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 * const 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|F, "replaygain" },
78 { "drop", "replaygain side data is dropped", 0, AV_OPT_TYPE_CONST, { .i64 = REPLAYGAIN_DROP }, 0, 0, A|F, "replaygain" },
79 { "ignore", "replaygain side data is ignored", 0, AV_OPT_TYPE_CONST, { .i64 = REPLAYGAIN_IGNORE }, 0, 0, A|F, "replaygain" },
80 { "track", "track gain is preferred", 0, AV_OPT_TYPE_CONST, { .i64 = REPLAYGAIN_TRACK }, 0, 0, A|F, "replaygain" },
81 { "album", "album gain is preferred", 0, AV_OPT_TYPE_CONST, { .i64 = REPLAYGAIN_ALBUM }, 0, 0, A|F, "replaygain" },
82 { "replaygain_preamp", "Apply replaygain pre-amplification",
83 OFFSET(replaygain_preamp), AV_OPT_TYPE_DOUBLE, { .dbl = 0.0 }, -15.0, 15.0, A|F },
84 { "replaygain_noclip", "Apply replaygain clipping prevention",
85 OFFSET(replaygain_noclip), AV_OPT_TYPE_BOOL, { .i64 = 1 }, 0, 1, A|F },
89 AVFILTER_DEFINE_CLASS(volume);
91 static int set_expr(AVExpr **pexpr, const char *expr, void *log_ctx)
98 ret = av_expr_parse(pexpr, expr, var_names,
99 NULL, NULL, NULL, NULL, 0, log_ctx);
101 av_log(log_ctx, AV_LOG_ERROR,
102 "Error when evaluating the volume expression '%s'\n", expr);
111 static av_cold int init(AVFilterContext *ctx)
113 VolumeContext *vol = ctx->priv;
115 vol->fdsp = avpriv_float_dsp_alloc(0);
117 return AVERROR(ENOMEM);
119 return set_expr(&vol->volume_pexpr, vol->volume_expr, ctx);
122 static av_cold void uninit(AVFilterContext *ctx)
124 VolumeContext *vol = ctx->priv;
125 av_expr_free(vol->volume_pexpr);
127 av_freep(&vol->fdsp);
130 static int query_formats(AVFilterContext *ctx)
132 VolumeContext *vol = ctx->priv;
133 AVFilterFormats *formats = NULL;
134 AVFilterChannelLayouts *layouts;
135 static const enum AVSampleFormat sample_fmts[][7] = {
136 [PRECISION_FIXED] = {
145 [PRECISION_FLOAT] = {
150 [PRECISION_DOUBLE] = {
158 layouts = ff_all_channel_counts();
160 return AVERROR(ENOMEM);
161 ret = ff_set_common_channel_layouts(ctx, layouts);
165 formats = ff_make_format_list(sample_fmts[vol->precision]);
167 return AVERROR(ENOMEM);
168 ret = ff_set_common_formats(ctx, formats);
172 formats = ff_all_samplerates();
174 return AVERROR(ENOMEM);
175 return ff_set_common_samplerates(ctx, formats);
178 static inline void scale_samples_u8(uint8_t *dst, const uint8_t *src,
179 int nb_samples, int volume)
182 for (i = 0; i < nb_samples; i++)
183 dst[i] = av_clip_uint8(((((int64_t)src[i] - 128) * volume + 128) >> 8) + 128);
186 static inline void scale_samples_u8_small(uint8_t *dst, const uint8_t *src,
187 int nb_samples, int volume)
190 for (i = 0; i < nb_samples; i++)
191 dst[i] = av_clip_uint8((((src[i] - 128) * volume + 128) >> 8) + 128);
194 static inline void scale_samples_s16(uint8_t *dst, const uint8_t *src,
195 int nb_samples, int volume)
198 int16_t *smp_dst = (int16_t *)dst;
199 const int16_t *smp_src = (const int16_t *)src;
200 for (i = 0; i < nb_samples; i++)
201 smp_dst[i] = av_clip_int16(((int64_t)smp_src[i] * volume + 128) >> 8);
204 static inline void scale_samples_s16_small(uint8_t *dst, const uint8_t *src,
205 int nb_samples, int volume)
208 int16_t *smp_dst = (int16_t *)dst;
209 const int16_t *smp_src = (const int16_t *)src;
210 for (i = 0; i < nb_samples; i++)
211 smp_dst[i] = av_clip_int16((smp_src[i] * volume + 128) >> 8);
214 static inline void scale_samples_s32(uint8_t *dst, const uint8_t *src,
215 int nb_samples, int volume)
218 int32_t *smp_dst = (int32_t *)dst;
219 const int32_t *smp_src = (const int32_t *)src;
220 for (i = 0; i < nb_samples; i++)
221 smp_dst[i] = av_clipl_int32((((int64_t)smp_src[i] * volume + 128) >> 8));
224 static av_cold void volume_init(VolumeContext *vol)
226 vol->samples_align = 1;
228 switch (av_get_packed_sample_fmt(vol->sample_fmt)) {
229 case AV_SAMPLE_FMT_U8:
230 if (vol->volume_i < 0x1000000)
231 vol->scale_samples = scale_samples_u8_small;
233 vol->scale_samples = scale_samples_u8;
235 case AV_SAMPLE_FMT_S16:
236 if (vol->volume_i < 0x10000)
237 vol->scale_samples = scale_samples_s16_small;
239 vol->scale_samples = scale_samples_s16;
241 case AV_SAMPLE_FMT_S32:
242 vol->scale_samples = scale_samples_s32;
244 case AV_SAMPLE_FMT_FLT:
245 vol->samples_align = 4;
247 case AV_SAMPLE_FMT_DBL:
248 vol->samples_align = 8;
253 ff_volume_init_x86(vol);
256 static int set_volume(AVFilterContext *ctx)
258 VolumeContext *vol = ctx->priv;
260 vol->volume = av_expr_eval(vol->volume_pexpr, vol->var_values, NULL);
261 if (isnan(vol->volume)) {
262 if (vol->eval_mode == EVAL_MODE_ONCE) {
263 av_log(ctx, AV_LOG_ERROR, "Invalid value NaN for volume\n");
264 return AVERROR(EINVAL);
266 av_log(ctx, AV_LOG_WARNING, "Invalid value NaN for volume, setting to 0\n");
270 vol->var_values[VAR_VOLUME] = vol->volume;
272 av_log(ctx, AV_LOG_VERBOSE, "n:%f t:%f pts:%f precision:%s ",
273 vol->var_values[VAR_N], vol->var_values[VAR_T], vol->var_values[VAR_PTS],
274 precision_str[vol->precision]);
276 if (vol->precision == PRECISION_FIXED) {
277 vol->volume_i = (int)(vol->volume * 256 + 0.5);
278 vol->volume = vol->volume_i / 256.0;
279 av_log(ctx, AV_LOG_VERBOSE, "volume_i:%d/255 ", vol->volume_i);
281 av_log(ctx, AV_LOG_VERBOSE, "volume:%f volume_dB:%f\n",
282 vol->volume, 20.0*log10(vol->volume));
288 static int config_output(AVFilterLink *outlink)
290 AVFilterContext *ctx = outlink->src;
291 VolumeContext *vol = ctx->priv;
292 AVFilterLink *inlink = ctx->inputs[0];
294 vol->sample_fmt = inlink->format;
295 vol->channels = inlink->channels;
296 vol->planes = av_sample_fmt_is_planar(inlink->format) ? vol->channels : 1;
298 vol->var_values[VAR_N] =
299 vol->var_values[VAR_NB_CONSUMED_SAMPLES] =
300 vol->var_values[VAR_NB_SAMPLES] =
301 vol->var_values[VAR_POS] =
302 vol->var_values[VAR_PTS] =
303 vol->var_values[VAR_STARTPTS] =
304 vol->var_values[VAR_STARTT] =
305 vol->var_values[VAR_T] =
306 vol->var_values[VAR_VOLUME] = NAN;
308 vol->var_values[VAR_NB_CHANNELS] = inlink->channels;
309 vol->var_values[VAR_TB] = av_q2d(inlink->time_base);
310 vol->var_values[VAR_SAMPLE_RATE] = inlink->sample_rate;
312 av_log(inlink->src, AV_LOG_VERBOSE, "tb:%f sample_rate:%f nb_channels:%f\n",
313 vol->var_values[VAR_TB],
314 vol->var_values[VAR_SAMPLE_RATE],
315 vol->var_values[VAR_NB_CHANNELS]);
317 return set_volume(ctx);
320 static int process_command(AVFilterContext *ctx, const char *cmd, const char *args,
321 char *res, int res_len, int flags)
323 VolumeContext *vol = ctx->priv;
324 int ret = AVERROR(ENOSYS);
326 if (!strcmp(cmd, "volume")) {
327 if ((ret = set_expr(&vol->volume_pexpr, args, ctx)) < 0)
329 if (vol->eval_mode == EVAL_MODE_ONCE)
336 #define D2TS(d) (isnan(d) ? AV_NOPTS_VALUE : (int64_t)(d))
337 #define TS2D(ts) ((ts) == AV_NOPTS_VALUE ? NAN : (double)(ts))
338 #define TS2T(ts, tb) ((ts) == AV_NOPTS_VALUE ? NAN : (double)(ts)*av_q2d(tb))
340 static int filter_frame(AVFilterLink *inlink, AVFrame *buf)
342 AVFilterContext *ctx = inlink->dst;
343 VolumeContext *vol = inlink->dst->priv;
344 AVFilterLink *outlink = inlink->dst->outputs[0];
345 int nb_samples = buf->nb_samples;
348 AVFrameSideData *sd = av_frame_get_side_data(buf, AV_FRAME_DATA_REPLAYGAIN);
351 if (sd && vol->replaygain != REPLAYGAIN_IGNORE) {
352 if (vol->replaygain != REPLAYGAIN_DROP) {
353 AVReplayGain *replaygain = (AVReplayGain*)sd->data;
354 int32_t gain = 100000;
355 uint32_t peak = 100000;
358 if (vol->replaygain == REPLAYGAIN_TRACK &&
359 replaygain->track_gain != INT32_MIN) {
360 gain = replaygain->track_gain;
362 if (replaygain->track_peak != 0)
363 peak = replaygain->track_peak;
364 } else if (replaygain->album_gain != INT32_MIN) {
365 gain = replaygain->album_gain;
367 if (replaygain->album_peak != 0)
368 peak = replaygain->album_peak;
370 av_log(inlink->dst, AV_LOG_WARNING, "Both ReplayGain gain "
371 "values are unknown.\n");
373 g = gain / 100000.0f;
374 p = peak / 100000.0f;
376 av_log(inlink->dst, AV_LOG_VERBOSE,
377 "Using gain %f dB from replaygain side data.\n", g);
379 vol->volume = ff_exp10((g + vol->replaygain_preamp) / 20);
380 if (vol->replaygain_noclip)
381 vol->volume = FFMIN(vol->volume, 1.0 / p);
382 vol->volume_i = (int)(vol->volume * 256 + 0.5);
386 av_frame_remove_side_data(buf, AV_FRAME_DATA_REPLAYGAIN);
389 if (isnan(vol->var_values[VAR_STARTPTS])) {
390 vol->var_values[VAR_STARTPTS] = TS2D(buf->pts);
391 vol->var_values[VAR_STARTT ] = TS2T(buf->pts, inlink->time_base);
393 vol->var_values[VAR_PTS] = TS2D(buf->pts);
394 vol->var_values[VAR_T ] = TS2T(buf->pts, inlink->time_base);
395 vol->var_values[VAR_N ] = inlink->frame_count;
397 pos = av_frame_get_pkt_pos(buf);
398 vol->var_values[VAR_POS] = pos == -1 ? NAN : pos;
399 if (vol->eval_mode == EVAL_MODE_FRAME)
402 if (vol->volume == 1.0 || vol->volume_i == 256) {
407 /* do volume scaling in-place if input buffer is writable */
408 if (av_frame_is_writable(buf)
409 && (vol->precision != PRECISION_FIXED || vol->volume_i > 0)) {
412 out_buf = ff_get_audio_buffer(inlink, nb_samples);
414 return AVERROR(ENOMEM);
415 ret = av_frame_copy_props(out_buf, buf);
417 av_frame_free(&out_buf);
423 if (vol->precision != PRECISION_FIXED || vol->volume_i > 0) {
424 int p, plane_samples;
426 if (av_sample_fmt_is_planar(buf->format))
427 plane_samples = FFALIGN(nb_samples, vol->samples_align);
429 plane_samples = FFALIGN(nb_samples * vol->channels, vol->samples_align);
431 if (vol->precision == PRECISION_FIXED) {
432 for (p = 0; p < vol->planes; p++) {
433 vol->scale_samples(out_buf->extended_data[p],
434 buf->extended_data[p], plane_samples,
437 } else if (av_get_packed_sample_fmt(vol->sample_fmt) == AV_SAMPLE_FMT_FLT) {
438 for (p = 0; p < vol->planes; p++) {
439 vol->fdsp->vector_fmul_scalar((float *)out_buf->extended_data[p],
440 (const float *)buf->extended_data[p],
441 vol->volume, plane_samples);
444 for (p = 0; p < vol->planes; p++) {
445 vol->fdsp->vector_dmul_scalar((double *)out_buf->extended_data[p],
446 (const double *)buf->extended_data[p],
447 vol->volume, plane_samples);
458 vol->var_values[VAR_NB_CONSUMED_SAMPLES] += out_buf->nb_samples;
459 return ff_filter_frame(outlink, out_buf);
462 static const AVFilterPad avfilter_af_volume_inputs[] = {
465 .type = AVMEDIA_TYPE_AUDIO,
466 .filter_frame = filter_frame,
471 static const AVFilterPad avfilter_af_volume_outputs[] = {
474 .type = AVMEDIA_TYPE_AUDIO,
475 .config_props = config_output,
480 AVFilter ff_af_volume = {
482 .description = NULL_IF_CONFIG_SMALL("Change input volume."),
483 .query_formats = query_formats,
484 .priv_size = sizeof(VolumeContext),
485 .priv_class = &volume_class,
488 .inputs = avfilter_af_volume_inputs,
489 .outputs = avfilter_af_volume_outputs,
490 .flags = AVFILTER_FLAG_SUPPORT_TIMELINE_GENERIC,
491 .process_command = process_command,