af_volume: preserve frame properties
[libav.git] / libavfilter / af_volume.c
1 /*
2 * Copyright (c) 2011 Stefano Sabatini
3 * Copyright (c) 2012 Justin Ruggles <justin.ruggles@gmail.com>
4 *
5 * This file is part of Libav.
6 *
7 * Libav 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 * Libav 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 Libav; 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 * audio volume filter
25 */
26
27 #include "libavutil/channel_layout.h"
28 #include "libavutil/common.h"
29 #include "libavutil/eval.h"
30 #include "libavutil/float_dsp.h"
31 #include "libavutil/opt.h"
32 #include "audio.h"
33 #include "avfilter.h"
34 #include "formats.h"
35 #include "internal.h"
36 #include "af_volume.h"
37
38 static const char *precision_str[] = {
39 "fixed", "float", "double"
40 };
41
42 #define OFFSET(x) offsetof(VolumeContext, x)
43 #define A AV_OPT_FLAG_AUDIO_PARAM
44
45 static const AVOption options[] = {
46 { "volume", "Volume adjustment.",
47 OFFSET(volume), AV_OPT_TYPE_DOUBLE, { .dbl = 1.0 }, 0, 0x7fffff, A },
48 { "precision", "Mathematical precision.",
49 OFFSET(precision), AV_OPT_TYPE_INT, { .i64 = PRECISION_FLOAT }, PRECISION_FIXED, PRECISION_DOUBLE, A, "precision" },
50 { "fixed", "8-bit fixed-point.", 0, AV_OPT_TYPE_CONST, { .i64 = PRECISION_FIXED }, INT_MIN, INT_MAX, A, "precision" },
51 { "float", "32-bit floating-point.", 0, AV_OPT_TYPE_CONST, { .i64 = PRECISION_FLOAT }, INT_MIN, INT_MAX, A, "precision" },
52 { "double", "64-bit floating-point.", 0, AV_OPT_TYPE_CONST, { .i64 = PRECISION_DOUBLE }, INT_MIN, INT_MAX, A, "precision" },
53 { NULL },
54 };
55
56 static const AVClass volume_class = {
57 .class_name = "volume filter",
58 .item_name = av_default_item_name,
59 .option = options,
60 .version = LIBAVUTIL_VERSION_INT,
61 };
62
63 static av_cold int init(AVFilterContext *ctx)
64 {
65 VolumeContext *vol = ctx->priv;
66
67 if (vol->precision == PRECISION_FIXED) {
68 vol->volume_i = (int)(vol->volume * 256 + 0.5);
69 vol->volume = vol->volume_i / 256.0;
70 av_log(ctx, AV_LOG_VERBOSE, "volume:(%d/256)(%f)(%1.2fdB) precision:fixed\n",
71 vol->volume_i, vol->volume, 20.0*log(vol->volume)/M_LN10);
72 } else {
73 av_log(ctx, AV_LOG_VERBOSE, "volume:(%f)(%1.2fdB) precision:%s\n",
74 vol->volume, 20.0*log(vol->volume)/M_LN10,
75 precision_str[vol->precision]);
76 }
77
78 return 0;
79 }
80
81 static int query_formats(AVFilterContext *ctx)
82 {
83 VolumeContext *vol = ctx->priv;
84 AVFilterFormats *formats = NULL;
85 AVFilterChannelLayouts *layouts;
86 static const enum AVSampleFormat sample_fmts[][7] = {
87 /* PRECISION_FIXED */
88 {
89 AV_SAMPLE_FMT_U8,
90 AV_SAMPLE_FMT_U8P,
91 AV_SAMPLE_FMT_S16,
92 AV_SAMPLE_FMT_S16P,
93 AV_SAMPLE_FMT_S32,
94 AV_SAMPLE_FMT_S32P,
95 AV_SAMPLE_FMT_NONE
96 },
97 /* PRECISION_FLOAT */
98 {
99 AV_SAMPLE_FMT_FLT,
100 AV_SAMPLE_FMT_FLTP,
101 AV_SAMPLE_FMT_NONE
102 },
103 /* PRECISION_DOUBLE */
104 {
105 AV_SAMPLE_FMT_DBL,
106 AV_SAMPLE_FMT_DBLP,
107 AV_SAMPLE_FMT_NONE
108 }
109 };
110
111 layouts = ff_all_channel_layouts();
112 if (!layouts)
113 return AVERROR(ENOMEM);
114 ff_set_common_channel_layouts(ctx, layouts);
115
116 formats = ff_make_format_list(sample_fmts[vol->precision]);
117 if (!formats)
118 return AVERROR(ENOMEM);
119 ff_set_common_formats(ctx, formats);
120
121 formats = ff_all_samplerates();
122 if (!formats)
123 return AVERROR(ENOMEM);
124 ff_set_common_samplerates(ctx, formats);
125
126 return 0;
127 }
128
129 static inline void scale_samples_u8(uint8_t *dst, const uint8_t *src,
130 int nb_samples, int volume)
131 {
132 int i;
133 for (i = 0; i < nb_samples; i++)
134 dst[i] = av_clip_uint8(((((int64_t)src[i] - 128) * volume + 128) >> 8) + 128);
135 }
136
137 static inline void scale_samples_u8_small(uint8_t *dst, const uint8_t *src,
138 int nb_samples, int volume)
139 {
140 int i;
141 for (i = 0; i < nb_samples; i++)
142 dst[i] = av_clip_uint8((((src[i] - 128) * volume + 128) >> 8) + 128);
143 }
144
145 static inline void scale_samples_s16(uint8_t *dst, const uint8_t *src,
146 int nb_samples, int volume)
147 {
148 int i;
149 int16_t *smp_dst = (int16_t *)dst;
150 const int16_t *smp_src = (const int16_t *)src;
151 for (i = 0; i < nb_samples; i++)
152 smp_dst[i] = av_clip_int16(((int64_t)smp_src[i] * volume + 128) >> 8);
153 }
154
155 static inline void scale_samples_s16_small(uint8_t *dst, const uint8_t *src,
156 int nb_samples, int volume)
157 {
158 int i;
159 int16_t *smp_dst = (int16_t *)dst;
160 const int16_t *smp_src = (const int16_t *)src;
161 for (i = 0; i < nb_samples; i++)
162 smp_dst[i] = av_clip_int16((smp_src[i] * volume + 128) >> 8);
163 }
164
165 static inline void scale_samples_s32(uint8_t *dst, const uint8_t *src,
166 int nb_samples, int volume)
167 {
168 int i;
169 int32_t *smp_dst = (int32_t *)dst;
170 const int32_t *smp_src = (const int32_t *)src;
171 for (i = 0; i < nb_samples; i++)
172 smp_dst[i] = av_clipl_int32((((int64_t)smp_src[i] * volume + 128) >> 8));
173 }
174
175
176
177 static av_cold void volume_init(VolumeContext *vol)
178 {
179 vol->samples_align = 1;
180
181 switch (av_get_packed_sample_fmt(vol->sample_fmt)) {
182 case AV_SAMPLE_FMT_U8:
183 if (vol->volume_i < 0x1000000)
184 vol->scale_samples = scale_samples_u8_small;
185 else
186 vol->scale_samples = scale_samples_u8;
187 break;
188 case AV_SAMPLE_FMT_S16:
189 if (vol->volume_i < 0x10000)
190 vol->scale_samples = scale_samples_s16_small;
191 else
192 vol->scale_samples = scale_samples_s16;
193 break;
194 case AV_SAMPLE_FMT_S32:
195 vol->scale_samples = scale_samples_s32;
196 break;
197 case AV_SAMPLE_FMT_FLT:
198 avpriv_float_dsp_init(&vol->fdsp, 0);
199 vol->samples_align = 4;
200 break;
201 case AV_SAMPLE_FMT_DBL:
202 avpriv_float_dsp_init(&vol->fdsp, 0);
203 vol->samples_align = 8;
204 break;
205 }
206
207 if (ARCH_X86)
208 ff_volume_init_x86(vol);
209 }
210
211 static int config_output(AVFilterLink *outlink)
212 {
213 AVFilterContext *ctx = outlink->src;
214 VolumeContext *vol = ctx->priv;
215 AVFilterLink *inlink = ctx->inputs[0];
216
217 vol->sample_fmt = inlink->format;
218 vol->channels = av_get_channel_layout_nb_channels(inlink->channel_layout);
219 vol->planes = av_sample_fmt_is_planar(inlink->format) ? vol->channels : 1;
220
221 volume_init(vol);
222
223 return 0;
224 }
225
226 static int filter_frame(AVFilterLink *inlink, AVFrame *buf)
227 {
228 VolumeContext *vol = inlink->dst->priv;
229 AVFilterLink *outlink = inlink->dst->outputs[0];
230 int nb_samples = buf->nb_samples;
231 AVFrame *out_buf;
232 int ret;
233
234 if (vol->volume == 1.0 || vol->volume_i == 256)
235 return ff_filter_frame(outlink, buf);
236
237 /* do volume scaling in-place if input buffer is writable */
238 if (av_frame_is_writable(buf)) {
239 out_buf = buf;
240 } else {
241 out_buf = ff_get_audio_buffer(inlink, nb_samples);
242 if (!out_buf)
243 return AVERROR(ENOMEM);
244 ret = av_frame_copy_props(out_buf, buf);
245 if (ret < 0) {
246 av_frame_free(&out_buf);
247 av_frame_free(&buf);
248 return ret;
249 }
250 }
251
252 if (vol->precision != PRECISION_FIXED || vol->volume_i > 0) {
253 int p, plane_samples;
254
255 if (av_sample_fmt_is_planar(buf->format))
256 plane_samples = FFALIGN(nb_samples, vol->samples_align);
257 else
258 plane_samples = FFALIGN(nb_samples * vol->channels, vol->samples_align);
259
260 if (vol->precision == PRECISION_FIXED) {
261 for (p = 0; p < vol->planes; p++) {
262 vol->scale_samples(out_buf->extended_data[p],
263 buf->extended_data[p], plane_samples,
264 vol->volume_i);
265 }
266 } else if (av_get_packed_sample_fmt(vol->sample_fmt) == AV_SAMPLE_FMT_FLT) {
267 for (p = 0; p < vol->planes; p++) {
268 vol->fdsp.vector_fmul_scalar((float *)out_buf->extended_data[p],
269 (const float *)buf->extended_data[p],
270 vol->volume, plane_samples);
271 }
272 } else {
273 for (p = 0; p < vol->planes; p++) {
274 vol->fdsp.vector_dmul_scalar((double *)out_buf->extended_data[p],
275 (const double *)buf->extended_data[p],
276 vol->volume, plane_samples);
277 }
278 }
279 }
280
281 if (buf != out_buf)
282 av_frame_free(&buf);
283
284 return ff_filter_frame(outlink, out_buf);
285 }
286
287 static const AVFilterPad avfilter_af_volume_inputs[] = {
288 {
289 .name = "default",
290 .type = AVMEDIA_TYPE_AUDIO,
291 .filter_frame = filter_frame,
292 },
293 { NULL }
294 };
295
296 static const AVFilterPad avfilter_af_volume_outputs[] = {
297 {
298 .name = "default",
299 .type = AVMEDIA_TYPE_AUDIO,
300 .config_props = config_output,
301 },
302 { NULL }
303 };
304
305 AVFilter ff_af_volume = {
306 .name = "volume",
307 .description = NULL_IF_CONFIG_SMALL("Change input volume."),
308 .query_formats = query_formats,
309 .priv_size = sizeof(VolumeContext),
310 .priv_class = &volume_class,
311 .init = init,
312 .inputs = avfilter_af_volume_inputs,
313 .outputs = avfilter_af_volume_outputs,
314 };