2 * Copyright (c) 2011 Stefano Sabatini
3 * Copyright (c) 2011 Mina Nagy Zaki
5 * This file is part of FFmpeg.
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.
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.
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
24 * resampling audio filter
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"
41 struct SwrContext *swr;
47 static av_cold int init_dict(AVFilterContext *ctx, AVDictionary **opts)
49 AResampleContext *aresample = ctx->priv;
52 aresample->next_pts = AV_NOPTS_VALUE;
53 aresample->swr = swr_alloc();
54 if (!aresample->swr) {
55 ret = AVERROR(ENOMEM);
60 AVDictionaryEntry *e = NULL;
62 while ((e = av_dict_get(*opts, "", e, AV_DICT_IGNORE_SUFFIX))) {
63 if ((ret = av_opt_set(aresample->swr, e->key, e->value, 0)) < 0)
68 if (aresample->sample_rate_arg > 0)
69 av_opt_set_int(aresample->swr, "osr", aresample->sample_rate_arg, 0);
74 static av_cold void uninit(AVFilterContext *ctx)
76 AResampleContext *aresample = ctx->priv;
77 swr_free(&aresample->swr);
80 static int query_formats(AVFilterContext *ctx)
82 AResampleContext *aresample = ctx->priv;
83 enum AVSampleFormat out_format;
84 int64_t out_rate, out_layout;
86 AVFilterLink *inlink = ctx->inputs[0];
87 AVFilterLink *outlink = ctx->outputs[0];
89 AVFilterFormats *in_formats, *out_formats;
90 AVFilterFormats *in_samplerates, *out_samplerates;
91 AVFilterChannelLayouts *in_layouts, *out_layouts;
93 av_opt_get_sample_fmt(aresample->swr, "osf", 0, &out_format);
94 av_opt_get_int(aresample->swr, "osr", 0, &out_rate);
95 av_opt_get_int(aresample->swr, "ocl", 0, &out_layout);
97 in_formats = ff_all_formats(AVMEDIA_TYPE_AUDIO);
99 return AVERROR(ENOMEM);
100 ff_formats_ref (in_formats, &inlink->out_formats);
102 in_samplerates = ff_all_samplerates();
104 return AVERROR(ENOMEM);
105 ff_formats_ref (in_samplerates, &inlink->out_samplerates);
107 in_layouts = ff_all_channel_counts();
109 return AVERROR(ENOMEM);
110 ff_channel_layouts_ref(in_layouts, &inlink->out_channel_layouts);
113 int ratelist[] = { out_rate, -1 };
114 out_samplerates = ff_make_format_list(ratelist);
116 out_samplerates = ff_all_samplerates();
118 if (!out_samplerates) {
119 av_log(ctx, AV_LOG_ERROR, "Cannot allocate output samplerates.\n");
120 return AVERROR(ENOMEM);
123 ff_formats_ref(out_samplerates, &outlink->in_samplerates);
125 if(out_format != AV_SAMPLE_FMT_NONE) {
126 int formatlist[] = { out_format, -1 };
127 out_formats = ff_make_format_list(formatlist);
129 out_formats = ff_all_formats(AVMEDIA_TYPE_AUDIO);
130 ff_formats_ref(out_formats, &outlink->in_formats);
133 int64_t layout_list[] = { out_layout, -1 };
134 out_layouts = avfilter_make_format64_list(layout_list);
136 out_layouts = ff_all_channel_counts();
137 ff_channel_layouts_ref(out_layouts, &outlink->in_channel_layouts);
143 static int config_output(AVFilterLink *outlink)
146 AVFilterContext *ctx = outlink->src;
147 AVFilterLink *inlink = ctx->inputs[0];
148 AResampleContext *aresample = ctx->priv;
149 int64_t out_rate, out_layout;
150 enum AVSampleFormat out_format;
151 char inchl_buf[128], outchl_buf[128];
153 aresample->swr = swr_alloc_set_opts(aresample->swr,
154 outlink->channel_layout, outlink->format, outlink->sample_rate,
155 inlink->channel_layout, inlink->format, inlink->sample_rate,
158 return AVERROR(ENOMEM);
159 if (!inlink->channel_layout)
160 av_opt_set_int(aresample->swr, "ich", inlink->channels, 0);
161 if (!outlink->channel_layout)
162 av_opt_set_int(aresample->swr, "och", outlink->channels, 0);
164 ret = swr_init(aresample->swr);
168 av_opt_get_int(aresample->swr, "osr", 0, &out_rate);
169 av_opt_get_int(aresample->swr, "ocl", 0, &out_layout);
170 av_opt_get_sample_fmt(aresample->swr, "osf", 0, &out_format);
171 outlink->time_base = (AVRational) {1, out_rate};
173 av_assert0(outlink->sample_rate == out_rate);
174 av_assert0(outlink->channel_layout == out_layout || !outlink->channel_layout);
175 av_assert0(outlink->format == out_format);
177 aresample->ratio = (double)outlink->sample_rate / inlink->sample_rate;
179 av_get_channel_layout_string(inchl_buf, sizeof(inchl_buf), inlink ->channels, inlink ->channel_layout);
180 av_get_channel_layout_string(outchl_buf, sizeof(outchl_buf), outlink->channels, outlink->channel_layout);
182 av_log(ctx, AV_LOG_VERBOSE, "ch:%d chl:%s fmt:%s r:%dHz -> ch:%d chl:%s fmt:%s r:%dHz\n",
183 inlink ->channels, inchl_buf, av_get_sample_fmt_name(inlink->format), inlink->sample_rate,
184 outlink->channels, outchl_buf, av_get_sample_fmt_name(outlink->format), outlink->sample_rate);
188 static int filter_frame(AVFilterLink *inlink, AVFrame *insamplesref)
190 AResampleContext *aresample = inlink->dst->priv;
191 const int n_in = insamplesref->nb_samples;
193 int n_out = n_in * aresample->ratio + 32;
194 AVFilterLink *const outlink = inlink->dst->outputs[0];
195 AVFrame *outsamplesref;
198 delay = swr_get_delay(aresample->swr, outlink->sample_rate);
200 n_out += FFMIN(delay, FFMAX(4096, n_out));
202 outsamplesref = ff_get_audio_buffer(outlink, n_out);
205 return AVERROR(ENOMEM);
207 av_frame_copy_props(outsamplesref, insamplesref);
208 outsamplesref->format = outlink->format;
209 av_frame_set_channels(outsamplesref, outlink->channels);
210 outsamplesref->channel_layout = outlink->channel_layout;
211 outsamplesref->sample_rate = outlink->sample_rate;
213 if(insamplesref->pts != AV_NOPTS_VALUE) {
214 int64_t inpts = av_rescale(insamplesref->pts, inlink->time_base.num * (int64_t)outlink->sample_rate * inlink->sample_rate, inlink->time_base.den);
215 int64_t outpts= swr_next_pts(aresample->swr, inpts);
216 aresample->next_pts =
217 outsamplesref->pts = ROUNDED_DIV(outpts, inlink->sample_rate);
219 outsamplesref->pts = AV_NOPTS_VALUE;
221 n_out = swr_convert(aresample->swr, outsamplesref->extended_data, n_out,
222 (void *)insamplesref->extended_data, n_in);
224 av_frame_free(&outsamplesref);
225 av_frame_free(&insamplesref);
229 aresample->more_data = outsamplesref->nb_samples == n_out; // Indicate that there is probably more data in our buffers
231 outsamplesref->nb_samples = n_out;
233 ret = ff_filter_frame(outlink, outsamplesref);
234 aresample->req_fullfilled= 1;
235 av_frame_free(&insamplesref);
239 static int flush_frame(AVFilterLink *outlink, int final, AVFrame **outsamplesref_ret)
241 AVFilterContext *ctx = outlink->src;
242 AResampleContext *aresample = ctx->priv;
243 AVFilterLink *const inlink = outlink->src->inputs[0];
244 AVFrame *outsamplesref;
248 outsamplesref = ff_get_audio_buffer(outlink, n_out);
249 *outsamplesref_ret = outsamplesref;
251 return AVERROR(ENOMEM);
253 pts = swr_next_pts(aresample->swr, INT64_MIN);
254 pts = ROUNDED_DIV(pts, inlink->sample_rate);
256 n_out = swr_convert(aresample->swr, outsamplesref->extended_data, n_out, final ? NULL : (void*)outsamplesref->extended_data, 0);
258 av_frame_free(&outsamplesref);
259 return (n_out == 0) ? AVERROR_EOF : n_out;
262 outsamplesref->sample_rate = outlink->sample_rate;
263 outsamplesref->nb_samples = n_out;
265 outsamplesref->pts = pts;
270 static int request_frame(AVFilterLink *outlink)
272 AVFilterContext *ctx = outlink->src;
273 AResampleContext *aresample = ctx->priv;
276 // First try to get data from the internal buffers
277 if (aresample->more_data) {
278 AVFrame *outsamplesref;
280 if (flush_frame(outlink, 0, &outsamplesref) >= 0) {
281 return ff_filter_frame(outlink, outsamplesref);
284 aresample->more_data = 0;
286 // Second request more data from the input
287 aresample->req_fullfilled = 0;
289 ret = ff_request_frame(ctx->inputs[0]);
290 }while(!aresample->req_fullfilled && ret>=0);
292 // Third if we hit the end flush
293 if (ret == AVERROR_EOF) {
294 AVFrame *outsamplesref;
296 if ((ret = flush_frame(outlink, 1, &outsamplesref)) < 0)
299 return ff_filter_frame(outlink, outsamplesref);
304 static const AVClass *resample_child_class_next(const AVClass *prev)
306 return prev ? NULL : swr_get_class();
309 static void *resample_child_next(void *obj, void *prev)
311 AResampleContext *s = obj;
312 return prev ? NULL : s->swr;
315 #define OFFSET(x) offsetof(AResampleContext, x)
316 #define FLAGS AV_OPT_FLAG_AUDIO_PARAM|AV_OPT_FLAG_FILTERING_PARAM
318 static const AVOption options[] = {
319 {"sample_rate", NULL, OFFSET(sample_rate_arg), AV_OPT_TYPE_INT, {.i64=0}, 0, INT_MAX, FLAGS },
323 static const AVClass aresample_class = {
324 .class_name = "aresample",
325 .item_name = av_default_item_name,
327 .version = LIBAVUTIL_VERSION_INT,
328 .child_class_next = resample_child_class_next,
329 .child_next = resample_child_next,
332 static const AVFilterPad aresample_inputs[] = {
335 .type = AVMEDIA_TYPE_AUDIO,
336 .filter_frame = filter_frame,
341 static const AVFilterPad aresample_outputs[] = {
344 .config_props = config_output,
345 .request_frame = request_frame,
346 .type = AVMEDIA_TYPE_AUDIO,
351 AVFilter ff_af_aresample = {
353 .description = NULL_IF_CONFIG_SMALL("Resample audio data."),
354 .init_dict = init_dict,
356 .query_formats = query_formats,
357 .priv_size = sizeof(AResampleContext),
358 .priv_class = &aresample_class,
359 .inputs = aresample_inputs,
360 .outputs = aresample_outputs,