| 1 | /* |
| 2 | * This file is part of FFmpeg. |
| 3 | * |
| 4 | * FFmpeg is free software; you can redistribute it and/or |
| 5 | * modify it under the terms of the GNU Lesser General Public |
| 6 | * License as published by the Free Software Foundation; either |
| 7 | * version 2.1 of the License, or (at your option) any later version. |
| 8 | * |
| 9 | * FFmpeg is distributed in the hope that it will be useful, |
| 10 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 11 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
| 12 | * Lesser General Public License for more details. |
| 13 | * |
| 14 | * You should have received a copy of the GNU Lesser General Public |
| 15 | * License along with FFmpeg; if not, write to the Free Software |
| 16 | * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
| 17 | */ |
| 18 | |
| 19 | /** |
| 20 | * @file |
| 21 | * sample format and channel layout conversion audio filter |
| 22 | */ |
| 23 | |
| 24 | #include "libavutil/avassert.h" |
| 25 | #include "libavutil/avstring.h" |
| 26 | #include "libavutil/common.h" |
| 27 | #include "libavutil/dict.h" |
| 28 | #include "libavutil/mathematics.h" |
| 29 | #include "libavutil/opt.h" |
| 30 | |
| 31 | #include "libavresample/avresample.h" |
| 32 | |
| 33 | #include "audio.h" |
| 34 | #include "avfilter.h" |
| 35 | #include "formats.h" |
| 36 | #include "internal.h" |
| 37 | |
| 38 | typedef struct ResampleContext { |
| 39 | const AVClass *class; |
| 40 | AVAudioResampleContext *avr; |
| 41 | AVDictionary *options; |
| 42 | |
| 43 | int64_t next_pts; |
| 44 | int64_t next_in_pts; |
| 45 | |
| 46 | /* set by filter_frame() to signal an output frame to request_frame() */ |
| 47 | int got_output; |
| 48 | } ResampleContext; |
| 49 | |
| 50 | static av_cold int init(AVFilterContext *ctx, AVDictionary **opts) |
| 51 | { |
| 52 | ResampleContext *s = ctx->priv; |
| 53 | const AVClass *avr_class = avresample_get_class(); |
| 54 | AVDictionaryEntry *e = NULL; |
| 55 | |
| 56 | while ((e = av_dict_get(*opts, "", e, AV_DICT_IGNORE_SUFFIX))) { |
| 57 | if (av_opt_find(&avr_class, e->key, NULL, 0, |
| 58 | AV_OPT_SEARCH_FAKE_OBJ | AV_OPT_SEARCH_CHILDREN)) |
| 59 | av_dict_set(&s->options, e->key, e->value, 0); |
| 60 | } |
| 61 | |
| 62 | e = NULL; |
| 63 | while ((e = av_dict_get(s->options, "", e, AV_DICT_IGNORE_SUFFIX))) |
| 64 | av_dict_set(opts, e->key, NULL, 0); |
| 65 | |
| 66 | /* do not allow the user to override basic format options */ |
| 67 | av_dict_set(&s->options, "in_channel_layout", NULL, 0); |
| 68 | av_dict_set(&s->options, "out_channel_layout", NULL, 0); |
| 69 | av_dict_set(&s->options, "in_sample_fmt", NULL, 0); |
| 70 | av_dict_set(&s->options, "out_sample_fmt", NULL, 0); |
| 71 | av_dict_set(&s->options, "in_sample_rate", NULL, 0); |
| 72 | av_dict_set(&s->options, "out_sample_rate", NULL, 0); |
| 73 | |
| 74 | return 0; |
| 75 | } |
| 76 | |
| 77 | static av_cold void uninit(AVFilterContext *ctx) |
| 78 | { |
| 79 | ResampleContext *s = ctx->priv; |
| 80 | |
| 81 | if (s->avr) { |
| 82 | avresample_close(s->avr); |
| 83 | avresample_free(&s->avr); |
| 84 | } |
| 85 | av_dict_free(&s->options); |
| 86 | } |
| 87 | |
| 88 | static int query_formats(AVFilterContext *ctx) |
| 89 | { |
| 90 | AVFilterLink *inlink = ctx->inputs[0]; |
| 91 | AVFilterLink *outlink = ctx->outputs[0]; |
| 92 | |
| 93 | AVFilterFormats *in_formats = ff_all_formats(AVMEDIA_TYPE_AUDIO); |
| 94 | AVFilterFormats *out_formats = ff_all_formats(AVMEDIA_TYPE_AUDIO); |
| 95 | AVFilterFormats *in_samplerates = ff_all_samplerates(); |
| 96 | AVFilterFormats *out_samplerates = ff_all_samplerates(); |
| 97 | AVFilterChannelLayouts *in_layouts = ff_all_channel_layouts(); |
| 98 | AVFilterChannelLayouts *out_layouts = ff_all_channel_layouts(); |
| 99 | |
| 100 | ff_formats_ref(in_formats, &inlink->out_formats); |
| 101 | ff_formats_ref(out_formats, &outlink->in_formats); |
| 102 | |
| 103 | ff_formats_ref(in_samplerates, &inlink->out_samplerates); |
| 104 | ff_formats_ref(out_samplerates, &outlink->in_samplerates); |
| 105 | |
| 106 | ff_channel_layouts_ref(in_layouts, &inlink->out_channel_layouts); |
| 107 | ff_channel_layouts_ref(out_layouts, &outlink->in_channel_layouts); |
| 108 | |
| 109 | return 0; |
| 110 | } |
| 111 | |
| 112 | static int config_output(AVFilterLink *outlink) |
| 113 | { |
| 114 | AVFilterContext *ctx = outlink->src; |
| 115 | AVFilterLink *inlink = ctx->inputs[0]; |
| 116 | ResampleContext *s = ctx->priv; |
| 117 | char buf1[64], buf2[64]; |
| 118 | int ret; |
| 119 | |
| 120 | if (s->avr) { |
| 121 | avresample_close(s->avr); |
| 122 | avresample_free(&s->avr); |
| 123 | } |
| 124 | |
| 125 | if (inlink->channel_layout == outlink->channel_layout && |
| 126 | inlink->sample_rate == outlink->sample_rate && |
| 127 | (inlink->format == outlink->format || |
| 128 | (av_get_channel_layout_nb_channels(inlink->channel_layout) == 1 && |
| 129 | av_get_channel_layout_nb_channels(outlink->channel_layout) == 1 && |
| 130 | av_get_planar_sample_fmt(inlink->format) == |
| 131 | av_get_planar_sample_fmt(outlink->format)))) |
| 132 | return 0; |
| 133 | |
| 134 | if (!(s->avr = avresample_alloc_context())) |
| 135 | return AVERROR(ENOMEM); |
| 136 | |
| 137 | if (s->options) { |
| 138 | AVDictionaryEntry *e = NULL; |
| 139 | while ((e = av_dict_get(s->options, "", e, AV_DICT_IGNORE_SUFFIX))) |
| 140 | av_log(ctx, AV_LOG_VERBOSE, "lavr option: %s=%s\n", e->key, e->value); |
| 141 | |
| 142 | av_opt_set_dict(s->avr, &s->options); |
| 143 | } |
| 144 | |
| 145 | av_opt_set_int(s->avr, "in_channel_layout", inlink ->channel_layout, 0); |
| 146 | av_opt_set_int(s->avr, "out_channel_layout", outlink->channel_layout, 0); |
| 147 | av_opt_set_int(s->avr, "in_sample_fmt", inlink ->format, 0); |
| 148 | av_opt_set_int(s->avr, "out_sample_fmt", outlink->format, 0); |
| 149 | av_opt_set_int(s->avr, "in_sample_rate", inlink ->sample_rate, 0); |
| 150 | av_opt_set_int(s->avr, "out_sample_rate", outlink->sample_rate, 0); |
| 151 | |
| 152 | if ((ret = avresample_open(s->avr)) < 0) |
| 153 | return ret; |
| 154 | |
| 155 | outlink->time_base = (AVRational){ 1, outlink->sample_rate }; |
| 156 | s->next_pts = AV_NOPTS_VALUE; |
| 157 | s->next_in_pts = AV_NOPTS_VALUE; |
| 158 | |
| 159 | av_get_channel_layout_string(buf1, sizeof(buf1), |
| 160 | -1, inlink ->channel_layout); |
| 161 | av_get_channel_layout_string(buf2, sizeof(buf2), |
| 162 | -1, outlink->channel_layout); |
| 163 | av_log(ctx, AV_LOG_VERBOSE, |
| 164 | "fmt:%s srate:%d cl:%s -> fmt:%s srate:%d cl:%s\n", |
| 165 | av_get_sample_fmt_name(inlink ->format), inlink ->sample_rate, buf1, |
| 166 | av_get_sample_fmt_name(outlink->format), outlink->sample_rate, buf2); |
| 167 | |
| 168 | return 0; |
| 169 | } |
| 170 | |
| 171 | static int request_frame(AVFilterLink *outlink) |
| 172 | { |
| 173 | AVFilterContext *ctx = outlink->src; |
| 174 | ResampleContext *s = ctx->priv; |
| 175 | int ret = 0; |
| 176 | |
| 177 | s->got_output = 0; |
| 178 | while (ret >= 0 && !s->got_output) |
| 179 | ret = ff_request_frame(ctx->inputs[0]); |
| 180 | |
| 181 | /* flush the lavr delay buffer */ |
| 182 | if (ret == AVERROR_EOF && s->avr) { |
| 183 | AVFrame *frame; |
| 184 | int nb_samples = avresample_get_out_samples(s->avr, 0); |
| 185 | |
| 186 | if (!nb_samples) |
| 187 | return ret; |
| 188 | |
| 189 | frame = ff_get_audio_buffer(outlink, nb_samples); |
| 190 | if (!frame) |
| 191 | return AVERROR(ENOMEM); |
| 192 | |
| 193 | ret = avresample_convert(s->avr, frame->extended_data, |
| 194 | frame->linesize[0], nb_samples, |
| 195 | NULL, 0, 0); |
| 196 | if (ret <= 0) { |
| 197 | av_frame_free(&frame); |
| 198 | return (ret == 0) ? AVERROR_EOF : ret; |
| 199 | } |
| 200 | |
| 201 | frame->pts = s->next_pts; |
| 202 | return ff_filter_frame(outlink, frame); |
| 203 | } |
| 204 | return ret; |
| 205 | } |
| 206 | |
| 207 | static int filter_frame(AVFilterLink *inlink, AVFrame *in) |
| 208 | { |
| 209 | AVFilterContext *ctx = inlink->dst; |
| 210 | ResampleContext *s = ctx->priv; |
| 211 | AVFilterLink *outlink = ctx->outputs[0]; |
| 212 | int ret; |
| 213 | |
| 214 | if (s->avr) { |
| 215 | AVFrame *out; |
| 216 | int delay, nb_samples; |
| 217 | |
| 218 | /* maximum possible samples lavr can output */ |
| 219 | delay = avresample_get_delay(s->avr); |
| 220 | nb_samples = avresample_get_out_samples(s->avr, in->nb_samples); |
| 221 | |
| 222 | out = ff_get_audio_buffer(outlink, nb_samples); |
| 223 | if (!out) { |
| 224 | ret = AVERROR(ENOMEM); |
| 225 | goto fail; |
| 226 | } |
| 227 | |
| 228 | ret = avresample_convert(s->avr, out->extended_data, out->linesize[0], |
| 229 | nb_samples, in->extended_data, in->linesize[0], |
| 230 | in->nb_samples); |
| 231 | if (ret <= 0) { |
| 232 | av_frame_free(&out); |
| 233 | if (ret < 0) |
| 234 | goto fail; |
| 235 | } |
| 236 | |
| 237 | av_assert0(!avresample_available(s->avr)); |
| 238 | |
| 239 | if (s->next_pts == AV_NOPTS_VALUE) { |
| 240 | if (in->pts == AV_NOPTS_VALUE) { |
| 241 | av_log(ctx, AV_LOG_WARNING, "First timestamp is missing, " |
| 242 | "assuming 0.\n"); |
| 243 | s->next_pts = 0; |
| 244 | } else |
| 245 | s->next_pts = av_rescale_q(in->pts, inlink->time_base, |
| 246 | outlink->time_base); |
| 247 | } |
| 248 | |
| 249 | if (ret > 0) { |
| 250 | out->nb_samples = ret; |
| 251 | |
| 252 | ret = av_frame_copy_props(out, in); |
| 253 | if (ret < 0) { |
| 254 | av_frame_free(&out); |
| 255 | goto fail; |
| 256 | } |
| 257 | |
| 258 | out->sample_rate = outlink->sample_rate; |
| 259 | /* Only convert in->pts if there is a discontinuous jump. |
| 260 | This ensures that out->pts tracks the number of samples actually |
| 261 | output by the resampler in the absence of such a jump. |
| 262 | Otherwise, the rounding in av_rescale_q() and av_rescale() |
| 263 | causes off-by-1 errors. */ |
| 264 | if (in->pts != AV_NOPTS_VALUE && in->pts != s->next_in_pts) { |
| 265 | out->pts = av_rescale_q(in->pts, inlink->time_base, |
| 266 | outlink->time_base) - |
| 267 | av_rescale(delay, outlink->sample_rate, |
| 268 | inlink->sample_rate); |
| 269 | } else |
| 270 | out->pts = s->next_pts; |
| 271 | |
| 272 | s->next_pts = out->pts + out->nb_samples; |
| 273 | s->next_in_pts = in->pts + in->nb_samples; |
| 274 | |
| 275 | ret = ff_filter_frame(outlink, out); |
| 276 | s->got_output = 1; |
| 277 | } |
| 278 | |
| 279 | fail: |
| 280 | av_frame_free(&in); |
| 281 | } else { |
| 282 | in->format = outlink->format; |
| 283 | ret = ff_filter_frame(outlink, in); |
| 284 | s->got_output = 1; |
| 285 | } |
| 286 | |
| 287 | return ret; |
| 288 | } |
| 289 | |
| 290 | static const AVClass *resample_child_class_next(const AVClass *prev) |
| 291 | { |
| 292 | return prev ? NULL : avresample_get_class(); |
| 293 | } |
| 294 | |
| 295 | static void *resample_child_next(void *obj, void *prev) |
| 296 | { |
| 297 | ResampleContext *s = obj; |
| 298 | return prev ? NULL : s->avr; |
| 299 | } |
| 300 | |
| 301 | static const AVClass resample_class = { |
| 302 | .class_name = "resample", |
| 303 | .item_name = av_default_item_name, |
| 304 | .version = LIBAVUTIL_VERSION_INT, |
| 305 | .child_class_next = resample_child_class_next, |
| 306 | .child_next = resample_child_next, |
| 307 | }; |
| 308 | |
| 309 | static const AVFilterPad avfilter_af_resample_inputs[] = { |
| 310 | { |
| 311 | .name = "default", |
| 312 | .type = AVMEDIA_TYPE_AUDIO, |
| 313 | .filter_frame = filter_frame, |
| 314 | }, |
| 315 | { NULL } |
| 316 | }; |
| 317 | |
| 318 | static const AVFilterPad avfilter_af_resample_outputs[] = { |
| 319 | { |
| 320 | .name = "default", |
| 321 | .type = AVMEDIA_TYPE_AUDIO, |
| 322 | .config_props = config_output, |
| 323 | .request_frame = request_frame |
| 324 | }, |
| 325 | { NULL } |
| 326 | }; |
| 327 | |
| 328 | AVFilter ff_af_resample = { |
| 329 | .name = "resample", |
| 330 | .description = NULL_IF_CONFIG_SMALL("Audio resampling and conversion."), |
| 331 | .priv_size = sizeof(ResampleContext), |
| 332 | .priv_class = &resample_class, |
| 333 | .init_dict = init, |
| 334 | .uninit = uninit, |
| 335 | .query_formats = query_formats, |
| 336 | .inputs = avfilter_af_resample_inputs, |
| 337 | .outputs = avfilter_af_resample_outputs, |
| 338 | }; |