1 /*
2 * Copyright (c) 2012 Nicolas George
3 *
4 * This file is part of FFmpeg.
5 *
6 * FFmpeg is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU Lesser General Public License
8 * as published by the Free Software Foundation; either
9 * version 2.1 of the License, or (at your option) any later version.
10 *
11 * FFmpeg is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14 * GNU Lesser General Public License for more details.
15 *
16 * You should have received a copy of the GNU Lesser General Public License
17 * along with FFmpeg; if not, write to the Free Software Foundation, Inc.,
18 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
19 */
20
21 #include "libavutil/channel_layout.h"
22 #include "libavutil/avassert.h"
23 #include "audio.h"
24 #include "avfilter.h"
25 #include "internal.h"
26
27 typedef struct VolDetectContext {
28 /**
29 * Number of samples at each PCM value.
30 * histogram[0x8000 + i] is the number of samples at value i.
31 * The extra element is there for symmetry.
32 */
33 uint64_t histogram[0x10001];
34 } VolDetectContext;
35
filter_frame(AVFilterLink * inlink,AVFrame * samples)36 static int filter_frame(AVFilterLink *inlink, AVFrame *samples)
37 {
38 AVFilterContext *ctx = inlink->dst;
39 VolDetectContext *vd = ctx->priv;
40 int nb_samples = samples->nb_samples;
41 int nb_channels = samples->ch_layout.nb_channels;
42 int nb_planes = nb_channels;
43 int plane, i;
44 int16_t *pcm;
45
46 if (!av_sample_fmt_is_planar(samples->format)) {
47 nb_samples *= nb_channels;
48 nb_planes = 1;
49 }
50 for (plane = 0; plane < nb_planes; plane++) {
51 pcm = (int16_t *)samples->extended_data[plane];
52 for (i = 0; i < nb_samples; i++)
53 vd->histogram[pcm[i] + 0x8000]++;
54 }
55
56 return ff_filter_frame(inlink->dst->outputs[0], samples);
57 }
58
59 #define MAX_DB 91
60
logdb(uint64_t v)61 static inline double logdb(uint64_t v)
62 {
63 double d = v / (double)(0x8000 * 0x8000);
64 if (!v)
65 return MAX_DB;
66 return -log10(d) * 10;
67 }
68
print_stats(AVFilterContext * ctx)69 static void print_stats(AVFilterContext *ctx)
70 {
71 VolDetectContext *vd = ctx->priv;
72 int i, max_volume, shift;
73 uint64_t nb_samples = 0, power = 0, nb_samples_shift = 0, sum = 0;
74 uint64_t histdb[MAX_DB + 1] = { 0 };
75
76 for (i = 0; i < 0x10000; i++)
77 nb_samples += vd->histogram[i];
78 av_log(ctx, AV_LOG_INFO, "n_samples: %"PRId64"\n", nb_samples);
79 if (!nb_samples)
80 return;
81
82 /* If nb_samples > 1<<34, there is a risk of overflow in the
83 multiplication or the sum: shift all histogram values to avoid that.
84 The total number of samples must be recomputed to avoid rounding
85 errors. */
86 shift = av_log2(nb_samples >> 33);
87 for (i = 0; i < 0x10000; i++) {
88 nb_samples_shift += vd->histogram[i] >> shift;
89 power += (i - 0x8000) * (i - 0x8000) * (vd->histogram[i] >> shift);
90 }
91 if (!nb_samples_shift)
92 return;
93 power = (power + nb_samples_shift / 2) / nb_samples_shift;
94 av_assert0(power <= 0x8000 * 0x8000);
95 av_log(ctx, AV_LOG_INFO, "mean_volume: %.1f dB\n", -logdb(power));
96
97 max_volume = 0x8000;
98 while (max_volume > 0 && !vd->histogram[0x8000 + max_volume] &&
99 !vd->histogram[0x8000 - max_volume])
100 max_volume--;
101 av_log(ctx, AV_LOG_INFO, "max_volume: %.1f dB\n", -logdb(max_volume * max_volume));
102
103 for (i = 0; i < 0x10000; i++)
104 histdb[(int)logdb((i - 0x8000) * (i - 0x8000))] += vd->histogram[i];
105 for (i = 0; i <= MAX_DB && !histdb[i]; i++);
106 for (; i <= MAX_DB && sum < nb_samples / 1000; i++) {
107 av_log(ctx, AV_LOG_INFO, "histogram_%ddb: %"PRId64"\n", i, histdb[i]);
108 sum += histdb[i];
109 }
110 }
111
uninit(AVFilterContext * ctx)112 static av_cold void uninit(AVFilterContext *ctx)
113 {
114 print_stats(ctx);
115 }
116
117 static const AVFilterPad volumedetect_inputs[] = {
118 {
119 .name = "default",
120 .type = AVMEDIA_TYPE_AUDIO,
121 .filter_frame = filter_frame,
122 },
123 };
124
125 static const AVFilterPad volumedetect_outputs[] = {
126 {
127 .name = "default",
128 .type = AVMEDIA_TYPE_AUDIO,
129 },
130 };
131
132 const AVFilter ff_af_volumedetect = {
133 .name = "volumedetect",
134 .description = NULL_IF_CONFIG_SMALL("Detect audio volume."),
135 .priv_size = sizeof(VolDetectContext),
136 .uninit = uninit,
137 .flags = AVFILTER_FLAG_METADATA_ONLY,
138 FILTER_INPUTS(volumedetect_inputs),
139 FILTER_OUTPUTS(volumedetect_outputs),
140 FILTER_SAMPLEFMTS(AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_S16P),
141 };
142