2 * Copyright (c) 2011 Stefano Sabatini
3 * Copyright (c) 2012 Justin Ruggles <justin.ruggles@gmail.com>
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
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/intreadwrite.h"
32 #include "libavutil/opt.h"
33 #include "libavutil/replaygain.h"
39 #include "af_volume.h"
41 static const char *precision_str[] = {
42 "fixed", "float", "double"
45 static const char *const var_names[] = {
46 "n", ///< frame number (starting at zero)
47 "nb_channels", ///< number of channels
48 "nb_consumed_samples", ///< number of samples consumed by the filter
49 "nb_samples", ///< number of samples in the current frame
50 "pos", ///< position in the file of the frame
51 "pts", ///< frame presentation timestamp
52 "sample_rate", ///< sample rate
53 "startpts", ///< PTS at start of stream
54 "startt", ///< time at start of stream
55 "t", ///< time in the file of the frame
57 "volume", ///< last set value
61 #define OFFSET(x) offsetof(VolumeContext, x)
62 #define A AV_OPT_FLAG_AUDIO_PARAM
63 #define F AV_OPT_FLAG_FILTERING_PARAM
65 static const AVOption volume_options[] = {
66 { "volume", "set volume adjustment expression",
67 OFFSET(volume_expr), AV_OPT_TYPE_STRING, { .str = "1.0" }, .flags = A|F },
68 { "precision", "select mathematical precision",
69 OFFSET(precision), AV_OPT_TYPE_INT, { .i64 = PRECISION_FLOAT }, PRECISION_FIXED, PRECISION_DOUBLE, A|F, "precision" },
70 { "fixed", "select 8-bit fixed-point", 0, AV_OPT_TYPE_CONST, { .i64 = PRECISION_FIXED }, INT_MIN, INT_MAX, A|F, "precision" },
71 { "float", "select 32-bit floating-point", 0, AV_OPT_TYPE_CONST, { .i64 = PRECISION_FLOAT }, INT_MIN, INT_MAX, A|F, "precision" },
72 { "double", "select 64-bit floating-point", 0, AV_OPT_TYPE_CONST, { .i64 = PRECISION_DOUBLE }, INT_MIN, INT_MAX, A|F, "precision" },
73 { "eval", "specify when to evaluate expressions", OFFSET(eval_mode), AV_OPT_TYPE_INT, {.i64 = EVAL_MODE_ONCE}, 0, EVAL_MODE_NB-1, .flags = A|F, "eval" },
74 { "once", "eval volume expression once", 0, AV_OPT_TYPE_CONST, {.i64=EVAL_MODE_ONCE}, .flags = A|F, .unit = "eval" },
75 { "frame", "eval volume expression per-frame", 0, AV_OPT_TYPE_CONST, {.i64=EVAL_MODE_FRAME}, .flags = A|F, .unit = "eval" },
76 { "replaygain", "Apply replaygain side data when present",
77 OFFSET(replaygain), AV_OPT_TYPE_INT, { .i64 = REPLAYGAIN_DROP }, REPLAYGAIN_DROP, REPLAYGAIN_ALBUM, A, "replaygain" },
78 { "drop", "replaygain side data is dropped", 0, AV_OPT_TYPE_CONST, { .i64 = REPLAYGAIN_DROP }, 0, 0, A, "replaygain" },
79 { "ignore", "replaygain side data is ignored", 0, AV_OPT_TYPE_CONST, { .i64 = REPLAYGAIN_IGNORE }, 0, 0, A, "replaygain" },
80 { "track", "track gain is preferred", 0, AV_OPT_TYPE_CONST, { .i64 = REPLAYGAIN_TRACK }, 0, 0, A, "replaygain" },
81 { "album", "album gain is preferred", 0, AV_OPT_TYPE_CONST, { .i64 = REPLAYGAIN_ALBUM }, 0, 0, A, "replaygain" },
85 AVFILTER_DEFINE_CLASS(volume);
87 static int set_expr(AVExpr **pexpr, const char *expr, void *log_ctx)
94 ret = av_expr_parse(pexpr, expr, var_names,
95 NULL, NULL, NULL, NULL, 0, log_ctx);
97 av_log(log_ctx, AV_LOG_ERROR,
98 "Error when evaluating the volume expression '%s'\n", expr);
107 static av_cold int init(AVFilterContext *ctx)
109 VolumeContext *vol = ctx->priv;
110 return set_expr(&vol->volume_pexpr, vol->volume_expr, ctx);
113 static av_cold void uninit(AVFilterContext *ctx)
115 VolumeContext *vol = ctx->priv;
116 av_expr_free(vol->volume_pexpr);
120 static int query_formats(AVFilterContext *ctx)
122 VolumeContext *vol = ctx->priv;
123 AVFilterFormats *formats = NULL;
124 AVFilterChannelLayouts *layouts;
125 static const enum AVSampleFormat sample_fmts[][7] = {
126 [PRECISION_FIXED] = {
135 [PRECISION_FLOAT] = {
140 [PRECISION_DOUBLE] = {
147 layouts = ff_all_channel_counts();
149 return AVERROR(ENOMEM);
150 ff_set_common_channel_layouts(ctx, layouts);
152 formats = ff_make_format_list(sample_fmts[vol->precision]);
154 return AVERROR(ENOMEM);
155 ff_set_common_formats(ctx, formats);
157 formats = ff_all_samplerates();
159 return AVERROR(ENOMEM);
160 ff_set_common_samplerates(ctx, formats);
165 static inline void scale_samples_u8(uint8_t *dst, const uint8_t *src,
166 int nb_samples, int volume)
169 for (i = 0; i < nb_samples; i++)
170 dst[i] = av_clip_uint8(((((int64_t)src[i] - 128) * volume + 128) >> 8) + 128);
173 static inline void scale_samples_u8_small(uint8_t *dst, const uint8_t *src,
174 int nb_samples, int volume)
177 for (i = 0; i < nb_samples; i++)
178 dst[i] = av_clip_uint8((((src[i] - 128) * volume + 128) >> 8) + 128);
181 static inline void scale_samples_s16(uint8_t *dst, const uint8_t *src,
182 int nb_samples, int volume)
185 int16_t *smp_dst = (int16_t *)dst;
186 const int16_t *smp_src = (const int16_t *)src;
187 for (i = 0; i < nb_samples; i++)
188 smp_dst[i] = av_clip_int16(((int64_t)smp_src[i] * volume + 128) >> 8);
191 static inline void scale_samples_s16_small(uint8_t *dst, const uint8_t *src,
192 int nb_samples, int volume)
195 int16_t *smp_dst = (int16_t *)dst;
196 const int16_t *smp_src = (const int16_t *)src;
197 for (i = 0; i < nb_samples; i++)
198 smp_dst[i] = av_clip_int16((smp_src[i] * volume + 128) >> 8);
201 static inline void scale_samples_s32(uint8_t *dst, const uint8_t *src,
202 int nb_samples, int volume)
205 int32_t *smp_dst = (int32_t *)dst;
206 const int32_t *smp_src = (const int32_t *)src;
207 for (i = 0; i < nb_samples; i++)
208 smp_dst[i] = av_clipl_int32((((int64_t)smp_src[i] * volume + 128) >> 8));
211 static av_cold void volume_init(VolumeContext *vol)
213 vol->samples_align = 1;
215 switch (av_get_packed_sample_fmt(vol->sample_fmt)) {
216 case AV_SAMPLE_FMT_U8:
217 if (vol->volume_i < 0x1000000)
218 vol->scale_samples = scale_samples_u8_small;
220 vol->scale_samples = scale_samples_u8;
222 case AV_SAMPLE_FMT_S16:
223 if (vol->volume_i < 0x10000)
224 vol->scale_samples = scale_samples_s16_small;
226 vol->scale_samples = scale_samples_s16;
228 case AV_SAMPLE_FMT_S32:
229 vol->scale_samples = scale_samples_s32;
231 case AV_SAMPLE_FMT_FLT:
232 avpriv_float_dsp_init(&vol->fdsp, 0);
233 vol->samples_align = 4;
235 case AV_SAMPLE_FMT_DBL:
236 avpriv_float_dsp_init(&vol->fdsp, 0);
237 vol->samples_align = 8;
242 ff_volume_init_x86(vol);
245 static int set_volume(AVFilterContext *ctx)
247 VolumeContext *vol = ctx->priv;
249 vol->volume = av_expr_eval(vol->volume_pexpr, vol->var_values, NULL);
250 if (isnan(vol->volume)) {
251 if (vol->eval_mode == EVAL_MODE_ONCE) {
252 av_log(ctx, AV_LOG_ERROR, "Invalid value NaN for volume\n");
253 return AVERROR(EINVAL);
255 av_log(ctx, AV_LOG_WARNING, "Invalid value NaN for volume, setting to 0\n");
259 vol->var_values[VAR_VOLUME] = vol->volume;
261 av_log(ctx, AV_LOG_VERBOSE, "n:%f t:%f pts:%f precision:%s ",
262 vol->var_values[VAR_N], vol->var_values[VAR_T], vol->var_values[VAR_PTS],
263 precision_str[vol->precision]);
265 if (vol->precision == PRECISION_FIXED) {
266 vol->volume_i = (int)(vol->volume * 256 + 0.5);
267 vol->volume = vol->volume_i / 256.0;
268 av_log(ctx, AV_LOG_VERBOSE, "volume_i:%d/255 ", vol->volume_i);
270 av_log(ctx, AV_LOG_VERBOSE, "volume:%f volume_dB:%f\n",
271 vol->volume, 20.0*log(vol->volume)/M_LN10);
277 static int config_output(AVFilterLink *outlink)
279 AVFilterContext *ctx = outlink->src;
280 VolumeContext *vol = ctx->priv;
281 AVFilterLink *inlink = ctx->inputs[0];
283 vol->sample_fmt = inlink->format;
284 vol->channels = inlink->channels;
285 vol->planes = av_sample_fmt_is_planar(inlink->format) ? vol->channels : 1;
287 vol->var_values[VAR_N] =
288 vol->var_values[VAR_NB_CONSUMED_SAMPLES] =
289 vol->var_values[VAR_NB_SAMPLES] =
290 vol->var_values[VAR_POS] =
291 vol->var_values[VAR_PTS] =
292 vol->var_values[VAR_STARTPTS] =
293 vol->var_values[VAR_STARTT] =
294 vol->var_values[VAR_T] =
295 vol->var_values[VAR_VOLUME] = NAN;
297 vol->var_values[VAR_NB_CHANNELS] = inlink->channels;
298 vol->var_values[VAR_TB] = av_q2d(inlink->time_base);
299 vol->var_values[VAR_SAMPLE_RATE] = inlink->sample_rate;
301 av_log(inlink->src, AV_LOG_VERBOSE, "tb:%f sample_rate:%f nb_channels:%f\n",
302 vol->var_values[VAR_TB],
303 vol->var_values[VAR_SAMPLE_RATE],
304 vol->var_values[VAR_NB_CHANNELS]);
306 return set_volume(ctx);
309 static int process_command(AVFilterContext *ctx, const char *cmd, const char *args,
310 char *res, int res_len, int flags)
312 VolumeContext *vol = ctx->priv;
313 int ret = AVERROR(ENOSYS);
315 if (!strcmp(cmd, "volume")) {
316 if ((ret = set_expr(&vol->volume_pexpr, args, ctx)) < 0)
318 if (vol->eval_mode == EVAL_MODE_ONCE)
325 #define D2TS(d) (isnan(d) ? AV_NOPTS_VALUE : (int64_t)(d))
326 #define TS2D(ts) ((ts) == AV_NOPTS_VALUE ? NAN : (double)(ts))
327 #define TS2T(ts, tb) ((ts) == AV_NOPTS_VALUE ? NAN : (double)(ts)*av_q2d(tb))
329 static int filter_frame(AVFilterLink *inlink, AVFrame *buf)
331 AVFilterContext *ctx = inlink->dst;
332 VolumeContext *vol = inlink->dst->priv;
333 AVFilterLink *outlink = inlink->dst->outputs[0];
334 int nb_samples = buf->nb_samples;
337 AVFrameSideData *sd = av_frame_get_side_data(buf, AV_FRAME_DATA_REPLAYGAIN);
340 if (sd && vol->replaygain != REPLAYGAIN_IGNORE) {
341 if (vol->replaygain != REPLAYGAIN_DROP) {
342 AVReplayGain *replaygain = (AVReplayGain*)sd->data;
346 if (vol->replaygain == REPLAYGAIN_TRACK &&
347 replaygain->track_gain != INT32_MIN)
348 gain = replaygain->track_gain;
349 else if (replaygain->album_gain != INT32_MIN)
350 gain = replaygain->album_gain;
352 av_log(inlink->dst, AV_LOG_WARNING, "Both ReplayGain gain "
353 "values are unknown.\n");
356 g = gain / 100000.0f;
358 av_log(inlink->dst, AV_LOG_VERBOSE,
359 "Using gain %f dB from replaygain side data.\n", g);
361 vol->volume = pow(10, g / 20);
362 vol->volume_i = (int)(vol->volume * 256 + 0.5);
366 av_frame_remove_side_data(buf, AV_FRAME_DATA_REPLAYGAIN);
369 if (isnan(vol->var_values[VAR_STARTPTS])) {
370 vol->var_values[VAR_STARTPTS] = TS2D(buf->pts);
371 vol->var_values[VAR_STARTT ] = TS2T(buf->pts, inlink->time_base);
373 vol->var_values[VAR_PTS] = TS2D(buf->pts);
374 vol->var_values[VAR_T ] = TS2T(buf->pts, inlink->time_base);
375 vol->var_values[VAR_N ] = inlink->frame_count;
377 pos = av_frame_get_pkt_pos(buf);
378 vol->var_values[VAR_POS] = pos == -1 ? NAN : pos;
379 if (vol->eval_mode == EVAL_MODE_FRAME)
382 if (vol->volume == 1.0 || vol->volume_i == 256) {
387 /* do volume scaling in-place if input buffer is writable */
388 if (av_frame_is_writable(buf)) {
391 out_buf = ff_get_audio_buffer(inlink, nb_samples);
393 return AVERROR(ENOMEM);
394 ret = av_frame_copy_props(out_buf, buf);
396 av_frame_free(&out_buf);
402 if (vol->precision != PRECISION_FIXED || vol->volume_i > 0) {
403 int p, plane_samples;
405 if (av_sample_fmt_is_planar(buf->format))
406 plane_samples = FFALIGN(nb_samples, vol->samples_align);
408 plane_samples = FFALIGN(nb_samples * vol->channels, vol->samples_align);
410 if (vol->precision == PRECISION_FIXED) {
411 for (p = 0; p < vol->planes; p++) {
412 vol->scale_samples(out_buf->extended_data[p],
413 buf->extended_data[p], plane_samples,
416 } else if (av_get_packed_sample_fmt(vol->sample_fmt) == AV_SAMPLE_FMT_FLT) {
417 for (p = 0; p < vol->planes; p++) {
418 vol->fdsp.vector_fmul_scalar((float *)out_buf->extended_data[p],
419 (const float *)buf->extended_data[p],
420 vol->volume, plane_samples);
423 for (p = 0; p < vol->planes; p++) {
424 vol->fdsp.vector_dmul_scalar((double *)out_buf->extended_data[p],
425 (const double *)buf->extended_data[p],
426 vol->volume, plane_samples);
437 vol->var_values[VAR_NB_CONSUMED_SAMPLES] += out_buf->nb_samples;
438 return ff_filter_frame(outlink, out_buf);
441 static const AVFilterPad avfilter_af_volume_inputs[] = {
444 .type = AVMEDIA_TYPE_AUDIO,
445 .filter_frame = filter_frame,
450 static const AVFilterPad avfilter_af_volume_outputs[] = {
453 .type = AVMEDIA_TYPE_AUDIO,
454 .config_props = config_output,
459 AVFilter ff_af_volume = {
461 .description = NULL_IF_CONFIG_SMALL("Change input volume."),
462 .query_formats = query_formats,
463 .priv_size = sizeof(VolumeContext),
464 .priv_class = &volume_class,
467 .inputs = avfilter_af_volume_inputs,
468 .outputs = avfilter_af_volume_outputs,
469 .flags = AVFILTER_FLAG_SUPPORT_TIMELINE_GENERIC,
470 .process_command = process_command,