blob: 028e105318488481203bce6ff2a0fbdf8959d8cb
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 { |
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 | |
46 | static av_cold int init_dict(AVFilterContext *ctx, AVDictionary **opts) |
47 | { |
48 | AResampleContext *aresample = ctx->priv; |
49 | int ret = 0; |
50 | |
51 | aresample->next_pts = AV_NOPTS_VALUE; |
52 | aresample->swr = swr_alloc(); |
53 | if (!aresample->swr) { |
54 | ret = AVERROR(ENOMEM); |
55 | goto end; |
56 | } |
57 | |
58 | if (opts) { |
59 | AVDictionaryEntry *e = NULL; |
60 | |
61 | while ((e = av_dict_get(*opts, "", e, AV_DICT_IGNORE_SUFFIX))) { |
62 | if ((ret = av_opt_set(aresample->swr, e->key, e->value, 0)) < 0) |
63 | goto end; |
64 | } |
65 | av_dict_free(opts); |
66 | } |
67 | if (aresample->sample_rate_arg > 0) |
68 | av_opt_set_int(aresample->swr, "osr", aresample->sample_rate_arg, 0); |
69 | end: |
70 | return ret; |
71 | } |
72 | |
73 | static av_cold void uninit(AVFilterContext *ctx) |
74 | { |
75 | AResampleContext *aresample = ctx->priv; |
76 | swr_free(&aresample->swr); |
77 | } |
78 | |
79 | static int query_formats(AVFilterContext *ctx) |
80 | { |
81 | AResampleContext *aresample = ctx->priv; |
82 | enum AVSampleFormat out_format; |
83 | int64_t out_rate, out_layout; |
84 | |
85 | AVFilterLink *inlink = ctx->inputs[0]; |
86 | AVFilterLink *outlink = ctx->outputs[0]; |
87 | |
88 | AVFilterFormats *in_formats, *out_formats; |
89 | AVFilterFormats *in_samplerates, *out_samplerates; |
90 | AVFilterChannelLayouts *in_layouts, *out_layouts; |
91 | int ret; |
92 | |
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); |
96 | |
97 | in_formats = ff_all_formats(AVMEDIA_TYPE_AUDIO); |
98 | if ((ret = ff_formats_ref(in_formats, &inlink->out_formats)) < 0) |
99 | return ret; |
100 | |
101 | in_samplerates = ff_all_samplerates(); |
102 | if ((ret = ff_formats_ref(in_samplerates, &inlink->out_samplerates)) < 0) |
103 | return ret; |
104 | |
105 | in_layouts = ff_all_channel_counts(); |
106 | if ((ret = ff_channel_layouts_ref(in_layouts, &inlink->out_channel_layouts)) < 0) |
107 | return ret; |
108 | |
109 | if(out_rate > 0) { |
110 | int ratelist[] = { out_rate, -1 }; |
111 | out_samplerates = ff_make_format_list(ratelist); |
112 | } else { |
113 | out_samplerates = ff_all_samplerates(); |
114 | } |
115 | |
116 | if ((ret = ff_formats_ref(out_samplerates, &outlink->in_samplerates)) < 0) |
117 | return ret; |
118 | |
119 | if(out_format != AV_SAMPLE_FMT_NONE) { |
120 | int formatlist[] = { out_format, -1 }; |
121 | out_formats = ff_make_format_list(formatlist); |
122 | } else |
123 | out_formats = ff_all_formats(AVMEDIA_TYPE_AUDIO); |
124 | if ((ret = ff_formats_ref(out_formats, &outlink->in_formats)) < 0) |
125 | return ret; |
126 | |
127 | if(out_layout) { |
128 | int64_t layout_list[] = { out_layout, -1 }; |
129 | out_layouts = avfilter_make_format64_list(layout_list); |
130 | } else |
131 | out_layouts = ff_all_channel_counts(); |
132 | |
133 | return ff_channel_layouts_ref(out_layouts, &outlink->in_channel_layouts); |
134 | } |
135 | |
136 | |
137 | static int config_output(AVFilterLink *outlink) |
138 | { |
139 | int ret; |
140 | AVFilterContext *ctx = outlink->src; |
141 | AVFilterLink *inlink = ctx->inputs[0]; |
142 | AResampleContext *aresample = ctx->priv; |
143 | int64_t out_rate, out_layout; |
144 | enum AVSampleFormat out_format; |
145 | char inchl_buf[128], outchl_buf[128]; |
146 | |
147 | aresample->swr = swr_alloc_set_opts(aresample->swr, |
148 | outlink->channel_layout, outlink->format, outlink->sample_rate, |
149 | inlink->channel_layout, inlink->format, inlink->sample_rate, |
150 | 0, ctx); |
151 | if (!aresample->swr) |
152 | return AVERROR(ENOMEM); |
153 | if (!inlink->channel_layout) |
154 | av_opt_set_int(aresample->swr, "ich", inlink->channels, 0); |
155 | if (!outlink->channel_layout) |
156 | av_opt_set_int(aresample->swr, "och", outlink->channels, 0); |
157 | |
158 | ret = swr_init(aresample->swr); |
159 | if (ret < 0) |
160 | return ret; |
161 | |
162 | av_opt_get_int(aresample->swr, "osr", 0, &out_rate); |
163 | av_opt_get_int(aresample->swr, "ocl", 0, &out_layout); |
164 | av_opt_get_sample_fmt(aresample->swr, "osf", 0, &out_format); |
165 | outlink->time_base = (AVRational) {1, out_rate}; |
166 | |
167 | av_assert0(outlink->sample_rate == out_rate); |
168 | av_assert0(outlink->channel_layout == out_layout || !outlink->channel_layout); |
169 | av_assert0(outlink->format == out_format); |
170 | |
171 | aresample->ratio = (double)outlink->sample_rate / inlink->sample_rate; |
172 | |
173 | av_get_channel_layout_string(inchl_buf, sizeof(inchl_buf), inlink ->channels, inlink ->channel_layout); |
174 | av_get_channel_layout_string(outchl_buf, sizeof(outchl_buf), outlink->channels, outlink->channel_layout); |
175 | |
176 | av_log(ctx, AV_LOG_VERBOSE, "ch:%d chl:%s fmt:%s r:%dHz -> ch:%d chl:%s fmt:%s r:%dHz\n", |
177 | inlink ->channels, inchl_buf, av_get_sample_fmt_name(inlink->format), inlink->sample_rate, |
178 | outlink->channels, outchl_buf, av_get_sample_fmt_name(outlink->format), outlink->sample_rate); |
179 | return 0; |
180 | } |
181 | |
182 | static int filter_frame(AVFilterLink *inlink, AVFrame *insamplesref) |
183 | { |
184 | AResampleContext *aresample = inlink->dst->priv; |
185 | const int n_in = insamplesref->nb_samples; |
186 | int64_t delay; |
187 | int n_out = n_in * aresample->ratio + 32; |
188 | AVFilterLink *const outlink = inlink->dst->outputs[0]; |
189 | AVFrame *outsamplesref; |
190 | int ret; |
191 | |
192 | delay = swr_get_delay(aresample->swr, outlink->sample_rate); |
193 | if (delay > 0) |
194 | n_out += FFMIN(delay, FFMAX(4096, n_out)); |
195 | |
196 | outsamplesref = ff_get_audio_buffer(outlink, n_out); |
197 | |
198 | if(!outsamplesref) |
199 | return AVERROR(ENOMEM); |
200 | |
201 | av_frame_copy_props(outsamplesref, insamplesref); |
202 | outsamplesref->format = outlink->format; |
203 | av_frame_set_channels(outsamplesref, outlink->channels); |
204 | outsamplesref->channel_layout = outlink->channel_layout; |
205 | outsamplesref->sample_rate = outlink->sample_rate; |
206 | |
207 | if(insamplesref->pts != AV_NOPTS_VALUE) { |
208 | int64_t inpts = av_rescale(insamplesref->pts, inlink->time_base.num * (int64_t)outlink->sample_rate * inlink->sample_rate, inlink->time_base.den); |
209 | int64_t outpts= swr_next_pts(aresample->swr, inpts); |
210 | aresample->next_pts = |
211 | outsamplesref->pts = ROUNDED_DIV(outpts, inlink->sample_rate); |
212 | } else { |
213 | outsamplesref->pts = AV_NOPTS_VALUE; |
214 | } |
215 | n_out = swr_convert(aresample->swr, outsamplesref->extended_data, n_out, |
216 | (void *)insamplesref->extended_data, n_in); |
217 | if (n_out <= 0) { |
218 | av_frame_free(&outsamplesref); |
219 | av_frame_free(&insamplesref); |
220 | return 0; |
221 | } |
222 | |
223 | aresample->more_data = outsamplesref->nb_samples == n_out; // Indicate that there is probably more data in our buffers |
224 | |
225 | outsamplesref->nb_samples = n_out; |
226 | |
227 | ret = ff_filter_frame(outlink, outsamplesref); |
228 | av_frame_free(&insamplesref); |
229 | return ret; |
230 | } |
231 | |
232 | static int flush_frame(AVFilterLink *outlink, int final, AVFrame **outsamplesref_ret) |
233 | { |
234 | AVFilterContext *ctx = outlink->src; |
235 | AResampleContext *aresample = ctx->priv; |
236 | AVFilterLink *const inlink = outlink->src->inputs[0]; |
237 | AVFrame *outsamplesref; |
238 | int n_out = 4096; |
239 | int64_t pts; |
240 | |
241 | outsamplesref = ff_get_audio_buffer(outlink, n_out); |
242 | *outsamplesref_ret = outsamplesref; |
243 | if (!outsamplesref) |
244 | return AVERROR(ENOMEM); |
245 | |
246 | pts = swr_next_pts(aresample->swr, INT64_MIN); |
247 | pts = ROUNDED_DIV(pts, inlink->sample_rate); |
248 | |
249 | n_out = swr_convert(aresample->swr, outsamplesref->extended_data, n_out, final ? NULL : (void*)outsamplesref->extended_data, 0); |
250 | if (n_out <= 0) { |
251 | av_frame_free(&outsamplesref); |
252 | return (n_out == 0) ? AVERROR_EOF : n_out; |
253 | } |
254 | |
255 | outsamplesref->sample_rate = outlink->sample_rate; |
256 | outsamplesref->nb_samples = n_out; |
257 | |
258 | outsamplesref->pts = pts; |
259 | |
260 | return 0; |
261 | } |
262 | |
263 | static int request_frame(AVFilterLink *outlink) |
264 | { |
265 | AVFilterContext *ctx = outlink->src; |
266 | AResampleContext *aresample = ctx->priv; |
267 | int ret; |
268 | |
269 | // First try to get data from the internal buffers |
270 | if (aresample->more_data) { |
271 | AVFrame *outsamplesref; |
272 | |
273 | if (flush_frame(outlink, 0, &outsamplesref) >= 0) { |
274 | return ff_filter_frame(outlink, outsamplesref); |
275 | } |
276 | } |
277 | aresample->more_data = 0; |
278 | |
279 | // Second request more data from the input |
280 | ret = ff_request_frame(ctx->inputs[0]); |
281 | |
282 | // Third if we hit the end flush |
283 | if (ret == AVERROR_EOF) { |
284 | AVFrame *outsamplesref; |
285 | |
286 | if ((ret = flush_frame(outlink, 1, &outsamplesref)) < 0) |
287 | return ret; |
288 | |
289 | return ff_filter_frame(outlink, outsamplesref); |
290 | } |
291 | return ret; |
292 | } |
293 | |
294 | static const AVClass *resample_child_class_next(const AVClass *prev) |
295 | { |
296 | return prev ? NULL : swr_get_class(); |
297 | } |
298 | |
299 | static void *resample_child_next(void *obj, void *prev) |
300 | { |
301 | AResampleContext *s = obj; |
302 | return prev ? NULL : s->swr; |
303 | } |
304 | |
305 | #define OFFSET(x) offsetof(AResampleContext, x) |
306 | #define FLAGS AV_OPT_FLAG_AUDIO_PARAM|AV_OPT_FLAG_FILTERING_PARAM |
307 | |
308 | static const AVOption options[] = { |
309 | {"sample_rate", NULL, OFFSET(sample_rate_arg), AV_OPT_TYPE_INT, {.i64=0}, 0, INT_MAX, FLAGS }, |
310 | {NULL} |
311 | }; |
312 | |
313 | static const AVClass aresample_class = { |
314 | .class_name = "aresample", |
315 | .item_name = av_default_item_name, |
316 | .option = options, |
317 | .version = LIBAVUTIL_VERSION_INT, |
318 | .child_class_next = resample_child_class_next, |
319 | .child_next = resample_child_next, |
320 | }; |
321 | |
322 | static const AVFilterPad aresample_inputs[] = { |
323 | { |
324 | .name = "default", |
325 | .type = AVMEDIA_TYPE_AUDIO, |
326 | .filter_frame = filter_frame, |
327 | }, |
328 | { NULL } |
329 | }; |
330 | |
331 | static const AVFilterPad aresample_outputs[] = { |
332 | { |
333 | .name = "default", |
334 | .config_props = config_output, |
335 | .request_frame = request_frame, |
336 | .type = AVMEDIA_TYPE_AUDIO, |
337 | }, |
338 | { NULL } |
339 | }; |
340 | |
341 | AVFilter ff_af_aresample = { |
342 | .name = "aresample", |
343 | .description = NULL_IF_CONFIG_SMALL("Resample audio data."), |
344 | .init_dict = init_dict, |
345 | .uninit = uninit, |
346 | .query_formats = query_formats, |
347 | .priv_size = sizeof(AResampleContext), |
348 | .priv_class = &aresample_class, |
349 | .inputs = aresample_inputs, |
350 | .outputs = aresample_outputs, |
351 | }; |
352 |