blob: 187cacf28afd5ebd0dd7ff8d763aff8714bf045d
1 | /* |
2 | * Copyright (c) 2013 Paul B Mahol |
3 | * |
4 | * This file is part of FFmpeg. |
5 | * |
6 | * FFmpeg is free software; you can redistribute it and/or |
7 | * modify it under the terms of the GNU Lesser General Public |
8 | * License as published by the Free Software Foundation; either |
9 | * version 2.1 of the License, or (at your option) any later version. |
10 | * |
11 | * FFmpeg is distributed in the hope that it will be useful, |
12 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
13 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
14 | * Lesser General Public License for more details. |
15 | * |
16 | * You should have received a copy of the GNU Lesser General Public |
17 | * License along with FFmpeg; if not, write to the Free Software |
18 | * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
19 | */ |
20 | |
21 | #include "libavutil/avstring.h" |
22 | #include "libavutil/opt.h" |
23 | #include "libavutil/samplefmt.h" |
24 | #include "avfilter.h" |
25 | #include "audio.h" |
26 | #include "internal.h" |
27 | |
28 | typedef struct ChanDelay { |
29 | int delay; |
30 | unsigned delay_index; |
31 | unsigned index; |
32 | uint8_t *samples; |
33 | } ChanDelay; |
34 | |
35 | typedef struct AudioDelayContext { |
36 | const AVClass *class; |
37 | char *delays; |
38 | ChanDelay *chandelay; |
39 | int nb_delays; |
40 | int block_align; |
41 | unsigned max_delay; |
42 | int64_t next_pts; |
43 | |
44 | void (*delay_channel)(ChanDelay *d, int nb_samples, |
45 | const uint8_t *src, uint8_t *dst); |
46 | } AudioDelayContext; |
47 | |
48 | #define OFFSET(x) offsetof(AudioDelayContext, x) |
49 | #define A AV_OPT_FLAG_AUDIO_PARAM|AV_OPT_FLAG_FILTERING_PARAM |
50 | |
51 | static const AVOption adelay_options[] = { |
52 | { "delays", "set list of delays for each channel", OFFSET(delays), AV_OPT_TYPE_STRING, {.str=NULL}, 0, 0, A }, |
53 | { NULL } |
54 | }; |
55 | |
56 | AVFILTER_DEFINE_CLASS(adelay); |
57 | |
58 | static int query_formats(AVFilterContext *ctx) |
59 | { |
60 | AVFilterChannelLayouts *layouts; |
61 | AVFilterFormats *formats; |
62 | static const enum AVSampleFormat sample_fmts[] = { |
63 | AV_SAMPLE_FMT_U8P, AV_SAMPLE_FMT_S16P, AV_SAMPLE_FMT_S32P, |
64 | AV_SAMPLE_FMT_FLTP, AV_SAMPLE_FMT_DBLP, |
65 | AV_SAMPLE_FMT_NONE |
66 | }; |
67 | int ret; |
68 | |
69 | layouts = ff_all_channel_counts(); |
70 | if (!layouts) |
71 | return AVERROR(ENOMEM); |
72 | ret = ff_set_common_channel_layouts(ctx, layouts); |
73 | if (ret < 0) |
74 | return ret; |
75 | |
76 | formats = ff_make_format_list(sample_fmts); |
77 | if (!formats) |
78 | return AVERROR(ENOMEM); |
79 | ret = ff_set_common_formats(ctx, formats); |
80 | if (ret < 0) |
81 | return ret; |
82 | |
83 | formats = ff_all_samplerates(); |
84 | if (!formats) |
85 | return AVERROR(ENOMEM); |
86 | return ff_set_common_samplerates(ctx, formats); |
87 | } |
88 | |
89 | #define DELAY(name, type, fill) \ |
90 | static void delay_channel_## name ##p(ChanDelay *d, int nb_samples, \ |
91 | const uint8_t *ssrc, uint8_t *ddst) \ |
92 | { \ |
93 | const type *src = (type *)ssrc; \ |
94 | type *dst = (type *)ddst; \ |
95 | type *samples = (type *)d->samples; \ |
96 | \ |
97 | while (nb_samples) { \ |
98 | if (d->delay_index < d->delay) { \ |
99 | const int len = FFMIN(nb_samples, d->delay - d->delay_index); \ |
100 | \ |
101 | memcpy(&samples[d->delay_index], src, len * sizeof(type)); \ |
102 | memset(dst, fill, len * sizeof(type)); \ |
103 | d->delay_index += len; \ |
104 | src += len; \ |
105 | dst += len; \ |
106 | nb_samples -= len; \ |
107 | } else { \ |
108 | *dst = samples[d->index]; \ |
109 | samples[d->index] = *src; \ |
110 | nb_samples--; \ |
111 | d->index++; \ |
112 | src++, dst++; \ |
113 | d->index = d->index >= d->delay ? 0 : d->index; \ |
114 | } \ |
115 | } \ |
116 | } |
117 | |
118 | DELAY(u8, uint8_t, 0x80) |
119 | DELAY(s16, int16_t, 0) |
120 | DELAY(s32, int32_t, 0) |
121 | DELAY(flt, float, 0) |
122 | DELAY(dbl, double, 0) |
123 | |
124 | static int config_input(AVFilterLink *inlink) |
125 | { |
126 | AVFilterContext *ctx = inlink->dst; |
127 | AudioDelayContext *s = ctx->priv; |
128 | char *p, *arg, *saveptr = NULL; |
129 | int i; |
130 | |
131 | s->chandelay = av_calloc(inlink->channels, sizeof(*s->chandelay)); |
132 | if (!s->chandelay) |
133 | return AVERROR(ENOMEM); |
134 | s->nb_delays = inlink->channels; |
135 | s->block_align = av_get_bytes_per_sample(inlink->format); |
136 | |
137 | p = s->delays; |
138 | for (i = 0; i < s->nb_delays; i++) { |
139 | ChanDelay *d = &s->chandelay[i]; |
140 | float delay; |
141 | char type = 0; |
142 | int ret; |
143 | |
144 | if (!(arg = av_strtok(p, "|", &saveptr))) |
145 | break; |
146 | |
147 | p = NULL; |
148 | |
149 | ret = sscanf(arg, "%d%c", &d->delay, &type); |
150 | if (ret != 2 || type != 'S') { |
151 | sscanf(arg, "%f", &delay); |
152 | d->delay = delay * inlink->sample_rate / 1000.0; |
153 | } |
154 | |
155 | if (d->delay < 0) { |
156 | av_log(ctx, AV_LOG_ERROR, "Delay must be non negative number.\n"); |
157 | return AVERROR(EINVAL); |
158 | } |
159 | } |
160 | |
161 | for (i = 0; i < s->nb_delays; i++) { |
162 | ChanDelay *d = &s->chandelay[i]; |
163 | |
164 | if (!d->delay) |
165 | continue; |
166 | |
167 | d->samples = av_malloc_array(d->delay, s->block_align); |
168 | if (!d->samples) |
169 | return AVERROR(ENOMEM); |
170 | |
171 | s->max_delay = FFMAX(s->max_delay, d->delay); |
172 | } |
173 | |
174 | if (!s->max_delay) { |
175 | av_log(ctx, AV_LOG_ERROR, "At least one delay >0 must be specified.\n"); |
176 | return AVERROR(EINVAL); |
177 | } |
178 | |
179 | switch (inlink->format) { |
180 | case AV_SAMPLE_FMT_U8P : s->delay_channel = delay_channel_u8p ; break; |
181 | case AV_SAMPLE_FMT_S16P: s->delay_channel = delay_channel_s16p; break; |
182 | case AV_SAMPLE_FMT_S32P: s->delay_channel = delay_channel_s32p; break; |
183 | case AV_SAMPLE_FMT_FLTP: s->delay_channel = delay_channel_fltp; break; |
184 | case AV_SAMPLE_FMT_DBLP: s->delay_channel = delay_channel_dblp; break; |
185 | } |
186 | |
187 | return 0; |
188 | } |
189 | |
190 | static int filter_frame(AVFilterLink *inlink, AVFrame *frame) |
191 | { |
192 | AVFilterContext *ctx = inlink->dst; |
193 | AudioDelayContext *s = ctx->priv; |
194 | AVFrame *out_frame; |
195 | int i; |
196 | |
197 | if (ctx->is_disabled || !s->delays) |
198 | return ff_filter_frame(ctx->outputs[0], frame); |
199 | |
200 | out_frame = ff_get_audio_buffer(inlink, frame->nb_samples); |
201 | if (!out_frame) { |
202 | av_frame_free(&frame); |
203 | return AVERROR(ENOMEM); |
204 | } |
205 | av_frame_copy_props(out_frame, frame); |
206 | |
207 | for (i = 0; i < s->nb_delays; i++) { |
208 | ChanDelay *d = &s->chandelay[i]; |
209 | const uint8_t *src = frame->extended_data[i]; |
210 | uint8_t *dst = out_frame->extended_data[i]; |
211 | |
212 | if (!d->delay) |
213 | memcpy(dst, src, frame->nb_samples * s->block_align); |
214 | else |
215 | s->delay_channel(d, frame->nb_samples, src, dst); |
216 | } |
217 | |
218 | s->next_pts = frame->pts + av_rescale_q(frame->nb_samples, (AVRational){1, inlink->sample_rate}, inlink->time_base); |
219 | av_frame_free(&frame); |
220 | return ff_filter_frame(ctx->outputs[0], out_frame); |
221 | } |
222 | |
223 | static int request_frame(AVFilterLink *outlink) |
224 | { |
225 | AVFilterContext *ctx = outlink->src; |
226 | AudioDelayContext *s = ctx->priv; |
227 | int ret; |
228 | |
229 | ret = ff_request_frame(ctx->inputs[0]); |
230 | if (ret == AVERROR_EOF && !ctx->is_disabled && s->max_delay) { |
231 | int nb_samples = FFMIN(s->max_delay, 2048); |
232 | AVFrame *frame; |
233 | |
234 | frame = ff_get_audio_buffer(outlink, nb_samples); |
235 | if (!frame) |
236 | return AVERROR(ENOMEM); |
237 | s->max_delay -= nb_samples; |
238 | |
239 | av_samples_set_silence(frame->extended_data, 0, |
240 | frame->nb_samples, |
241 | outlink->channels, |
242 | frame->format); |
243 | |
244 | frame->pts = s->next_pts; |
245 | if (s->next_pts != AV_NOPTS_VALUE) |
246 | s->next_pts += av_rescale_q(nb_samples, (AVRational){1, outlink->sample_rate}, outlink->time_base); |
247 | |
248 | ret = filter_frame(ctx->inputs[0], frame); |
249 | } |
250 | |
251 | return ret; |
252 | } |
253 | |
254 | static av_cold void uninit(AVFilterContext *ctx) |
255 | { |
256 | AudioDelayContext *s = ctx->priv; |
257 | int i; |
258 | |
259 | for (i = 0; i < s->nb_delays; i++) |
260 | av_freep(&s->chandelay[i].samples); |
261 | av_freep(&s->chandelay); |
262 | } |
263 | |
264 | static const AVFilterPad adelay_inputs[] = { |
265 | { |
266 | .name = "default", |
267 | .type = AVMEDIA_TYPE_AUDIO, |
268 | .config_props = config_input, |
269 | .filter_frame = filter_frame, |
270 | }, |
271 | { NULL } |
272 | }; |
273 | |
274 | static const AVFilterPad adelay_outputs[] = { |
275 | { |
276 | .name = "default", |
277 | .request_frame = request_frame, |
278 | .type = AVMEDIA_TYPE_AUDIO, |
279 | }, |
280 | { NULL } |
281 | }; |
282 | |
283 | AVFilter ff_af_adelay = { |
284 | .name = "adelay", |
285 | .description = NULL_IF_CONFIG_SMALL("Delay one or more audio channels."), |
286 | .query_formats = query_formats, |
287 | .priv_size = sizeof(AudioDelayContext), |
288 | .priv_class = &adelay_class, |
289 | .uninit = uninit, |
290 | .inputs = adelay_inputs, |
291 | .outputs = adelay_outputs, |
292 | .flags = AVFILTER_FLAG_SUPPORT_TIMELINE_INTERNAL, |
293 | }; |
294 |