root", 0, AV_OPT_TYPE_CONST, {.i64 = SQU }, 0, 0, FLAGS, "curve" }, |
| 74 | { "cbr", "cubic root", 0, AV_OPT_TYPE_CONST, {.i64 = CBR }, 0, 0, FLAGS, "curve" }, |
| 75 | { NULL } |
| 76 | }; |
| 77 | |
| 78 | AVFILTER_DEFINE_CLASS(afade); |
| 79 | |
| 80 | static av_cold int init(AVFilterContext *ctx) |
| 81 | { |
| 82 | AudioFadeContext *s = ctx->priv; |
| 83 | |
| 84 | if (INT64_MAX - s->nb_samples < s->start_sample) |
| 85 | return AVERROR(EINVAL); |
| 86 | |
| 87 | return 0; |
| 88 | } |
| 89 | |
| 90 | static int query_formats(AVFilterContext *ctx) |
| 91 | { |
| 92 | AVFilterFormats *formats; |
| 93 | AVFilterChannelLayouts *layouts; |
| 94 | static const enum AVSampleFormat sample_fmts[] = { |
| 95 | AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_S16P, |
| 96 | AV_SAMPLE_FMT_S32, AV_SAMPLE_FMT_S32P, |
| 97 | AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_FLTP, |
| 98 | AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_DBLP, |
| 99 | AV_SAMPLE_FMT_NONE |
| 100 | }; |
| 101 | |
| 102 | layouts = ff_all_channel_layouts(); |
| 103 | if (!layouts) |
| 104 | return AVERROR(ENOMEM); |
| 105 | ff_set_common_channel_layouts(ctx, layouts); |
| 106 | |
| 107 | formats = ff_make_format_list(sample_fmts); |
| 108 | if (!formats) |
| 109 | return AVERROR(ENOMEM); |
| 110 | ff_set_common_formats(ctx, formats); |
| 111 | |
| 112 | formats = ff_all_samplerates(); |
| 113 | if (!formats) |
| 114 | return AVERROR(ENOMEM); |
| 115 | ff_set_common_samplerates(ctx, formats); |
| 116 | |
| 117 | return 0; |
| 118 | } |
| 119 | |
| 120 | static double fade_gain(int curve, int64_t index, int range) |
| 121 | { |
| 122 | double gain; |
| 123 | |
| 124 | gain = FFMAX(0.0, FFMIN(1.0, 1.0 * index / range)); |
| 125 | |
| 126 | switch (curve) { |
| 127 | case QSIN: |
| 128 | gain = sin(gain * M_PI / 2.0); |
| 129 | break; |
| 130 | case ESIN: |
| 131 | gain = 1.0 - cos(M_PI / 4.0 * (pow(2.0*gain - 1, 3) + 1)); |
| 132 | break; |
| 133 | case HSIN: |
| 134 | gain = (1.0 - cos(gain * M_PI)) / 2.0; |
| 135 | break; |
| 136 | case LOG: |
| 137 | gain = pow(0.1, (1 - gain) * 5.0); |
| 138 | break; |
| 139 | case PAR: |
| 140 | gain = (1 - (1 - gain) * (1 - gain)); |
| 141 | break; |
| 142 | case QUA: |
| 143 | gain *= gain; |
| 144 | break; |
| 145 | case CUB: |
| 146 | gain = gain * gain * gain; |
| 147 | break; |
| 148 | case SQU: |
| 149 | gain = sqrt(gain); |
| 150 | break; |
| 151 | case CBR: |
| 152 | gain = cbrt(gain); |
| 153 | break; |
| 154 | } |
| 155 | |
| 156 | return gain; |
| 157 | } |
| 158 | |
| 159 | #define FADE_PLANAR(name, type) \ |
| 160 | static void fade_samples_## name ##p(uint8_t **dst, uint8_t * const *src, \ |
| 161 | int nb_samples, int channels, int dir, \ |
| 162 | int64_t start, int range, int curve) \ |
| 163 | { \ |
| 164 | int i, c; \ |
| 165 | \ |
| 166 | for (i = 0; i < nb_samples; i++) { \ |
| 167 | double gain = fade_gain(curve, start + i * dir, range); \ |
| 168 | for (c = 0; c < channels; c++) { \ |
| 169 | type *d = (type *)dst[c]; \ |
| 170 | const type *s = (type *)src[c]; \ |
| 171 | \ |
| 172 | d[i] = s[i] * gain; \ |
| 173 | } \ |
| 174 | } \ |
| 175 | } |
| 176 | |
| 177 | #define FADE(name, type) \ |
| 178 | static void fade_samples_## name (uint8_t **dst, uint8_t * const *src, \ |
| 179 | int nb_samples, int channels, int dir, \ |
| 180 | int64_t start, int range, int curve) \ |
| 181 | { \ |
| 182 | type *d = (type *)dst[0]; \ |
| 183 | const type *s = (type *)src[0]; \ |
| 184 | int i, c, k = 0; \ |
| 185 | \ |
| 186 | for (i = 0; i < nb_samples; i++) { \ |
| 187 | double gain = fade_gain(curve, start + i * dir, range); \ |
| 188 | for (c = 0; c < channels; c++, k++) \ |
| 189 | d[k] = s[k] * gain; \ |
| 190 | } \ |
| 191 | } |
| 192 | |
| 193 | FADE_PLANAR(dbl, double) |
| 194 | FADE_PLANAR(flt, float) |
| 195 | FADE_PLANAR(s16, int16_t) |
| 196 | FADE_PLANAR(s32, int32_t) |
| 197 | |
| 198 | FADE(dbl, double) |
| 199 | FADE(flt, float) |
| 200 | FADE(s16, int16_t) |
| 201 | FADE(s32, int32_t) |
| 202 | |
| 203 | static int config_input(AVFilterLink *inlink) |
| 204 | { |
| 205 | AVFilterContext *ctx = inlink->dst; |
| 206 | AudioFadeContext *s = ctx->priv; |
| 207 | |
| 208 | switch (inlink->format) { |
| 209 | case AV_SAMPLE_FMT_DBL: s->fade_samples = fade_samples_dbl; break; |
| 210 | case AV_SAMPLE_FMT_DBLP: s->fade_samples = fade_samples_dblp; break; |
| 211 | case AV_SAMPLE_FMT_FLT: s->fade_samples = fade_samples_flt; break; |
| 212 | case AV_SAMPLE_FMT_FLTP: s->fade_samples = fade_samples_fltp; break; |
| 213 | case AV_SAMPLE_FMT_S16: s->fade_samples = fade_samples_s16; break; |
| 214 | case AV_SAMPLE_FMT_S16P: s->fade_samples = fade_samples_s16p; break; |
| 215 | case AV_SAMPLE_FMT_S32: s->fade_samples = fade_samples_s32; break; |
| 216 | case AV_SAMPLE_FMT_S32P: s->fade_samples = fade_samples_s32p; break; |
| 217 | } |
| 218 | |
| 219 | if (s->duration) |
| 220 | s->nb_samples = av_rescale(s->duration, inlink->sample_rate, AV_TIME_BASE); |
| 221 | if (s->start_time) |
| 222 | s->start_sample = av_rescale(s->start_time, inlink->sample_rate, AV_TIME_BASE); |
| 223 | |
| 224 | return 0; |
| 225 | } |
| 226 | |
| 227 | static int filter_frame(AVFilterLink *inlink, AVFrame *buf) |
| 228 | { |
| 229 | AudioFadeContext *s = inlink->dst->priv; |
| 230 | AVFilterLink *outlink = inlink->dst->outputs[0]; |
| 231 | int nb_samples = buf->nb_samples; |
| 232 | AVFrame *out_buf; |
| 233 | int64_t cur_sample = av_rescale_q(buf->pts, inlink->time_base, (AVRational){1, inlink->sample_rate}); |
| 234 | |
| 235 | if ((!s->type && (s->start_sample + s->nb_samples < cur_sample)) || |
| 236 | ( s->type && (cur_sample + s->nb_samples < s->start_sample))) |
| 237 | return ff_filter_frame(outlink, buf); |
| 238 | |
| 239 | if (av_frame_is_writable(buf)) { |
| 240 | out_buf = buf; |
| 241 | } else { |
| 242 | out_buf = ff_get_audio_buffer(inlink, nb_samples); |
| 243 | if (!out_buf) |
| 244 | return AVERROR(ENOMEM); |
| 245 | av_frame_copy_props(out_buf, buf); |
| 246 | } |
| 247 | |
| 248 | if ((!s->type && (cur_sample + nb_samples < s->start_sample)) || |
| 249 | ( s->type && (s->start_sample + s->nb_samples < cur_sample))) { |
| 250 | av_samples_set_silence(out_buf->extended_data, 0, nb_samples, |
| 251 | av_frame_get_channels(out_buf), out_buf->format); |
| 252 | } else { |
| 253 | int64_t start; |
| 254 | |
| 255 | if (!s->type) |
| 256 | start = cur_sample - s->start_sample; |
| 257 | else |
| 258 | start = s->start_sample + s->nb_samples - cur_sample; |
| 259 | |
| 260 | s->fade_samples(out_buf->extended_data, buf->extended_data, |
| 261 | nb_samples, av_frame_get_channels(buf), |
| 262 | s->type ? -1 : 1, start, |
| 263 | s->nb_samples, s->curve); |
| 264 | } |
| 265 | |
| 266 | if (buf != out_buf) |
| 267 | av_frame_free(&buf); |
| 268 | |
| 269 | return ff_filter_frame(outlink, out_buf); |
| 270 | } |
| 271 | |
| 272 | static const AVFilterPad avfilter_af_afade_inputs[] = { |
| 273 | { |
| 274 | .name = "default", |
| 275 | .type = AVMEDIA_TYPE_AUDIO, |
| 276 | .filter_frame = filter_frame, |
| 277 | .config_props = config_input, |
| 278 | }, |
| 279 | { NULL } |
| 280 | }; |
| 281 | |
| 282 | static const AVFilterPad avfilter_af_afade_outputs[] = { |
| 283 | { |
| 284 | .name = "default", |
| 285 | .type = AVMEDIA_TYPE_AUDIO, |
| 286 | }, |
| 287 | { NULL } |
| 288 | }; |
| 289 | |
| 290 | AVFilter ff_af_afade = { |
| 291 | .name = "afade", |
| 292 | .description = NULL_IF_CONFIG_SMALL("Fade in/out input audio."), |
| 293 | .query_formats = query_formats, |
| 294 | .priv_size = sizeof(AudioFadeContext), |
| 295 | .init = init, |
| 296 | .inputs = avfilter_af_afade_inputs, |
| 297 | .outputs = avfilter_af_afade_outputs, |
| 298 | .priv_class = &afade_class, |
| 299 | .flags = AVFILTER_FLAG_SUPPORT_TIMELINE_GENERIC, |
| 300 | }; |