| Line | Branch | Exec | Source |
|---|---|---|---|
| 1 | /* | ||
| 2 | * Audio Mix Filter | ||
| 3 | * Copyright (c) 2012 Justin Ruggles <justin.ruggles@gmail.com> | ||
| 4 | * | ||
| 5 | * This file is part of FFmpeg. | ||
| 6 | * | ||
| 7 | * FFmpeg is free software; you can redistribute it and/or | ||
| 8 | * modify it under the terms of the GNU Lesser General Public | ||
| 9 | * License as published by the Free Software Foundation; either | ||
| 10 | * version 2.1 of the License, or (at your option) any later version. | ||
| 11 | * | ||
| 12 | * FFmpeg is distributed in the hope that it will be useful, | ||
| 13 | * but WITHOUT ANY WARRANTY; without even the implied warranty of | ||
| 14 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU | ||
| 15 | * Lesser General Public License for more details. | ||
| 16 | * | ||
| 17 | * You should have received a copy of the GNU Lesser General Public | ||
| 18 | * License along with FFmpeg; if not, write to the Free Software | ||
| 19 | * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA | ||
| 20 | */ | ||
| 21 | |||
| 22 | /** | ||
| 23 | * @file | ||
| 24 | * Audio Mix Filter | ||
| 25 | * | ||
| 26 | * Mixes audio from multiple sources into a single output. The channel layout, | ||
| 27 | * sample rate, and sample format will be the same for all inputs and the | ||
| 28 | * output. | ||
| 29 | */ | ||
| 30 | |||
| 31 | #include "libavutil/attributes.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/common.h" | ||
| 37 | #include "libavutil/eval.h" | ||
| 38 | #include "libavutil/float_dsp.h" | ||
| 39 | #include "libavutil/mathematics.h" | ||
| 40 | #include "libavutil/mem.h" | ||
| 41 | #include "libavutil/opt.h" | ||
| 42 | #include "libavutil/samplefmt.h" | ||
| 43 | |||
| 44 | #include "audio.h" | ||
| 45 | #include "avfilter.h" | ||
| 46 | #include "filters.h" | ||
| 47 | |||
| 48 | #define INPUT_ON 1 /**< input is active */ | ||
| 49 | #define INPUT_EOF 2 /**< input has reached EOF (may still be active) */ | ||
| 50 | |||
| 51 | #define DURATION_LONGEST 0 | ||
| 52 | #define DURATION_SHORTEST 1 | ||
| 53 | #define DURATION_FIRST 2 | ||
| 54 | |||
| 55 | |||
| 56 | typedef struct FrameInfo { | ||
| 57 | int nb_samples; | ||
| 58 | int64_t pts; | ||
| 59 | struct FrameInfo *next; | ||
| 60 | } FrameInfo; | ||
| 61 | |||
| 62 | /** | ||
| 63 | * Linked list used to store timestamps and frame sizes of all frames in the | ||
| 64 | * FIFO for the first input. | ||
| 65 | * | ||
| 66 | * This is needed to keep timestamps synchronized for the case where multiple | ||
| 67 | * input frames are pushed to the filter for processing before a frame is | ||
| 68 | * requested by the output link. | ||
| 69 | */ | ||
| 70 | typedef struct FrameList { | ||
| 71 | int nb_frames; | ||
| 72 | int nb_samples; | ||
| 73 | FrameInfo *list; | ||
| 74 | FrameInfo *end; | ||
| 75 | } FrameList; | ||
| 76 | |||
| 77 | 947 | static void frame_list_clear(FrameList *frame_list) | |
| 78 | { | ||
| 79 |
2/2✓ Branch 0 taken 941 times.
✓ Branch 1 taken 6 times.
|
947 | if (frame_list) { |
| 80 |
2/2✓ Branch 0 taken 702 times.
✓ Branch 1 taken 941 times.
|
1643 | while (frame_list->list) { |
| 81 | 702 | FrameInfo *info = frame_list->list; | |
| 82 | 702 | frame_list->list = info->next; | |
| 83 | 702 | av_free(info); | |
| 84 | } | ||
| 85 | 941 | frame_list->nb_frames = 0; | |
| 86 | 941 | frame_list->nb_samples = 0; | |
| 87 | 941 | frame_list->end = NULL; | |
| 88 | } | ||
| 89 | 947 | } | |
| 90 | |||
| 91 | 2480 | static int frame_list_next_frame_size(FrameList *frame_list) | |
| 92 | { | ||
| 93 |
2/2✓ Branch 0 taken 233 times.
✓ Branch 1 taken 2247 times.
|
2480 | if (!frame_list->list) |
| 94 | 233 | return 0; | |
| 95 | 2247 | return frame_list->list->nb_samples; | |
| 96 | } | ||
| 97 | |||
| 98 | 1142 | static int64_t frame_list_next_pts(FrameList *frame_list) | |
| 99 | { | ||
| 100 |
2/2✓ Branch 0 taken 233 times.
✓ Branch 1 taken 909 times.
|
1142 | if (!frame_list->list) |
| 101 | 233 | return AV_NOPTS_VALUE; | |
| 102 | 909 | return frame_list->list->pts; | |
| 103 | } | ||
| 104 | |||
| 105 | 1142 | static void frame_list_remove_samples(FrameList *frame_list, int nb_samples) | |
| 106 | { | ||
| 107 |
2/2✓ Branch 0 taken 935 times.
✓ Branch 1 taken 207 times.
|
1142 | if (nb_samples >= frame_list->nb_samples) { |
| 108 | 935 | frame_list_clear(frame_list); | |
| 109 | } else { | ||
| 110 | 207 | int samples = nb_samples; | |
| 111 |
2/2✓ Branch 0 taken 207 times.
✓ Branch 1 taken 207 times.
|
414 | while (samples > 0) { |
| 112 | 207 | FrameInfo *info = frame_list->list; | |
| 113 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 207 times.
|
207 | av_assert0(info); |
| 114 |
2/2✓ Branch 0 taken 201 times.
✓ Branch 1 taken 6 times.
|
207 | if (info->nb_samples <= samples) { |
| 115 | 201 | samples -= info->nb_samples; | |
| 116 | 201 | frame_list->list = info->next; | |
| 117 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 201 times.
|
201 | if (!frame_list->list) |
| 118 | ✗ | frame_list->end = NULL; | |
| 119 | 201 | frame_list->nb_frames--; | |
| 120 | 201 | frame_list->nb_samples -= info->nb_samples; | |
| 121 | 201 | av_free(info); | |
| 122 | } else { | ||
| 123 | 6 | info->nb_samples -= samples; | |
| 124 | 6 | info->pts += samples; | |
| 125 | 6 | frame_list->nb_samples -= samples; | |
| 126 | 6 | samples = 0; | |
| 127 | } | ||
| 128 | } | ||
| 129 | } | ||
| 130 | 1142 | } | |
| 131 | |||
| 132 | 903 | static int frame_list_add_frame(FrameList *frame_list, int nb_samples, int64_t pts) | |
| 133 | { | ||
| 134 | 903 | FrameInfo *info = av_malloc(sizeof(*info)); | |
| 135 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 903 times.
|
903 | if (!info) |
| 136 | ✗ | return AVERROR(ENOMEM); | |
| 137 | 903 | info->nb_samples = nb_samples; | |
| 138 | 903 | info->pts = pts; | |
| 139 | 903 | info->next = NULL; | |
| 140 | |||
| 141 |
2/2✓ Branch 0 taken 702 times.
✓ Branch 1 taken 201 times.
|
903 | if (!frame_list->list) { |
| 142 | 702 | frame_list->list = info; | |
| 143 | 702 | frame_list->end = info; | |
| 144 | } else { | ||
| 145 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 201 times.
|
201 | av_assert0(frame_list->end); |
| 146 | 201 | frame_list->end->next = info; | |
| 147 | 201 | frame_list->end = info; | |
| 148 | } | ||
| 149 | 903 | frame_list->nb_frames++; | |
| 150 | 903 | frame_list->nb_samples += nb_samples; | |
| 151 | |||
| 152 | 903 | return 0; | |
| 153 | } | ||
| 154 | |||
| 155 | /* FIXME: use directly links fifo */ | ||
| 156 | |||
| 157 | typedef struct MixContext { | ||
| 158 | const AVClass *class; /**< class for AVOptions */ | ||
| 159 | AVFloatDSPContext *fdsp; | ||
| 160 | |||
| 161 | int nb_inputs; /**< number of inputs */ | ||
| 162 | int active_inputs; /**< number of input currently active */ | ||
| 163 | int duration_mode; /**< mode for determining duration */ | ||
| 164 | float dropout_transition; /**< transition time when an input drops out */ | ||
| 165 | char *weights_str; /**< string for custom weights for every input */ | ||
| 166 | int normalize; /**< if inputs are scaled */ | ||
| 167 | |||
| 168 | int nb_channels; /**< number of channels */ | ||
| 169 | int sample_rate; /**< sample rate */ | ||
| 170 | int planar; | ||
| 171 | AVAudioFifo **fifos; /**< audio fifo for each input */ | ||
| 172 | uint8_t *input_state; /**< current state of each input */ | ||
| 173 | float *input_scale; /**< mixing scale factor for each input */ | ||
| 174 | float *weights; /**< custom weights for every input */ | ||
| 175 | float weight_sum; /**< sum of custom weights for every input */ | ||
| 176 | float *scale_norm; /**< normalization factor for every input */ | ||
| 177 | int64_t next_pts; /**< calculated pts for next output frame */ | ||
| 178 | FrameList *frame_list; /**< list of frame info for the first input */ | ||
| 179 | } MixContext; | ||
| 180 | |||
| 181 | #define OFFSET(x) offsetof(MixContext, x) | ||
| 182 | #define A AV_OPT_FLAG_AUDIO_PARAM | ||
| 183 | #define F AV_OPT_FLAG_FILTERING_PARAM | ||
| 184 | #define T AV_OPT_FLAG_RUNTIME_PARAM | ||
| 185 | static const AVOption amix_options[] = { | ||
| 186 | { "inputs", "Number of inputs.", | ||
| 187 | OFFSET(nb_inputs), AV_OPT_TYPE_INT, { .i64 = 2 }, 1, INT16_MAX, A|F }, | ||
| 188 | { "duration", "How to determine the end-of-stream.", | ||
| 189 | OFFSET(duration_mode), AV_OPT_TYPE_INT, { .i64 = DURATION_LONGEST }, 0, 2, A|F, .unit = "duration" }, | ||
| 190 | { "longest", "Duration of longest input.", 0, AV_OPT_TYPE_CONST, { .i64 = DURATION_LONGEST }, 0, 0, A|F, .unit = "duration" }, | ||
| 191 | { "shortest", "Duration of shortest input.", 0, AV_OPT_TYPE_CONST, { .i64 = DURATION_SHORTEST }, 0, 0, A|F, .unit = "duration" }, | ||
| 192 | { "first", "Duration of first input.", 0, AV_OPT_TYPE_CONST, { .i64 = DURATION_FIRST }, 0, 0, A|F, .unit = "duration" }, | ||
| 193 | { "dropout_transition", "Transition time, in seconds, for volume " | ||
| 194 | "renormalization when an input stream ends.", | ||
| 195 | OFFSET(dropout_transition), AV_OPT_TYPE_FLOAT, { .dbl = 2.0 }, 0, INT_MAX, A|F }, | ||
| 196 | { "weights", "Set weight for each input.", | ||
| 197 | OFFSET(weights_str), AV_OPT_TYPE_STRING, {.str="1 1"}, 0, 0, A|F|T }, | ||
| 198 | { "normalize", "Scale inputs", | ||
| 199 | OFFSET(normalize), AV_OPT_TYPE_BOOL, {.i64=1}, 0, 1, A|F|T }, | ||
| 200 | { NULL } | ||
| 201 | }; | ||
| 202 | |||
| 203 | AVFILTER_DEFINE_CLASS(amix); | ||
| 204 | |||
| 205 | /** | ||
| 206 | * Update the scaling factors to apply to each input during mixing. | ||
| 207 | * | ||
| 208 | * This balances the full volume range between active inputs and handles | ||
| 209 | * volume transitions when EOF is encountered on an input but mixing continues | ||
| 210 | * with the remaining inputs. | ||
| 211 | */ | ||
| 212 | 1148 | static void calculate_scales(MixContext *s, int nb_samples) | |
| 213 | { | ||
| 214 | 1148 | float weight_sum = 0.f; | |
| 215 | int i; | ||
| 216 | |||
| 217 |
2/2✓ Branch 0 taken 2652 times.
✓ Branch 1 taken 1148 times.
|
3800 | for (i = 0; i < s->nb_inputs; i++) |
| 218 |
2/2✓ Branch 0 taken 2055 times.
✓ Branch 1 taken 597 times.
|
2652 | if (s->input_state[i] & INPUT_ON) |
| 219 | 2055 | weight_sum += FFABS(s->weights[i]); | |
| 220 | |||
| 221 |
2/2✓ Branch 0 taken 2652 times.
✓ Branch 1 taken 1148 times.
|
3800 | for (i = 0; i < s->nb_inputs; i++) { |
| 222 |
2/2✓ Branch 0 taken 2055 times.
✓ Branch 1 taken 597 times.
|
2652 | if (s->input_state[i] & INPUT_ON) { |
| 223 |
2/2✓ Branch 0 taken 346 times.
✓ Branch 1 taken 1709 times.
|
2055 | if (s->scale_norm[i] > weight_sum / FFABS(s->weights[i])) { |
| 224 | 346 | s->scale_norm[i] -= ((s->weight_sum / FFABS(s->weights[i])) / s->nb_inputs) * | |
| 225 | 346 | nb_samples / (s->dropout_transition * s->sample_rate); | |
| 226 |
2/2✓ Branch 0 taken 342 times.
✓ Branch 1 taken 4 times.
|
346 | s->scale_norm[i] = FFMAX(s->scale_norm[i], weight_sum / FFABS(s->weights[i])); |
| 227 | } | ||
| 228 | } | ||
| 229 | } | ||
| 230 | |||
| 231 |
2/2✓ Branch 0 taken 2652 times.
✓ Branch 1 taken 1148 times.
|
3800 | for (i = 0; i < s->nb_inputs; i++) { |
| 232 |
2/2✓ Branch 0 taken 2055 times.
✓ Branch 1 taken 597 times.
|
2652 | if (s->input_state[i] & INPUT_ON) { |
| 233 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 2055 times.
|
2055 | if (!s->normalize) |
| 234 | ✗ | s->input_scale[i] = FFABS(s->weights[i]); | |
| 235 | else | ||
| 236 |
1/2✓ Branch 0 taken 2055 times.
✗ Branch 1 not taken.
|
2055 | s->input_scale[i] = 1.0f / s->scale_norm[i] * FFSIGN(s->weights[i]); |
| 237 | } else { | ||
| 238 | 597 | s->input_scale[i] = 0.0f; | |
| 239 | } | ||
| 240 | } | ||
| 241 | 1148 | } | |
| 242 | |||
| 243 | 6 | static int config_output(AVFilterLink *outlink) | |
| 244 | { | ||
| 245 | 6 | AVFilterContext *ctx = outlink->src; | |
| 246 | 6 | MixContext *s = ctx->priv; | |
| 247 | int i; | ||
| 248 | char buf[64]; | ||
| 249 | |||
| 250 | 6 | s->planar = av_sample_fmt_is_planar(outlink->format); | |
| 251 | 6 | s->sample_rate = outlink->sample_rate; | |
| 252 | 6 | outlink->time_base = (AVRational){ 1, outlink->sample_rate }; | |
| 253 | 6 | s->next_pts = AV_NOPTS_VALUE; | |
| 254 | |||
| 255 | 6 | s->frame_list = av_mallocz(sizeof(*s->frame_list)); | |
| 256 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 6 times.
|
6 | if (!s->frame_list) |
| 257 | ✗ | return AVERROR(ENOMEM); | |
| 258 | |||
| 259 | 6 | s->fifos = av_calloc(s->nb_inputs, sizeof(*s->fifos)); | |
| 260 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 6 times.
|
6 | if (!s->fifos) |
| 261 | ✗ | return AVERROR(ENOMEM); | |
| 262 | |||
| 263 | 6 | s->nb_channels = outlink->ch_layout.nb_channels; | |
| 264 |
2/2✓ Branch 0 taken 13 times.
✓ Branch 1 taken 6 times.
|
19 | for (i = 0; i < s->nb_inputs; i++) { |
| 265 | 13 | s->fifos[i] = av_audio_fifo_alloc(outlink->format, s->nb_channels, 1024); | |
| 266 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 13 times.
|
13 | if (!s->fifos[i]) |
| 267 | ✗ | return AVERROR(ENOMEM); | |
| 268 | } | ||
| 269 | |||
| 270 | 6 | s->input_state = av_malloc(s->nb_inputs); | |
| 271 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 6 times.
|
6 | if (!s->input_state) |
| 272 | ✗ | return AVERROR(ENOMEM); | |
| 273 | 6 | memset(s->input_state, INPUT_ON, s->nb_inputs); | |
| 274 | 6 | s->active_inputs = s->nb_inputs; | |
| 275 | |||
| 276 | 6 | s->input_scale = av_calloc(s->nb_inputs, sizeof(*s->input_scale)); | |
| 277 | 6 | s->scale_norm = av_calloc(s->nb_inputs, sizeof(*s->scale_norm)); | |
| 278 |
2/4✓ Branch 0 taken 6 times.
✗ Branch 1 not taken.
✗ Branch 2 not taken.
✓ Branch 3 taken 6 times.
|
6 | if (!s->input_scale || !s->scale_norm) |
| 279 | ✗ | return AVERROR(ENOMEM); | |
| 280 |
2/2✓ Branch 0 taken 13 times.
✓ Branch 1 taken 6 times.
|
19 | for (i = 0; i < s->nb_inputs; i++) |
| 281 | 13 | s->scale_norm[i] = s->weight_sum / FFABS(s->weights[i]); | |
| 282 | 6 | calculate_scales(s, 0); | |
| 283 | |||
| 284 | 6 | av_channel_layout_describe(&outlink->ch_layout, buf, sizeof(buf)); | |
| 285 | |||
| 286 | 6 | av_log(ctx, AV_LOG_VERBOSE, | |
| 287 | "inputs:%d fmt:%s srate:%d cl:%s\n", s->nb_inputs, | ||
| 288 | 6 | av_get_sample_fmt_name(outlink->format), outlink->sample_rate, buf); | |
| 289 | |||
| 290 | 6 | return 0; | |
| 291 | } | ||
| 292 | |||
| 293 | /** | ||
| 294 | * Read samples from the input FIFOs, mix, and write to the output link. | ||
| 295 | */ | ||
| 296 | 1525 | static int output_frame(AVFilterLink *outlink) | |
| 297 | { | ||
| 298 | 1525 | AVFilterContext *ctx = outlink->src; | |
| 299 | 1525 | MixContext *s = ctx->priv; | |
| 300 | AVFrame *out_buf, *in_buf; | ||
| 301 | int nb_samples, ns, i; | ||
| 302 | |||
| 303 |
1/2✓ Branch 0 taken 1525 times.
✗ Branch 1 not taken.
|
1525 | if (s->input_state[0] & INPUT_ON) { |
| 304 | /* first input live: use the corresponding frame size */ | ||
| 305 | 1525 | nb_samples = frame_list_next_frame_size(s->frame_list); | |
| 306 |
2/2✓ Branch 0 taken 1948 times.
✓ Branch 1 taken 1142 times.
|
3090 | for (i = 1; i < s->nb_inputs; i++) { |
| 307 |
2/2✓ Branch 0 taken 1351 times.
✓ Branch 1 taken 597 times.
|
1948 | if (s->input_state[i] & INPUT_ON) { |
| 308 | 1351 | ns = av_audio_fifo_size(s->fifos[i]); | |
| 309 |
2/2✓ Branch 0 taken 389 times.
✓ Branch 1 taken 962 times.
|
1351 | if (ns < nb_samples) { |
| 310 |
2/2✓ Branch 0 taken 383 times.
✓ Branch 1 taken 6 times.
|
389 | if (!(s->input_state[i] & INPUT_EOF)) |
| 311 | /* unclosed input with not enough samples */ | ||
| 312 | 383 | return 0; | |
| 313 | /* closed input to drain */ | ||
| 314 | 6 | nb_samples = ns; | |
| 315 | } | ||
| 316 | } | ||
| 317 | } | ||
| 318 | |||
| 319 | 1142 | s->next_pts = frame_list_next_pts(s->frame_list); | |
| 320 | } else { | ||
| 321 | /* first input closed: use the available samples */ | ||
| 322 | ✗ | nb_samples = INT_MAX; | |
| 323 | ✗ | for (i = 1; i < s->nb_inputs; i++) { | |
| 324 | ✗ | if (s->input_state[i] & INPUT_ON) { | |
| 325 | ✗ | ns = av_audio_fifo_size(s->fifos[i]); | |
| 326 | ✗ | nb_samples = FFMIN(nb_samples, ns); | |
| 327 | } | ||
| 328 | } | ||
| 329 | ✗ | if (nb_samples == INT_MAX) { | |
| 330 | ✗ | ff_outlink_set_status(outlink, AVERROR_EOF, s->next_pts); | |
| 331 | ✗ | return 0; | |
| 332 | } | ||
| 333 | } | ||
| 334 | |||
| 335 | 1142 | frame_list_remove_samples(s->frame_list, nb_samples); | |
| 336 | |||
| 337 | 1142 | calculate_scales(s, nb_samples); | |
| 338 | |||
| 339 |
2/2✓ Branch 0 taken 233 times.
✓ Branch 1 taken 909 times.
|
1142 | if (nb_samples == 0) |
| 340 | 233 | return 0; | |
| 341 | |||
| 342 | 909 | out_buf = ff_get_audio_buffer(outlink, nb_samples); | |
| 343 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 909 times.
|
909 | if (!out_buf) |
| 344 | ✗ | return AVERROR(ENOMEM); | |
| 345 | |||
| 346 | 909 | in_buf = ff_get_audio_buffer(outlink, nb_samples); | |
| 347 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 909 times.
|
909 | if (!in_buf) { |
| 348 | ✗ | av_frame_free(&out_buf); | |
| 349 | ✗ | return AVERROR(ENOMEM); | |
| 350 | } | ||
| 351 | |||
| 352 |
2/2✓ Branch 0 taken 2079 times.
✓ Branch 1 taken 909 times.
|
2988 | for (i = 0; i < s->nb_inputs; i++) { |
| 353 |
2/2✓ Branch 0 taken 1519 times.
✓ Branch 1 taken 560 times.
|
2079 | if (s->input_state[i] & INPUT_ON) { |
| 354 | int planes, plane_size, p; | ||
| 355 | |||
| 356 | 1519 | av_audio_fifo_read(s->fifos[i], (void **)in_buf->extended_data, | |
| 357 | nb_samples); | ||
| 358 | |||
| 359 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 1519 times.
|
1519 | planes = s->planar ? s->nb_channels : 1; |
| 360 |
1/2✓ Branch 0 taken 1519 times.
✗ Branch 1 not taken.
|
1519 | plane_size = nb_samples * (s->planar ? 1 : s->nb_channels); |
| 361 | 1519 | plane_size = FFALIGN(plane_size, 16); | |
| 362 | |||
| 363 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 1519 times.
|
1519 | if (out_buf->format == AV_SAMPLE_FMT_FLT || |
| 364 | ✗ | out_buf->format == AV_SAMPLE_FMT_FLTP) { | |
| 365 |
2/2✓ Branch 0 taken 1519 times.
✓ Branch 1 taken 1519 times.
|
3038 | for (p = 0; p < planes; p++) { |
| 366 | 1519 | s->fdsp->vector_fmac_scalar((float *)out_buf->extended_data[p], | |
| 367 | 1519 | (float *) in_buf->extended_data[p], | |
| 368 | 1519 | s->input_scale[i], plane_size); | |
| 369 | } | ||
| 370 | } else { | ||
| 371 | ✗ | for (p = 0; p < planes; p++) { | |
| 372 | ✗ | s->fdsp->vector_dmac_scalar((double *)out_buf->extended_data[p], | |
| 373 | ✗ | (double *) in_buf->extended_data[p], | |
| 374 | ✗ | s->input_scale[i], plane_size); | |
| 375 | } | ||
| 376 | } | ||
| 377 | } | ||
| 378 | } | ||
| 379 | 909 | av_frame_free(&in_buf); | |
| 380 | |||
| 381 | 909 | out_buf->pts = s->next_pts; | |
| 382 | 909 | out_buf->duration = av_rescale_q(out_buf->nb_samples, av_make_q(1, outlink->sample_rate), | |
| 383 | outlink->time_base); | ||
| 384 | |||
| 385 |
1/2✓ Branch 0 taken 909 times.
✗ Branch 1 not taken.
|
909 | if (s->next_pts != AV_NOPTS_VALUE) |
| 386 | 909 | s->next_pts += nb_samples; | |
| 387 | |||
| 388 | 909 | return ff_filter_frame(outlink, out_buf); | |
| 389 | } | ||
| 390 | |||
| 391 | /** | ||
| 392 | * Requests a frame, if needed, from each input link other than the first. | ||
| 393 | */ | ||
| 394 | 955 | static int request_samples(AVFilterContext *ctx, int min_samples) | |
| 395 | { | ||
| 396 | 955 | MixContext *s = ctx->priv; | |
| 397 | int i; | ||
| 398 | |||
| 399 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 955 times.
|
955 | av_assert0(s->nb_inputs > 1); |
| 400 |
1/4✗ Branch 0 not taken.
✓ Branch 1 taken 955 times.
✗ Branch 2 not taken.
✗ Branch 3 not taken.
|
955 | if (min_samples == 1 && s->duration_mode == DURATION_FIRST) |
| 401 | ✗ | min_samples = av_audio_fifo_size(s->fifos[0]); | |
| 402 | |||
| 403 |
2/2✓ Branch 0 taken 1106 times.
✓ Branch 1 taken 11 times.
|
1117 | for (i = 1; i < s->nb_inputs; i++) { |
| 404 |
2/2✓ Branch 0 taken 1096 times.
✓ Branch 1 taken 10 times.
|
1106 | if (!(s->input_state[i] & INPUT_ON) || |
| 405 |
2/2✓ Branch 0 taken 2 times.
✓ Branch 1 taken 1094 times.
|
1096 | (s->input_state[i] & INPUT_EOF)) |
| 406 | 12 | continue; | |
| 407 |
2/2✓ Branch 1 taken 150 times.
✓ Branch 2 taken 944 times.
|
1094 | if (av_audio_fifo_size(s->fifos[i]) >= min_samples) |
| 408 | 150 | continue; | |
| 409 | 944 | ff_inlink_request_frame(ctx->inputs[i]); | |
| 410 | 944 | return 0; | |
| 411 | } | ||
| 412 | 11 | return output_frame(ctx->outputs[0]); | |
| 413 | } | ||
| 414 | |||
| 415 | /** | ||
| 416 | * Calculates the number of active inputs and determines EOF based on the | ||
| 417 | * duration option. | ||
| 418 | * | ||
| 419 | * @return 0 if mixing should continue, or AVERROR_EOF if mixing should stop. | ||
| 420 | */ | ||
| 421 | 3036 | static int calc_active_inputs(MixContext *s) | |
| 422 | { | ||
| 423 | int i; | ||
| 424 | 3036 | int active_inputs = 0; | |
| 425 |
2/2✓ Branch 0 taken 7113 times.
✓ Branch 1 taken 3036 times.
|
10149 | for (i = 0; i < s->nb_inputs; i++) |
| 426 | 7113 | active_inputs += !!(s->input_state[i] & INPUT_ON); | |
| 427 | 3036 | s->active_inputs = active_inputs; | |
| 428 | |||
| 429 |
2/2✓ Branch 0 taken 3032 times.
✓ Branch 1 taken 4 times.
|
3036 | if (!active_inputs || |
| 430 |
4/4✓ Branch 0 taken 351 times.
✓ Branch 1 taken 2681 times.
✓ Branch 2 taken 350 times.
✓ Branch 3 taken 1 times.
|
3032 | (s->duration_mode == DURATION_FIRST && !(s->input_state[0] & INPUT_ON)) || |
| 431 |
1/4✗ Branch 0 not taken.
✓ Branch 1 taken 3031 times.
✗ Branch 2 not taken.
✗ Branch 3 not taken.
|
3031 | (s->duration_mode == DURATION_SHORTEST && active_inputs != s->nb_inputs)) |
| 432 | 5 | return AVERROR_EOF; | |
| 433 | 3031 | return 0; | |
| 434 | } | ||
| 435 | |||
| 436 | 3039 | static int activate(AVFilterContext *ctx) | |
| 437 | { | ||
| 438 | 3039 | AVFilterLink *outlink = ctx->outputs[0]; | |
| 439 | 3039 | MixContext *s = ctx->priv; | |
| 440 | 3039 | AVFrame *buf = NULL; | |
| 441 | int i, ret; | ||
| 442 | |||
| 443 |
4/4✓ Branch 1 taken 3 times.
✓ Branch 2 taken 3036 times.
✓ Branch 4 taken 7 times.
✓ Branch 5 taken 3 times.
|
3046 | FF_FILTER_FORWARD_STATUS_BACK_ALL(outlink, ctx); |
| 444 | |||
| 445 |
2/2✓ Branch 0 taken 7113 times.
✓ Branch 1 taken 3036 times.
|
10149 | for (i = 0; i < s->nb_inputs; i++) { |
| 446 | 7113 | AVFilterLink *inlink = ctx->inputs[i]; | |
| 447 | |||
| 448 |
2/2✓ Branch 1 taken 1514 times.
✓ Branch 2 taken 5599 times.
|
7113 | if ((ret = ff_inlink_consume_frame(ctx->inputs[i], &buf)) > 0) { |
| 449 |
2/2✓ Branch 0 taken 903 times.
✓ Branch 1 taken 611 times.
|
1514 | if (i == 0) { |
| 450 | 903 | int64_t pts = av_rescale_q(buf->pts, inlink->time_base, | |
| 451 | outlink->time_base); | ||
| 452 | 903 | ret = frame_list_add_frame(s->frame_list, buf->nb_samples, pts); | |
| 453 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 903 times.
|
903 | if (ret < 0) { |
| 454 | ✗ | av_frame_free(&buf); | |
| 455 | ✗ | return ret; | |
| 456 | } | ||
| 457 | } | ||
| 458 | |||
| 459 | 1514 | ret = av_audio_fifo_write(s->fifos[i], (void **)buf->extended_data, | |
| 460 | 1514 | buf->nb_samples); | |
| 461 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 1514 times.
|
1514 | if (ret < 0) { |
| 462 | ✗ | av_frame_free(&buf); | |
| 463 | ✗ | return ret; | |
| 464 | } | ||
| 465 | |||
| 466 | 1514 | av_frame_free(&buf); | |
| 467 | |||
| 468 | 1514 | ret = output_frame(outlink); | |
| 469 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 1514 times.
|
1514 | if (ret < 0) |
| 470 | ✗ | return ret; | |
| 471 | } | ||
| 472 | } | ||
| 473 | |||
| 474 |
2/2✓ Branch 0 taken 7113 times.
✓ Branch 1 taken 3036 times.
|
10149 | for (i = 0; i < s->nb_inputs; i++) { |
| 475 | int64_t pts; | ||
| 476 | int status; | ||
| 477 | |||
| 478 |
2/2✓ Branch 1 taken 1315 times.
✓ Branch 2 taken 5798 times.
|
7113 | if (ff_inlink_acknowledge_status(ctx->inputs[i], &status, &pts)) { |
| 479 |
1/2✓ Branch 0 taken 1315 times.
✗ Branch 1 not taken.
|
1315 | if (status == AVERROR_EOF) { |
| 480 | 1315 | s->input_state[i] |= INPUT_EOF; | |
| 481 |
2/2✓ Branch 1 taken 1301 times.
✓ Branch 2 taken 14 times.
|
1315 | if (av_audio_fifo_size(s->fifos[i]) == 0) { |
| 482 | 1301 | s->input_state[i] &= ~INPUT_ON; | |
| 483 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 1301 times.
|
1301 | if (s->nb_inputs == 1) { |
| 484 | ✗ | ff_outlink_set_status(outlink, status, pts); | |
| 485 | ✗ | return 0; | |
| 486 | } | ||
| 487 | } | ||
| 488 | } | ||
| 489 | } | ||
| 490 | } | ||
| 491 | |||
| 492 |
2/2✓ Branch 1 taken 5 times.
✓ Branch 2 taken 3031 times.
|
3036 | if (calc_active_inputs(s)) { |
| 493 | 5 | ff_outlink_set_status(outlink, AVERROR_EOF, s->next_pts); | |
| 494 | 5 | return 0; | |
| 495 | } | ||
| 496 | |||
| 497 |
2/2✓ Branch 1 taken 2126 times.
✓ Branch 2 taken 905 times.
|
3031 | if (ff_outlink_frame_wanted(outlink)) { |
| 498 | int wanted_samples; | ||
| 499 | |||
| 500 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 2126 times.
|
2126 | if (!(s->input_state[0] & INPUT_ON)) |
| 501 | ✗ | return request_samples(ctx, 1); | |
| 502 | |||
| 503 |
2/2✓ Branch 0 taken 1171 times.
✓ Branch 1 taken 955 times.
|
2126 | if (s->frame_list->nb_frames == 0) { |
| 504 | 1171 | ff_inlink_request_frame(ctx->inputs[0]); | |
| 505 | 1171 | return 0; | |
| 506 | } | ||
| 507 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 955 times.
|
955 | av_assert0(s->frame_list->nb_frames > 0); |
| 508 | |||
| 509 | 955 | wanted_samples = frame_list_next_frame_size(s->frame_list); | |
| 510 | |||
| 511 | 955 | return request_samples(ctx, wanted_samples); | |
| 512 | } | ||
| 513 | |||
| 514 | 905 | return 0; | |
| 515 | } | ||
| 516 | |||
| 517 | 12 | static void parse_weights(AVFilterContext *ctx) | |
| 518 | { | ||
| 519 | 12 | MixContext *s = ctx->priv; | |
| 520 | 12 | float last_weight = 1.f; | |
| 521 | char *p; | ||
| 522 | int i; | ||
| 523 | |||
| 524 | 12 | s->weight_sum = 0.f; | |
| 525 | 12 | p = s->weights_str; | |
| 526 |
1/2✓ Branch 0 taken 24 times.
✗ Branch 1 not taken.
|
24 | for (i = 0; i < s->nb_inputs; i++) { |
| 527 | 24 | last_weight = av_strtod(p, &p); | |
| 528 | 24 | s->weights[i] = last_weight; | |
| 529 | 24 | s->weight_sum += FFABS(last_weight); | |
| 530 |
3/4✓ Branch 0 taken 24 times.
✗ Branch 1 not taken.
✓ Branch 2 taken 12 times.
✓ Branch 3 taken 12 times.
|
24 | if (p && *p) { |
| 531 | 12 | p++; | |
| 532 | } else { | ||
| 533 | 12 | i++; | |
| 534 | 12 | break; | |
| 535 | } | ||
| 536 | } | ||
| 537 | |||
| 538 |
2/2✓ Branch 0 taken 2 times.
✓ Branch 1 taken 12 times.
|
14 | for (; i < s->nb_inputs; i++) { |
| 539 | 2 | s->weights[i] = last_weight; | |
| 540 | 2 | s->weight_sum += FFABS(last_weight); | |
| 541 | } | ||
| 542 | 12 | } | |
| 543 | |||
| 544 | 12 | static av_cold int init(AVFilterContext *ctx) | |
| 545 | { | ||
| 546 | 12 | MixContext *s = ctx->priv; | |
| 547 | int i, ret; | ||
| 548 | |||
| 549 |
2/2✓ Branch 0 taken 26 times.
✓ Branch 1 taken 12 times.
|
38 | for (i = 0; i < s->nb_inputs; i++) { |
| 550 | 26 | AVFilterPad pad = { 0 }; | |
| 551 | |||
| 552 | 26 | pad.type = AVMEDIA_TYPE_AUDIO; | |
| 553 | 26 | pad.name = av_asprintf("input%d", i); | |
| 554 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 26 times.
|
26 | if (!pad.name) |
| 555 | ✗ | return AVERROR(ENOMEM); | |
| 556 | |||
| 557 |
1/2✗ Branch 1 not taken.
✓ Branch 2 taken 26 times.
|
26 | if ((ret = ff_append_inpad_free_name(ctx, &pad)) < 0) |
| 558 | ✗ | return ret; | |
| 559 | } | ||
| 560 | |||
| 561 | 12 | s->fdsp = avpriv_float_dsp_alloc(0); | |
| 562 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 12 times.
|
12 | if (!s->fdsp) |
| 563 | ✗ | return AVERROR(ENOMEM); | |
| 564 | |||
| 565 | 12 | s->weights = av_calloc(s->nb_inputs, sizeof(*s->weights)); | |
| 566 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 12 times.
|
12 | if (!s->weights) |
| 567 | ✗ | return AVERROR(ENOMEM); | |
| 568 | |||
| 569 | 12 | parse_weights(ctx); | |
| 570 | |||
| 571 | 12 | return 0; | |
| 572 | } | ||
| 573 | |||
| 574 | 12 | static av_cold void uninit(AVFilterContext *ctx) | |
| 575 | { | ||
| 576 | int i; | ||
| 577 | 12 | MixContext *s = ctx->priv; | |
| 578 | |||
| 579 |
2/2✓ Branch 0 taken 6 times.
✓ Branch 1 taken 6 times.
|
12 | if (s->fifos) { |
| 580 |
2/2✓ Branch 0 taken 13 times.
✓ Branch 1 taken 6 times.
|
19 | for (i = 0; i < s->nb_inputs; i++) |
| 581 | 13 | av_audio_fifo_free(s->fifos[i]); | |
| 582 | 6 | av_freep(&s->fifos); | |
| 583 | } | ||
| 584 | 12 | frame_list_clear(s->frame_list); | |
| 585 | 12 | av_freep(&s->frame_list); | |
| 586 | 12 | av_freep(&s->input_state); | |
| 587 | 12 | av_freep(&s->input_scale); | |
| 588 | 12 | av_freep(&s->scale_norm); | |
| 589 | 12 | av_freep(&s->weights); | |
| 590 | 12 | av_freep(&s->fdsp); | |
| 591 | 12 | } | |
| 592 | |||
| 593 | ✗ | static int process_command(AVFilterContext *ctx, const char *cmd, const char *args, | |
| 594 | char *res, int res_len, int flags) | ||
| 595 | { | ||
| 596 | ✗ | MixContext *s = ctx->priv; | |
| 597 | int ret; | ||
| 598 | |||
| 599 | ✗ | ret = ff_filter_process_command(ctx, cmd, args, res, res_len, flags); | |
| 600 | ✗ | if (ret < 0) | |
| 601 | ✗ | return ret; | |
| 602 | |||
| 603 | ✗ | parse_weights(ctx); | |
| 604 | ✗ | for (int i = 0; i < s->nb_inputs; i++) | |
| 605 | ✗ | s->scale_norm[i] = s->weight_sum / FFABS(s->weights[i]); | |
| 606 | ✗ | calculate_scales(s, 0); | |
| 607 | |||
| 608 | ✗ | return 0; | |
| 609 | } | ||
| 610 | |||
| 611 | static const AVFilterPad avfilter_af_amix_outputs[] = { | ||
| 612 | { | ||
| 613 | .name = "default", | ||
| 614 | .type = AVMEDIA_TYPE_AUDIO, | ||
| 615 | .config_props = config_output, | ||
| 616 | }, | ||
| 617 | }; | ||
| 618 | |||
| 619 | const FFFilter ff_af_amix = { | ||
| 620 | .p.name = "amix", | ||
| 621 | .p.description = NULL_IF_CONFIG_SMALL("Audio mixing."), | ||
| 622 | .p.priv_class = &amix_class, | ||
| 623 | .p.inputs = NULL, | ||
| 624 | .p.flags = AVFILTER_FLAG_DYNAMIC_INPUTS, | ||
| 625 | .priv_size = sizeof(MixContext), | ||
| 626 | .init = init, | ||
| 627 | .uninit = uninit, | ||
| 628 | .activate = activate, | ||
| 629 | FILTER_OUTPUTS(avfilter_af_amix_outputs), | ||
| 630 | FILTER_SAMPLEFMTS(AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_FLTP, | ||
| 631 | AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_DBLP), | ||
| 632 | .process_command = process_command, | ||
| 633 | }; | ||
| 634 |