1 /*
2 * Copyright (c) 2011 Stefano Sabatini
3 * Copyright (c) 2011 Mina Nagy Zaki
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 * resampling audio filter
25 */
26
27 #include "libavutil/avstring.h"
28 #include "libavutil/channel_layout.h"
29 #include "libavutil/opt.h"
30 #include "libavutil/samplefmt.h"
31 #include "libavutil/avassert.h"
32 #include "libswresample/swresample.h"
33 #include "avfilter.h"
34 #include "audio.h"
35 #include "internal.h"
36
37 typedef struct AResampleContext {
38 const AVClass *class;
39 int sample_rate_arg;
40 double ratio;
41 struct SwrContext *swr;
42 int64_t next_pts;
43 int more_data;
44 } AResampleContext;
45
preinit(AVFilterContext * ctx)46 static av_cold int preinit(AVFilterContext *ctx)
47 {
48 AResampleContext *aresample = ctx->priv;
49
50 aresample->next_pts = AV_NOPTS_VALUE;
51 aresample->swr = swr_alloc();
52 if (!aresample->swr)
53 return AVERROR(ENOMEM);
54
55 return 0;
56 }
57
uninit(AVFilterContext * ctx)58 static av_cold void uninit(AVFilterContext *ctx)
59 {
60 AResampleContext *aresample = ctx->priv;
61 swr_free(&aresample->swr);
62 }
63
query_formats(AVFilterContext * ctx)64 static int query_formats(AVFilterContext *ctx)
65 {
66 AResampleContext *aresample = ctx->priv;
67 enum AVSampleFormat out_format;
68 AVChannelLayout out_layout = { 0 };
69 int64_t out_rate;
70
71 AVFilterLink *inlink = ctx->inputs[0];
72 AVFilterLink *outlink = ctx->outputs[0];
73
74 AVFilterFormats *in_formats, *out_formats;
75 AVFilterFormats *in_samplerates, *out_samplerates;
76 AVFilterChannelLayouts *in_layouts, *out_layouts;
77 int ret;
78
79 if (aresample->sample_rate_arg > 0)
80 av_opt_set_int(aresample->swr, "osr", aresample->sample_rate_arg, 0);
81 av_opt_get_sample_fmt(aresample->swr, "osf", 0, &out_format);
82 av_opt_get_int(aresample->swr, "osr", 0, &out_rate);
83
84 in_formats = ff_all_formats(AVMEDIA_TYPE_AUDIO);
85 if ((ret = ff_formats_ref(in_formats, &inlink->outcfg.formats)) < 0)
86 return ret;
87
88 in_samplerates = ff_all_samplerates();
89 if ((ret = ff_formats_ref(in_samplerates, &inlink->outcfg.samplerates)) < 0)
90 return ret;
91
92 in_layouts = ff_all_channel_counts();
93 if ((ret = ff_channel_layouts_ref(in_layouts, &inlink->outcfg.channel_layouts)) < 0)
94 return ret;
95
96 if(out_rate > 0) {
97 int ratelist[] = { out_rate, -1 };
98 out_samplerates = ff_make_format_list(ratelist);
99 } else {
100 out_samplerates = ff_all_samplerates();
101 }
102
103 if ((ret = ff_formats_ref(out_samplerates, &outlink->incfg.samplerates)) < 0)
104 return ret;
105
106 if(out_format != AV_SAMPLE_FMT_NONE) {
107 int formatlist[] = { out_format, -1 };
108 out_formats = ff_make_format_list(formatlist);
109 } else
110 out_formats = ff_all_formats(AVMEDIA_TYPE_AUDIO);
111 if ((ret = ff_formats_ref(out_formats, &outlink->incfg.formats)) < 0)
112 return ret;
113
114 av_opt_get_chlayout(aresample->swr, "ochl", 0, &out_layout);
115 if (av_channel_layout_check(&out_layout)) {
116 const AVChannelLayout layout_list[] = { out_layout, { 0 } };
117 out_layouts = ff_make_channel_layout_list(layout_list);
118 } else
119 out_layouts = ff_all_channel_counts();
120 av_channel_layout_uninit(&out_layout);
121
122 return ff_channel_layouts_ref(out_layouts, &outlink->incfg.channel_layouts);
123 }
124
125
config_output(AVFilterLink * outlink)126 static int config_output(AVFilterLink *outlink)
127 {
128 int ret;
129 AVFilterContext *ctx = outlink->src;
130 AVFilterLink *inlink = ctx->inputs[0];
131 AResampleContext *aresample = ctx->priv;
132 AVChannelLayout out_layout = { 0 };
133 int64_t out_rate;
134 enum AVSampleFormat out_format;
135 char inchl_buf[128], outchl_buf[128];
136
137 ret = swr_alloc_set_opts2(&aresample->swr,
138 &outlink->ch_layout, outlink->format, outlink->sample_rate,
139 &inlink->ch_layout, inlink->format, inlink->sample_rate,
140 0, ctx);
141 if (ret < 0)
142 return ret;
143
144 ret = swr_init(aresample->swr);
145 if (ret < 0)
146 return ret;
147
148 av_opt_get_int(aresample->swr, "osr", 0, &out_rate);
149 av_opt_get_chlayout(aresample->swr, "ochl", 0, &out_layout);
150 av_opt_get_sample_fmt(aresample->swr, "osf", 0, &out_format);
151 outlink->time_base = (AVRational) {1, out_rate};
152
153 av_assert0(outlink->sample_rate == out_rate);
154 av_assert0(!av_channel_layout_compare(&outlink->ch_layout, &out_layout));
155 av_assert0(outlink->format == out_format);
156
157 av_channel_layout_uninit(&out_layout);
158
159 aresample->ratio = (double)outlink->sample_rate / inlink->sample_rate;
160
161 av_channel_layout_describe(&inlink ->ch_layout, inchl_buf, sizeof(inchl_buf));
162 av_channel_layout_describe(&outlink->ch_layout, outchl_buf, sizeof(outchl_buf));
163
164 av_log(ctx, AV_LOG_VERBOSE, "ch:%d chl:%s fmt:%s r:%dHz -> ch:%d chl:%s fmt:%s r:%dHz\n",
165 inlink ->ch_layout.nb_channels, inchl_buf, av_get_sample_fmt_name(inlink->format), inlink->sample_rate,
166 outlink->ch_layout.nb_channels, outchl_buf, av_get_sample_fmt_name(outlink->format), outlink->sample_rate);
167 return 0;
168 }
169
filter_frame(AVFilterLink * inlink,AVFrame * insamplesref)170 static int filter_frame(AVFilterLink *inlink, AVFrame *insamplesref)
171 {
172 AResampleContext *aresample = inlink->dst->priv;
173 const int n_in = insamplesref->nb_samples;
174 int64_t delay;
175 int n_out = n_in * aresample->ratio + 32;
176 AVFilterLink *const outlink = inlink->dst->outputs[0];
177 AVFrame *outsamplesref;
178 int ret;
179
180 delay = swr_get_delay(aresample->swr, outlink->sample_rate);
181 if (delay > 0)
182 n_out += FFMIN(delay, FFMAX(4096, n_out));
183
184 outsamplesref = ff_get_audio_buffer(outlink, n_out);
185
186 if(!outsamplesref) {
187 av_frame_free(&insamplesref);
188 return AVERROR(ENOMEM);
189 }
190
191 av_frame_copy_props(outsamplesref, insamplesref);
192 outsamplesref->format = outlink->format;
193 #if FF_API_OLD_CHANNEL_LAYOUT
194 FF_DISABLE_DEPRECATION_WARNINGS
195 outsamplesref->channels = outlink->ch_layout.nb_channels;
196 outsamplesref->channel_layout = outlink->channel_layout;
197 FF_ENABLE_DEPRECATION_WARNINGS
198 #endif
199 ret = av_channel_layout_copy(&outsamplesref->ch_layout, &outlink->ch_layout);
200 if (ret < 0)
201 return ret;
202 outsamplesref->sample_rate = outlink->sample_rate;
203
204 if(insamplesref->pts != AV_NOPTS_VALUE) {
205 int64_t inpts = av_rescale(insamplesref->pts, inlink->time_base.num * (int64_t)outlink->sample_rate * inlink->sample_rate, inlink->time_base.den);
206 int64_t outpts= swr_next_pts(aresample->swr, inpts);
207 aresample->next_pts =
208 outsamplesref->pts = ROUNDED_DIV(outpts, inlink->sample_rate);
209 } else {
210 outsamplesref->pts = AV_NOPTS_VALUE;
211 }
212 n_out = swr_convert(aresample->swr, outsamplesref->extended_data, n_out,
213 (void *)insamplesref->extended_data, n_in);
214 if (n_out <= 0) {
215 av_frame_free(&outsamplesref);
216 av_frame_free(&insamplesref);
217 return 0;
218 }
219
220 aresample->more_data = outsamplesref->nb_samples == n_out; // Indicate that there is probably more data in our buffers
221
222 outsamplesref->nb_samples = n_out;
223
224 ret = ff_filter_frame(outlink, outsamplesref);
225 av_frame_free(&insamplesref);
226 return ret;
227 }
228
flush_frame(AVFilterLink * outlink,int final,AVFrame ** outsamplesref_ret)229 static int flush_frame(AVFilterLink *outlink, int final, AVFrame **outsamplesref_ret)
230 {
231 AVFilterContext *ctx = outlink->src;
232 AResampleContext *aresample = ctx->priv;
233 AVFilterLink *const inlink = outlink->src->inputs[0];
234 AVFrame *outsamplesref;
235 int n_out = 4096;
236 int64_t pts;
237
238 outsamplesref = ff_get_audio_buffer(outlink, n_out);
239 *outsamplesref_ret = outsamplesref;
240 if (!outsamplesref)
241 return AVERROR(ENOMEM);
242
243 pts = swr_next_pts(aresample->swr, INT64_MIN);
244 pts = ROUNDED_DIV(pts, inlink->sample_rate);
245
246 n_out = swr_convert(aresample->swr, outsamplesref->extended_data, n_out, final ? NULL : (void*)outsamplesref->extended_data, 0);
247 if (n_out <= 0) {
248 av_frame_free(&outsamplesref);
249 return (n_out == 0) ? AVERROR_EOF : n_out;
250 }
251
252 outsamplesref->sample_rate = outlink->sample_rate;
253 outsamplesref->nb_samples = n_out;
254
255 outsamplesref->pts = pts;
256
257 return 0;
258 }
259
request_frame(AVFilterLink * outlink)260 static int request_frame(AVFilterLink *outlink)
261 {
262 AVFilterContext *ctx = outlink->src;
263 AResampleContext *aresample = ctx->priv;
264 int ret;
265
266 // First try to get data from the internal buffers
267 if (aresample->more_data) {
268 AVFrame *outsamplesref;
269
270 if (flush_frame(outlink, 0, &outsamplesref) >= 0) {
271 return ff_filter_frame(outlink, outsamplesref);
272 }
273 }
274 aresample->more_data = 0;
275
276 // Second request more data from the input
277 ret = ff_request_frame(ctx->inputs[0]);
278
279 // Third if we hit the end flush
280 if (ret == AVERROR_EOF) {
281 AVFrame *outsamplesref;
282
283 if ((ret = flush_frame(outlink, 1, &outsamplesref)) < 0)
284 return ret;
285
286 return ff_filter_frame(outlink, outsamplesref);
287 }
288 return ret;
289 }
290
resample_child_class_iterate(void ** iter)291 static const AVClass *resample_child_class_iterate(void **iter)
292 {
293 const AVClass *c = *iter ? NULL : swr_get_class();
294 *iter = (void*)(uintptr_t)c;
295 return c;
296 }
297
resample_child_next(void * obj,void * prev)298 static void *resample_child_next(void *obj, void *prev)
299 {
300 AResampleContext *s = obj;
301 return prev ? NULL : s->swr;
302 }
303
304 #define OFFSET(x) offsetof(AResampleContext, x)
305 #define FLAGS AV_OPT_FLAG_AUDIO_PARAM|AV_OPT_FLAG_FILTERING_PARAM
306
307 static const AVOption options[] = {
308 {"sample_rate", NULL, OFFSET(sample_rate_arg), AV_OPT_TYPE_INT, {.i64=0}, 0, INT_MAX, FLAGS },
309 {NULL}
310 };
311
312 static const AVClass aresample_class = {
313 .class_name = "aresample",
314 .item_name = av_default_item_name,
315 .option = options,
316 .version = LIBAVUTIL_VERSION_INT,
317 .child_class_iterate = resample_child_class_iterate,
318 .child_next = resample_child_next,
319 };
320
321 static const AVFilterPad aresample_inputs[] = {
322 {
323 .name = "default",
324 .type = AVMEDIA_TYPE_AUDIO,
325 .filter_frame = filter_frame,
326 },
327 };
328
329 static const AVFilterPad aresample_outputs[] = {
330 {
331 .name = "default",
332 .config_props = config_output,
333 .request_frame = request_frame,
334 .type = AVMEDIA_TYPE_AUDIO,
335 },
336 };
337
338 const AVFilter ff_af_aresample = {
339 .name = "aresample",
340 .description = NULL_IF_CONFIG_SMALL("Resample audio data."),
341 .preinit = preinit,
342 .uninit = uninit,
343 .priv_size = sizeof(AResampleContext),
344 .priv_class = &aresample_class,
345 FILTER_INPUTS(aresample_inputs),
346 FILTER_OUTPUTS(aresample_outputs),
347 FILTER_QUERY_FUNC(query_formats),
348 };
349