FFmpeg
 All Data Structures Namespaces Files Functions Variables Typedefs Enumerations Enumerator Macros Groups Pages
af_resample.c
Go to the documentation of this file.
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 
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;
42 
43  int64_t next_pts;
44 
45  /* set by filter_frame() to signal an output frame to request_frame() */
48 
49 static av_cold int init(AVFilterContext *ctx, AVDictionary **opts)
50 {
51  ResampleContext *s = ctx->priv;
52  const AVClass *avr_class = avresample_get_class();
53  AVDictionaryEntry *e = NULL;
54 
55  while ((e = av_dict_get(*opts, "", e, AV_DICT_IGNORE_SUFFIX))) {
56  if (av_opt_find(&avr_class, e->key, NULL, 0,
58  av_dict_set(&s->options, e->key, e->value, 0);
59  }
60 
61  e = NULL;
62  while ((e = av_dict_get(s->options, "", e, AV_DICT_IGNORE_SUFFIX)))
63  av_dict_set(opts, e->key, NULL, 0);
64 
65  /* do not allow the user to override basic format options */
66  av_dict_set(&s->options, "in_channel_layout", NULL, 0);
67  av_dict_set(&s->options, "out_channel_layout", NULL, 0);
68  av_dict_set(&s->options, "in_sample_fmt", NULL, 0);
69  av_dict_set(&s->options, "out_sample_fmt", NULL, 0);
70  av_dict_set(&s->options, "in_sample_rate", NULL, 0);
71  av_dict_set(&s->options, "out_sample_rate", NULL, 0);
72 
73  return 0;
74 }
75 
76 static av_cold void uninit(AVFilterContext *ctx)
77 {
78  ResampleContext *s = ctx->priv;
79 
80  if (s->avr) {
82  avresample_free(&s->avr);
83  }
84  av_dict_free(&s->options);
85 }
86 
88 {
89  AVFilterLink *inlink = ctx->inputs[0];
90  AVFilterLink *outlink = ctx->outputs[0];
91 
94  AVFilterFormats *in_samplerates = ff_all_samplerates();
95  AVFilterFormats *out_samplerates = ff_all_samplerates();
98 
99  ff_formats_ref(in_formats, &inlink->out_formats);
100  ff_formats_ref(out_formats, &outlink->in_formats);
101 
102  ff_formats_ref(in_samplerates, &inlink->out_samplerates);
103  ff_formats_ref(out_samplerates, &outlink->in_samplerates);
104 
105  ff_channel_layouts_ref(in_layouts, &inlink->out_channel_layouts);
106  ff_channel_layouts_ref(out_layouts, &outlink->in_channel_layouts);
107 
108  return 0;
109 }
110 
111 static int config_output(AVFilterLink *outlink)
112 {
113  AVFilterContext *ctx = outlink->src;
114  AVFilterLink *inlink = ctx->inputs[0];
115  ResampleContext *s = ctx->priv;
116  char buf1[64], buf2[64];
117  int ret;
118 
119  if (s->avr) {
120  avresample_close(s->avr);
121  avresample_free(&s->avr);
122  }
123 
124  if (inlink->channel_layout == outlink->channel_layout &&
125  inlink->sample_rate == outlink->sample_rate &&
126  (inlink->format == outlink->format ||
129  av_get_planar_sample_fmt(inlink->format) ==
130  av_get_planar_sample_fmt(outlink->format))))
131  return 0;
132 
133  if (!(s->avr = avresample_alloc_context()))
134  return AVERROR(ENOMEM);
135 
136  if (s->options) {
137  AVDictionaryEntry *e = NULL;
138  while ((e = av_dict_get(s->options, "", e, AV_DICT_IGNORE_SUFFIX)))
139  av_log(ctx, AV_LOG_VERBOSE, "lavr option: %s=%s\n", e->key, e->value);
140 
141  av_opt_set_dict(s->avr, &s->options);
142  }
143 
144  av_opt_set_int(s->avr, "in_channel_layout", inlink ->channel_layout, 0);
145  av_opt_set_int(s->avr, "out_channel_layout", outlink->channel_layout, 0);
146  av_opt_set_int(s->avr, "in_sample_fmt", inlink ->format, 0);
147  av_opt_set_int(s->avr, "out_sample_fmt", outlink->format, 0);
148  av_opt_set_int(s->avr, "in_sample_rate", inlink ->sample_rate, 0);
149  av_opt_set_int(s->avr, "out_sample_rate", outlink->sample_rate, 0);
150 
151  if ((ret = avresample_open(s->avr)) < 0)
152  return ret;
153 
154  outlink->time_base = (AVRational){ 1, outlink->sample_rate };
156 
157  av_get_channel_layout_string(buf1, sizeof(buf1),
158  -1, inlink ->channel_layout);
159  av_get_channel_layout_string(buf2, sizeof(buf2),
160  -1, outlink->channel_layout);
161  av_log(ctx, AV_LOG_VERBOSE,
162  "fmt:%s srate:%d cl:%s -> fmt:%s srate:%d cl:%s\n",
163  av_get_sample_fmt_name(inlink ->format), inlink ->sample_rate, buf1,
164  av_get_sample_fmt_name(outlink->format), outlink->sample_rate, buf2);
165 
166  return 0;
167 }
168 
169 static int request_frame(AVFilterLink *outlink)
170 {
171  AVFilterContext *ctx = outlink->src;
172  ResampleContext *s = ctx->priv;
173  int ret = 0;
174 
175  s->got_output = 0;
176  while (ret >= 0 && !s->got_output)
177  ret = ff_request_frame(ctx->inputs[0]);
178 
179  /* flush the lavr delay buffer */
180  if (ret == AVERROR_EOF && s->avr) {
181  AVFrame *frame;
182  int nb_samples = av_rescale_rnd(avresample_get_delay(s->avr),
183  outlink->sample_rate,
184  ctx->inputs[0]->sample_rate,
185  AV_ROUND_UP);
186 
187  if (!nb_samples)
188  return ret;
189 
190  frame = ff_get_audio_buffer(outlink, nb_samples);
191  if (!frame)
192  return AVERROR(ENOMEM);
193 
194  ret = avresample_convert(s->avr, frame->extended_data,
195  frame->linesize[0], nb_samples,
196  NULL, 0, 0);
197  if (ret <= 0) {
198  av_frame_free(&frame);
199  return (ret == 0) ? AVERROR_EOF : ret;
200  }
201 
202  frame->pts = s->next_pts;
203  return ff_filter_frame(outlink, frame);
204  }
205  return ret;
206 }
207 
208 static int filter_frame(AVFilterLink *inlink, AVFrame *in)
209 {
210  AVFilterContext *ctx = inlink->dst;
211  ResampleContext *s = ctx->priv;
212  AVFilterLink *outlink = ctx->outputs[0];
213  int ret;
214 
215  if (s->avr) {
216  AVFrame *out;
217  int delay, nb_samples;
218 
219  /* maximum possible samples lavr can output */
220  delay = avresample_get_delay(s->avr);
221  nb_samples = av_rescale_rnd(in->nb_samples + delay,
222  outlink->sample_rate, inlink->sample_rate,
223  AV_ROUND_UP);
224 
225  out = ff_get_audio_buffer(outlink, nb_samples);
226  if (!out) {
227  ret = AVERROR(ENOMEM);
228  goto fail;
229  }
230 
231  ret = avresample_convert(s->avr, out->extended_data, out->linesize[0],
232  nb_samples, in->extended_data, in->linesize[0],
233  in->nb_samples);
234  if (ret <= 0) {
235  av_frame_free(&out);
236  if (ret < 0)
237  goto fail;
238  }
239 
241 
242  if (s->next_pts == AV_NOPTS_VALUE) {
243  if (in->pts == AV_NOPTS_VALUE) {
244  av_log(ctx, AV_LOG_WARNING, "First timestamp is missing, "
245  "assuming 0.\n");
246  s->next_pts = 0;
247  } else
248  s->next_pts = av_rescale_q(in->pts, inlink->time_base,
249  outlink->time_base);
250  }
251 
252  if (ret > 0) {
253  out->nb_samples = ret;
254  if (in->pts != AV_NOPTS_VALUE) {
255  out->pts = av_rescale_q(in->pts, inlink->time_base,
256  outlink->time_base) -
257  av_rescale(delay, outlink->sample_rate,
258  inlink->sample_rate);
259  } else
260  out->pts = s->next_pts;
261 
262  s->next_pts = out->pts + out->nb_samples;
263 
264  ret = ff_filter_frame(outlink, out);
265  s->got_output = 1;
266  }
267 
268 fail:
269  av_frame_free(&in);
270  } else {
271  in->format = outlink->format;
272  ret = ff_filter_frame(outlink, in);
273  s->got_output = 1;
274  }
275 
276  return ret;
277 }
278 
279 static const AVClass *resample_child_class_next(const AVClass *prev)
280 {
281  return prev ? NULL : avresample_get_class();
282 }
283 
284 static void *resample_child_next(void *obj, void *prev)
285 {
286  ResampleContext *s = obj;
287  return prev ? NULL : s->avr;
288 }
289 
290 static const AVClass resample_class = {
291  .class_name = "resample",
292  .item_name = av_default_item_name,
293  .version = LIBAVUTIL_VERSION_INT,
294  .child_class_next = resample_child_class_next,
296 };
297 
299  {
300  .name = "default",
301  .type = AVMEDIA_TYPE_AUDIO,
302  .filter_frame = filter_frame,
303  },
304  { NULL }
305 };
306 
308  {
309  .name = "default",
310  .type = AVMEDIA_TYPE_AUDIO,
311  .config_props = config_output,
312  .request_frame = request_frame
313  },
314  { NULL }
315 };
316 
318  .name = "resample",
319  .description = NULL_IF_CONFIG_SMALL("Audio resampling and conversion."),
320  .priv_size = sizeof(ResampleContext),
321  .priv_class = &resample_class,
322  .init_dict = init,
323  .uninit = uninit,
325  .inputs = avfilter_af_resample_inputs,
326  .outputs = avfilter_af_resample_outputs,
327 };