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/audioconvert.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, const char *args)
  64 {
  65     VolumeContext *vol = ctx->priv;
  66     int ret;
  67
  68     vol->class = &volume_class;
  69     av_opt_set_defaults(vol);
  70
  71     if ((ret = av_set_options_string(vol, args, "=", ":")) < 0) {
  72         av_log(ctx, AV_LOG_ERROR, "Error parsing options string '%s'.\n", args);
  73         return ret;
  74     }
  75
  76     if (vol->precision == PRECISION_FIXED) {
  77         vol->volume_i = (int)(vol->volume * 256 + 0.5);
  78         vol->volume   = vol->volume_i / 256.0;
  79         av_log(ctx, AV_LOG_VERBOSE, "volume:(%d/256)(%f)(%1.2fdB) precision:fixed\n",
  80                vol->volume_i, vol->volume, 20.0*log(vol->volume)/M_LN10);
  81     } else {
  82         av_log(ctx, AV_LOG_VERBOSE, "volume:(%f)(%1.2fdB) precision:%s\n",
  83                vol->volume, 20.0*log(vol->volume)/M_LN10,
  84                precision_str[vol->precision]);
  85     }
  86
  87     av_opt_free(vol);
  88     return ret;
  89 }
  90
  91 static int query_formats(AVFilterContext *ctx)
  92 {
  93     VolumeContext *vol = ctx->priv;
  94     AVFilterFormats *formats = NULL;
  95     AVFilterChannelLayouts *layouts;
  96     static const enum AVSampleFormat sample_fmts[][7] = {
  97         /* PRECISION_FIXED */
  98         {
  99             AV_SAMPLE_FMT_U8,
 100             AV_SAMPLE_FMT_U8P,
 101             AV_SAMPLE_FMT_S16,
 102             AV_SAMPLE_FMT_S16P,
 103             AV_SAMPLE_FMT_S32,
 104             AV_SAMPLE_FMT_S32P,
 105             AV_SAMPLE_FMT_NONE
 106         },
 107         /* PRECISION_FLOAT */
 108         {
 109             AV_SAMPLE_FMT_FLT,
 110             AV_SAMPLE_FMT_FLTP,
 111             AV_SAMPLE_FMT_NONE
 112         },
 113         /* PRECISION_DOUBLE */
 114         {
 115             AV_SAMPLE_FMT_DBL,
 116             AV_SAMPLE_FMT_DBLP,
 117             AV_SAMPLE_FMT_NONE
 118         }
 119     };
 120
 121     layouts = ff_all_channel_layouts();
 122     if (!layouts)
 123         return AVERROR(ENOMEM);
 124     ff_set_common_channel_layouts(ctx, layouts);
 125
 126     formats = ff_make_format_list(sample_fmts[vol->precision]);
 127     if (!formats)
 128         return AVERROR(ENOMEM);
 129     ff_set_common_formats(ctx, formats);
 130
 131     formats = ff_all_samplerates();
 132     if (!formats)
 133         return AVERROR(ENOMEM);
 134     ff_set_common_samplerates(ctx, formats);
 135
 136     return 0;
 137 }
 138
 139 static inline void scale_samples_u8(uint8_t *dst, const uint8_t *src,
 140                                     int nb_samples, int volume)
 141 {
 142     int i;
 143     for (i = 0; i < nb_samples; i++)
 144         dst[i] = av_clip_uint8(((((int64_t)src[i] - 128) * volume + 128) >> 8) + 128);
 145 }
 146
 147 static inline void scale_samples_u8_small(uint8_t *dst, const uint8_t *src,
 148                                           int nb_samples, int volume)
 149 {
 150     int i;
 151     for (i = 0; i < nb_samples; i++)
 152         dst[i] = av_clip_uint8((((src[i] - 128) * volume + 128) >> 8) + 128);
 153 }
 154
 155 static inline void scale_samples_s16(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(((int64_t)smp_src[i] * volume + 128) >> 8);
 163 }
 164
 165 static inline void scale_samples_s16_small(uint8_t *dst, const uint8_t *src,
 166                                            int nb_samples, int volume)
 167 {
 168     int i;
 169     int16_t *smp_dst       = (int16_t *)dst;
 170     const int16_t *smp_src = (const int16_t *)src;
 171     for (i = 0; i < nb_samples; i++)
 172         smp_dst[i] = av_clip_int16((smp_src[i] * volume + 128) >> 8);
 173 }
 174
 175 static inline void scale_samples_s32(uint8_t *dst, const uint8_t *src,
 176                                      int nb_samples, int volume)
 177 {
 178     int i;
 179     int32_t *smp_dst       = (int32_t *)dst;
 180     const int32_t *smp_src = (const int32_t *)src;
 181     for (i = 0; i < nb_samples; i++)
 182         smp_dst[i] = av_clipl_int32((((int64_t)smp_src[i] * volume + 128) >> 8));
 183 }
 184
 185
 186
 187 static void volume_init(VolumeContext *vol)
 188 {
 189     vol->samples_align = 1;
 190
 191     switch (av_get_packed_sample_fmt(vol->sample_fmt)) {
 192     case AV_SAMPLE_FMT_U8:
 193         if (vol->volume_i < 0x1000000)
 194             vol->scale_samples = scale_samples_u8_small;
 195         else
 196             vol->scale_samples = scale_samples_u8;
 197         break;
 198     case AV_SAMPLE_FMT_S16:
 199         if (vol->volume_i < 0x10000)
 200             vol->scale_samples = scale_samples_s16_small;
 201         else
 202             vol->scale_samples = scale_samples_s16;
 203         break;
 204     case AV_SAMPLE_FMT_S32:
 205         vol->scale_samples = scale_samples_s32;
 206         break;
 207     case AV_SAMPLE_FMT_FLT:
 208         avpriv_float_dsp_init(&vol->fdsp, 0);
 209         vol->samples_align = 4;
 210         break;
 211     case AV_SAMPLE_FMT_DBL:
 212         avpriv_float_dsp_init(&vol->fdsp, 0);
 213         vol->samples_align = 8;
 214         break;
 215     }
 216
 217     if (ARCH_X86)
 218         ff_volume_init_x86(vol);
 219 }
 220
 221 static int config_output(AVFilterLink *outlink)
 222 {
 223     AVFilterContext *ctx = outlink->src;
 224     VolumeContext *vol   = ctx->priv;
 225     AVFilterLink *inlink = ctx->inputs[0];
 226
 227     vol->sample_fmt = inlink->format;
 228     vol->channels   = av_get_channel_layout_nb_channels(inlink->channel_layout);
 229     vol->planes     = av_sample_fmt_is_planar(inlink->format) ? vol->channels : 1;
 230
 231     volume_init(vol);
 232
 233     return 0;
 234 }
 235
 236 static int filter_frame(AVFilterLink *inlink, AVFrame *buf)
 237 {
 238     VolumeContext *vol    = inlink->dst->priv;
 239     AVFilterLink *outlink = inlink->dst->outputs[0];
 240     int nb_samples        = buf->nb_samples;
 241     AVFrame *out_buf;
 242
 243     if (vol->volume == 1.0 || vol->volume_i == 256)
 244         return ff_filter_frame(outlink, buf);
 245
 246     /* do volume scaling in-place if input buffer is writable */
 247     if (av_frame_is_writable(buf)) {
 248         out_buf = buf;
 249     } else {
 250         out_buf = ff_get_audio_buffer(inlink, nb_samples);
 251         if (!out_buf)
 252             return AVERROR(ENOMEM);
 253         out_buf->pts = buf->pts;
 254     }
 255
 256     if (vol->precision != PRECISION_FIXED || vol->volume_i > 0) {
 257         int p, plane_samples;
 258
 259         if (av_sample_fmt_is_planar(buf->format))
 260             plane_samples = FFALIGN(nb_samples, vol->samples_align);
 261         else
 262             plane_samples = FFALIGN(nb_samples * vol->channels, vol->samples_align);
 263
 264         if (vol->precision == PRECISION_FIXED) {
 265             for (p = 0; p < vol->planes; p++) {
 266                 vol->scale_samples(out_buf->extended_data[p],
 267                                    buf->extended_data[p], plane_samples,
 268                                    vol->volume_i);
 269             }
 270         } else if (av_get_packed_sample_fmt(vol->sample_fmt) == AV_SAMPLE_FMT_FLT) {
 271             for (p = 0; p < vol->planes; p++) {
 272                 vol->fdsp.vector_fmul_scalar((float *)out_buf->extended_data[p],
 273                                              (const float *)buf->extended_data[p],
 274                                              vol->volume, plane_samples);
 275             }
 276         } else {
 277             for (p = 0; p < vol->planes; p++) {
 278                 vol->fdsp.vector_dmul_scalar((double *)out_buf->extended_data[p],
 279                                              (const double *)buf->extended_data[p],
 280                                              vol->volume, plane_samples);
 281             }
 282         }
 283     }
 284
 285     if (buf != out_buf)
 286         av_frame_free(&buf);
 287
 288     return ff_filter_frame(outlink, out_buf);
 289 }
 290
 291 static const AVFilterPad avfilter_af_volume_inputs[] = {
 292     {
 293         .name           = "default",
 294         .type           = AVMEDIA_TYPE_AUDIO,
 295         .filter_frame   = filter_frame,
 296     },
 297     { NULL }
 298 };
 299
 300 static const AVFilterPad avfilter_af_volume_outputs[] = {
 301     {
 302         .name         = "default",
 303         .type         = AVMEDIA_TYPE_AUDIO,
 304         .config_props = config_output,
 305     },
 306     { NULL }
 307 };
 308
 309 AVFilter avfilter_af_volume = {
 310     .name           = "volume",
 311     .description    = NULL_IF_CONFIG_SMALL("Change input volume."),
 312     .query_formats  = query_formats,
 313     .priv_size      = sizeof(VolumeContext),
 314     .init           = init,
 315     .inputs         = avfilter_af_volume_inputs,
 316     .outputs        = avfilter_af_volume_outputs,
 317 };