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/opt.h"
36 #include "af_volume.h"
38 static const char *precision_str[] = {
39 "fixed", "float", "double"
42 static const char *const var_names[] = {
43 "n", ///< frame number (starting at zero)
44 "nb_channels", ///< number of channels
45 "nb_consumed_samples", ///< number of samples consumed by the filter
46 "nb_samples", ///< number of samples in the current frame
47 "pos", ///< position in the file of the frame
48 "pts", ///< frame presentation timestamp
49 "sample_rate", ///< sample rate
50 "startpts", ///< PTS at start of stream
51 "startt", ///< time at start of stream
52 "t", ///< time in the file of the frame
54 "volume", ///< last set value
58 #define OFFSET(x) offsetof(VolumeContext, x)
59 #define A AV_OPT_FLAG_AUDIO_PARAM
60 #define F AV_OPT_FLAG_FILTERING_PARAM
62 static const AVOption volume_options[] = {
63 { "volume", "set volume adjustment expression",
64 OFFSET(volume_expr), AV_OPT_TYPE_STRING, { .str = "1.0" }, .flags = A|F },
65 { "precision", "select mathematical precision",
66 OFFSET(precision), AV_OPT_TYPE_INT, { .i64 = PRECISION_FLOAT }, PRECISION_FIXED, PRECISION_DOUBLE, A|F, "precision" },
67 { "fixed", "select 8-bit fixed-point", 0, AV_OPT_TYPE_CONST, { .i64 = PRECISION_FIXED }, INT_MIN, INT_MAX, A|F, "precision" },
68 { "float", "select 32-bit floating-point", 0, AV_OPT_TYPE_CONST, { .i64 = PRECISION_FLOAT }, INT_MIN, INT_MAX, A|F, "precision" },
69 { "double", "select 64-bit floating-point", 0, AV_OPT_TYPE_CONST, { .i64 = PRECISION_DOUBLE }, INT_MIN, INT_MAX, A|F, "precision" },
70 { "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" },
71 { "once", "eval volume expression once", 0, AV_OPT_TYPE_CONST, {.i64=EVAL_MODE_ONCE}, .flags = A|F, .unit = "eval" },
72 { "frame", "eval volume expression per-frame", 0, AV_OPT_TYPE_CONST, {.i64=EVAL_MODE_FRAME}, .flags = A|F, .unit = "eval" },
76 AVFILTER_DEFINE_CLASS(volume);
78 static int set_expr(AVExpr **pexpr, const char *expr, void *log_ctx)
85 ret = av_expr_parse(pexpr, expr, var_names,
86 NULL, NULL, NULL, NULL, 0, log_ctx);
88 av_log(log_ctx, AV_LOG_ERROR,
89 "Error when evaluating the volume expression '%s'\n", expr);
98 static av_cold int init(AVFilterContext *ctx)
100 VolumeContext *vol = ctx->priv;
101 return set_expr(&vol->volume_pexpr, vol->volume_expr, ctx);
104 static av_cold void uninit(AVFilterContext *ctx)
106 VolumeContext *vol = ctx->priv;
107 av_expr_free(vol->volume_pexpr);
111 static int query_formats(AVFilterContext *ctx)
113 VolumeContext *vol = ctx->priv;
114 AVFilterFormats *formats = NULL;
115 AVFilterChannelLayouts *layouts;
116 static const enum AVSampleFormat sample_fmts[][7] = {
117 [PRECISION_FIXED] = {
126 [PRECISION_FLOAT] = {
131 [PRECISION_DOUBLE] = {
138 layouts = ff_all_channel_counts();
140 return AVERROR(ENOMEM);
141 ff_set_common_channel_layouts(ctx, layouts);
143 formats = ff_make_format_list(sample_fmts[vol->precision]);
145 return AVERROR(ENOMEM);
146 ff_set_common_formats(ctx, formats);
148 formats = ff_all_samplerates();
150 return AVERROR(ENOMEM);
151 ff_set_common_samplerates(ctx, formats);
156 static inline void scale_samples_u8(uint8_t *dst, const uint8_t *src,
157 int nb_samples, int volume)
160 for (i = 0; i < nb_samples; i++)
161 dst[i] = av_clip_uint8(((((int64_t)src[i] - 128) * volume + 128) >> 8) + 128);
164 static inline void scale_samples_u8_small(uint8_t *dst, const uint8_t *src,
165 int nb_samples, int volume)
168 for (i = 0; i < nb_samples; i++)
169 dst[i] = av_clip_uint8((((src[i] - 128) * volume + 128) >> 8) + 128);
172 static inline void scale_samples_s16(uint8_t *dst, const uint8_t *src,
173 int nb_samples, int volume)
176 int16_t *smp_dst = (int16_t *)dst;
177 const int16_t *smp_src = (const int16_t *)src;
178 for (i = 0; i < nb_samples; i++)
179 smp_dst[i] = av_clip_int16(((int64_t)smp_src[i] * volume + 128) >> 8);
182 static inline void scale_samples_s16_small(uint8_t *dst, const uint8_t *src,
183 int nb_samples, int volume)
186 int16_t *smp_dst = (int16_t *)dst;
187 const int16_t *smp_src = (const int16_t *)src;
188 for (i = 0; i < nb_samples; i++)
189 smp_dst[i] = av_clip_int16((smp_src[i] * volume + 128) >> 8);
192 static inline void scale_samples_s32(uint8_t *dst, const uint8_t *src,
193 int nb_samples, int volume)
196 int32_t *smp_dst = (int32_t *)dst;
197 const int32_t *smp_src = (const int32_t *)src;
198 for (i = 0; i < nb_samples; i++)
199 smp_dst[i] = av_clipl_int32((((int64_t)smp_src[i] * volume + 128) >> 8));
202 static av_cold void volume_init(VolumeContext *vol)
204 vol->samples_align = 1;
206 switch (av_get_packed_sample_fmt(vol->sample_fmt)) {
207 case AV_SAMPLE_FMT_U8:
208 if (vol->volume_i < 0x1000000)
209 vol->scale_samples = scale_samples_u8_small;
211 vol->scale_samples = scale_samples_u8;
213 case AV_SAMPLE_FMT_S16:
214 if (vol->volume_i < 0x10000)
215 vol->scale_samples = scale_samples_s16_small;
217 vol->scale_samples = scale_samples_s16;
219 case AV_SAMPLE_FMT_S32:
220 vol->scale_samples = scale_samples_s32;
222 case AV_SAMPLE_FMT_FLT:
223 avpriv_float_dsp_init(&vol->fdsp, 0);
224 vol->samples_align = 4;
226 case AV_SAMPLE_FMT_DBL:
227 avpriv_float_dsp_init(&vol->fdsp, 0);
228 vol->samples_align = 8;
233 ff_volume_init_x86(vol);
236 static int set_volume(AVFilterContext *ctx)
238 VolumeContext *vol = ctx->priv;
240 vol->volume = av_expr_eval(vol->volume_pexpr, vol->var_values, NULL);
241 if (isnan(vol->volume)) {
242 if (vol->eval_mode == EVAL_MODE_ONCE) {
243 av_log(ctx, AV_LOG_ERROR, "Invalid value NaN for volume\n");
244 return AVERROR(EINVAL);
246 av_log(ctx, AV_LOG_WARNING, "Invalid value NaN for volume, setting to 0\n");
250 vol->var_values[VAR_VOLUME] = vol->volume;
252 av_log(ctx, AV_LOG_VERBOSE, "n:%f t:%f pts:%f precision:%s ",
253 vol->var_values[VAR_N], vol->var_values[VAR_T], vol->var_values[VAR_PTS],
254 precision_str[vol->precision]);
256 if (vol->precision == PRECISION_FIXED) {
257 vol->volume_i = (int)(vol->volume * 256 + 0.5);
258 vol->volume = vol->volume_i / 256.0;
259 av_log(ctx, AV_LOG_VERBOSE, "volume_i:%d/255 ", vol->volume_i);
261 av_log(ctx, AV_LOG_VERBOSE, "volume:%f volume_dB:%f\n",
262 vol->volume, 20.0*log(vol->volume)/M_LN10);
268 static int config_output(AVFilterLink *outlink)
270 AVFilterContext *ctx = outlink->src;
271 VolumeContext *vol = ctx->priv;
272 AVFilterLink *inlink = ctx->inputs[0];
274 vol->sample_fmt = inlink->format;
275 vol->channels = inlink->channels;
276 vol->planes = av_sample_fmt_is_planar(inlink->format) ? vol->channels : 1;
278 vol->var_values[VAR_N] =
279 vol->var_values[VAR_NB_CONSUMED_SAMPLES] =
280 vol->var_values[VAR_NB_SAMPLES] =
281 vol->var_values[VAR_POS] =
282 vol->var_values[VAR_PTS] =
283 vol->var_values[VAR_STARTPTS] =
284 vol->var_values[VAR_STARTT] =
285 vol->var_values[VAR_T] =
286 vol->var_values[VAR_VOLUME] = NAN;
288 vol->var_values[VAR_NB_CHANNELS] = inlink->channels;
289 vol->var_values[VAR_TB] = av_q2d(inlink->time_base);
290 vol->var_values[VAR_SAMPLE_RATE] = inlink->sample_rate;
292 av_log(inlink->src, AV_LOG_VERBOSE, "tb:%f sample_rate:%f nb_channels:%f\n",
293 vol->var_values[VAR_TB],
294 vol->var_values[VAR_SAMPLE_RATE],
295 vol->var_values[VAR_NB_CHANNELS]);
297 return set_volume(ctx);
300 static int process_command(AVFilterContext *ctx, const char *cmd, const char *args,
301 char *res, int res_len, int flags)
303 VolumeContext *vol = ctx->priv;
304 int ret = AVERROR(ENOSYS);
306 if (!strcmp(cmd, "volume")) {
307 if ((ret = set_expr(&vol->volume_pexpr, args, ctx)) < 0)
309 if (vol->eval_mode == EVAL_MODE_ONCE)
316 #define D2TS(d) (isnan(d) ? AV_NOPTS_VALUE : (int64_t)(d))
317 #define TS2D(ts) ((ts) == AV_NOPTS_VALUE ? NAN : (double)(ts))
318 #define TS2T(ts, tb) ((ts) == AV_NOPTS_VALUE ? NAN : (double)(ts)*av_q2d(tb))
320 static int filter_frame(AVFilterLink *inlink, AVFrame *buf)
322 AVFilterContext *ctx = inlink->dst;
323 VolumeContext *vol = inlink->dst->priv;
324 AVFilterLink *outlink = inlink->dst->outputs[0];
325 int nb_samples = buf->nb_samples;
329 if (isnan(vol->var_values[VAR_STARTPTS])) {
330 vol->var_values[VAR_STARTPTS] = TS2D(buf->pts);
331 vol->var_values[VAR_STARTT ] = TS2T(buf->pts, inlink->time_base);
333 vol->var_values[VAR_PTS] = TS2D(buf->pts);
334 vol->var_values[VAR_T ] = TS2T(buf->pts, inlink->time_base);
335 vol->var_values[VAR_N ] = inlink->frame_count;
337 pos = av_frame_get_pkt_pos(buf);
338 vol->var_values[VAR_POS] = pos == -1 ? NAN : pos;
339 if (vol->eval_mode == EVAL_MODE_FRAME)
342 if (vol->volume == 1.0 || vol->volume_i == 256) {
347 /* do volume scaling in-place if input buffer is writable */
348 if (av_frame_is_writable(buf)) {
351 out_buf = ff_get_audio_buffer(inlink, nb_samples);
353 return AVERROR(ENOMEM);
354 av_frame_copy_props(out_buf, buf);
357 if (vol->precision != PRECISION_FIXED || vol->volume_i > 0) {
358 int p, plane_samples;
360 if (av_sample_fmt_is_planar(buf->format))
361 plane_samples = FFALIGN(nb_samples, vol->samples_align);
363 plane_samples = FFALIGN(nb_samples * vol->channels, vol->samples_align);
365 if (vol->precision == PRECISION_FIXED) {
366 for (p = 0; p < vol->planes; p++) {
367 vol->scale_samples(out_buf->extended_data[p],
368 buf->extended_data[p], plane_samples,
371 } else if (av_get_packed_sample_fmt(vol->sample_fmt) == AV_SAMPLE_FMT_FLT) {
372 for (p = 0; p < vol->planes; p++) {
373 vol->fdsp.vector_fmul_scalar((float *)out_buf->extended_data[p],
374 (const float *)buf->extended_data[p],
375 vol->volume, plane_samples);
378 for (p = 0; p < vol->planes; p++) {
379 vol->fdsp.vector_dmul_scalar((double *)out_buf->extended_data[p],
380 (const double *)buf->extended_data[p],
381 vol->volume, plane_samples);
390 vol->var_values[VAR_NB_CONSUMED_SAMPLES] += out_buf->nb_samples;
391 return ff_filter_frame(outlink, out_buf);
394 static const AVFilterPad avfilter_af_volume_inputs[] = {
397 .type = AVMEDIA_TYPE_AUDIO,
398 .filter_frame = filter_frame,
403 static const AVFilterPad avfilter_af_volume_outputs[] = {
406 .type = AVMEDIA_TYPE_AUDIO,
407 .config_props = config_output,
412 AVFilter ff_af_volume = {
414 .description = NULL_IF_CONFIG_SMALL("Change input volume."),
415 .query_formats = query_formats,
416 .priv_size = sizeof(VolumeContext),
417 .priv_class = &volume_class,
420 .inputs = avfilter_af_volume_inputs,
421 .outputs = avfilter_af_volume_outputs,
422 .flags = AVFILTER_FLAG_SUPPORT_TIMELINE_GENERIC,
423 .process_command = process_command,