X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=libavfilter%2Faf_astats.c;h=e38d66b008404d0b31994c0b3447aebb8e443bff;hb=694f9ab1ef141cf80de8f81db6955a17606182af;hp=a89ac6d8f58f5b182a71fecbbeac5fbb29d1e1e9;hpb=a137e50ee54cd902c436d15463c6e621623cc9ad;p=ffmpeg

diff --git a/libavfilter/af_astats.c b/libavfilter/af_astats.c
index a89ac6d8f58..e38d66b0084 100644
--- a/libavfilter/af_astats.c
+++ b/libavfilter/af_astats.c
@@ -33,6 +33,9 @@ typedef struct ChannelStats {
     double min, max;
     double min_run, max_run;
     double min_runs, max_runs;
+    double min_diff, max_diff;
+    double diff1_sum;
+    uint64_t mask;
     uint64_t min_count, max_count;
     uint64_t nb_samples;
 } ChannelStats;
@@ -102,6 +105,7 @@ static void reset_stats(AudioStatsContext *s)
 
         p->min = p->min_sigma_x2 = DBL_MAX;
         p->max = p->max_sigma_x2 = DBL_MIN;
+        p->min_diff = p->max_diff = -1;
     }
 }
 
@@ -121,6 +125,15 @@ static int config_output(AVFilterLink *outlink)
     return 0;
 }
 
+static unsigned bit_depth(uint64_t mask)
+{
+    unsigned result = 64;
+
+    for (; result && !(mask & 1); --result, mask >>= 1);
+
+    return result;
+}
+
 static inline void update_stat(AudioStatsContext *s, ChannelStats *p, double d)
 {
     if (d < p->min) {
@@ -150,7 +163,11 @@ static inline void update_stat(AudioStatsContext *s, ChannelStats *p, double d)
     p->sigma_x += d;
     p->sigma_x2 += d * d;
     p->avg_sigma_x2 = p->avg_sigma_x2 * s->mult + (1.0 - s->mult) * d * d;
+    p->min_diff = FFMIN(p->min_diff == -1 ? DBL_MAX : p->min_diff, FFABS(d - (p->min_diff == -1 ? DBL_MAX : p->last)));
+    p->max_diff = FFMAX(p->max_diff, FFABS(d - (p->max_diff == -1 ? d : p->last)));
+    p->diff1_sum += FFABS(d - p->last);
     p->last = d;
+    p->mask |= llrint(d * (1LLU<<63));
 
     if (p->nb_samples >= s->tc_samples) {
         p->max_sigma_x2 = FFMAX(p->max_sigma_x2, p->avg_sigma_x2);
@@ -177,10 +194,11 @@ static void set_meta(AVDictionary **metadata, int chan, const char *key,
 
 static void set_metadata(AudioStatsContext *s, AVDictionary **metadata)
 {
-    uint64_t min_count = 0, max_count = 0, nb_samples = 0;
+    uint64_t mask = 0, min_count = 0, max_count = 0, nb_samples = 0;
     double min_runs = 0, max_runs = 0,
-           min = DBL_MAX, max = DBL_MIN,
+           min = DBL_MAX, max = DBL_MIN, min_diff = DBL_MAX, max_diff = 0,
            max_sigma_x = 0,
+           diff1_sum = 0,
            sigma_x = 0,
            sigma_x2 = 0,
            min_sigma_x2 = DBL_MAX,
@@ -195,6 +213,9 @@ static void set_metadata(AudioStatsContext *s, AVDictionary **metadata)
 
         min = FFMIN(min, p->min);
         max = FFMAX(max, p->max);
+        min_diff = FFMIN(min_diff, p->min_diff);
+        max_diff = FFMAX(max_diff, p->max_diff);
+        diff1_sum += p->diff1_sum,
         min_sigma_x2 = FFMIN(min_sigma_x2, p->min_sigma_x2);
         max_sigma_x2 = FFMAX(max_sigma_x2, p->max_sigma_x2);
         sigma_x += p->sigma_x;
@@ -203,6 +224,7 @@ static void set_metadata(AudioStatsContext *s, AVDictionary **metadata)
         max_count += p->max_count;
         min_runs += p->min_runs;
         max_runs += p->max_runs;
+        mask |= p->mask;
         nb_samples += p->nb_samples;
         if (fabs(p->sigma_x) > fabs(max_sigma_x))
             max_sigma_x = p->sigma_x;
@@ -210,6 +232,9 @@ static void set_metadata(AudioStatsContext *s, AVDictionary **metadata)
         set_meta(metadata, c + 1, "DC_offset", "%f", p->sigma_x / p->nb_samples);
         set_meta(metadata, c + 1, "Min_level", "%f", p->min);
         set_meta(metadata, c + 1, "Max_level", "%f", p->max);
+        set_meta(metadata, c + 1, "Min_difference", "%f", p->min_diff);
+        set_meta(metadata, c + 1, "Max_difference", "%f", p->max_diff);
+        set_meta(metadata, c + 1, "Mean_difference", "%f", p->diff1_sum / (p->nb_samples - 1));
         set_meta(metadata, c + 1, "Peak_level", "%f", LINEAR_TO_DB(FFMAX(-p->min, p->max)));
         set_meta(metadata, c + 1, "RMS_level", "%f", LINEAR_TO_DB(sqrt(p->sigma_x2 / p->nb_samples)));
         set_meta(metadata, c + 1, "RMS_peak", "%f", LINEAR_TO_DB(sqrt(p->max_sigma_x2)));
@@ -217,17 +242,22 @@ static void set_metadata(AudioStatsContext *s, AVDictionary **metadata)
         set_meta(metadata, c + 1, "Crest_factor", "%f", p->sigma_x2 ? FFMAX(-p->min, p->max) / sqrt(p->sigma_x2 / p->nb_samples) : 1);
         set_meta(metadata, c + 1, "Flat_factor", "%f", LINEAR_TO_DB((p->min_runs + p->max_runs) / (p->min_count + p->max_count)));
         set_meta(metadata, c + 1, "Peak_count", "%f", (float)(p->min_count + p->max_count));
+        set_meta(metadata, c + 1, "Bit_depth", "%f", bit_depth(p->mask));
     }
 
     set_meta(metadata, 0, "Overall.DC_offset", "%f", max_sigma_x / (nb_samples / s->nb_channels));
     set_meta(metadata, 0, "Overall.Min_level", "%f", min);
     set_meta(metadata, 0, "Overall.Max_level", "%f", max);
+    set_meta(metadata, 0, "Overall.Min_difference", "%f", min_diff);
+    set_meta(metadata, 0, "Overall.Max_difference", "%f", max_diff);
+    set_meta(metadata, 0, "Overall.Mean_difference", "%f", diff1_sum / (nb_samples - s->nb_channels));
     set_meta(metadata, 0, "Overall.Peak_level", "%f", LINEAR_TO_DB(FFMAX(-min, max)));
     set_meta(metadata, 0, "Overall.RMS_level", "%f", LINEAR_TO_DB(sqrt(sigma_x2 / nb_samples)));
     set_meta(metadata, 0, "Overall.RMS_peak", "%f", LINEAR_TO_DB(sqrt(max_sigma_x2)));
     set_meta(metadata, 0, "Overall.RMS_trough", "%f", LINEAR_TO_DB(sqrt(min_sigma_x2)));
     set_meta(metadata, 0, "Overall.Flat_factor", "%f", LINEAR_TO_DB((min_runs + max_runs) / (min_count + max_count)));
     set_meta(metadata, 0, "Overall.Peak_count", "%f", (float)(min_count + max_count) / (double)s->nb_channels);
+    set_meta(metadata, 0, "Overall.Bit_depth", "%f", bit_depth(mask));
     set_meta(metadata, 0, "Overall.Number_of_samples", "%f", nb_samples / s->nb_channels);
 }
 
@@ -276,10 +306,11 @@ static int filter_frame(AVFilterLink *inlink, AVFrame *buf)
 static void print_stats(AVFilterContext *ctx)
 {
     AudioStatsContext *s = ctx->priv;
-    uint64_t min_count = 0, max_count = 0, nb_samples = 0;
+    uint64_t mask = 0, min_count = 0, max_count = 0, nb_samples = 0;
     double min_runs = 0, max_runs = 0,
-           min = DBL_MAX, max = DBL_MIN,
+           min = DBL_MAX, max = DBL_MIN, min_diff = DBL_MAX, max_diff = 0,
            max_sigma_x = 0,
+           diff1_sum = 0,
            sigma_x = 0,
            sigma_x2 = 0,
            min_sigma_x2 = DBL_MAX,
@@ -294,6 +325,9 @@ static void print_stats(AVFilterContext *ctx)
 
         min = FFMIN(min, p->min);
         max = FFMAX(max, p->max);
+        min_diff = FFMIN(min_diff, p->min_diff);
+        max_diff = FFMAX(max_diff, p->max_diff);
+        diff1_sum += p->diff1_sum,
         min_sigma_x2 = FFMIN(min_sigma_x2, p->min_sigma_x2);
         max_sigma_x2 = FFMAX(max_sigma_x2, p->max_sigma_x2);
         sigma_x += p->sigma_x;
@@ -302,6 +336,7 @@ static void print_stats(AVFilterContext *ctx)
         max_count += p->max_count;
         min_runs += p->min_runs;
         max_runs += p->max_runs;
+        mask |= p->mask;
         nb_samples += p->nb_samples;
         if (fabs(p->sigma_x) > fabs(max_sigma_x))
             max_sigma_x = p->sigma_x;
@@ -310,6 +345,9 @@ static void print_stats(AVFilterContext *ctx)
         av_log(ctx, AV_LOG_INFO, "DC offset: %f\n", p->sigma_x / p->nb_samples);
         av_log(ctx, AV_LOG_INFO, "Min level: %f\n", p->min);
         av_log(ctx, AV_LOG_INFO, "Max level: %f\n", p->max);
+        av_log(ctx, AV_LOG_INFO, "Min difference: %f\n", p->min_diff);
+        av_log(ctx, AV_LOG_INFO, "Max difference: %f\n", p->max_diff);
+        av_log(ctx, AV_LOG_INFO, "Mean difference: %f\n", p->diff1_sum / (p->nb_samples - 1));
         av_log(ctx, AV_LOG_INFO, "Peak level dB: %f\n", LINEAR_TO_DB(FFMAX(-p->min, p->max)));
         av_log(ctx, AV_LOG_INFO, "RMS level dB: %f\n", LINEAR_TO_DB(sqrt(p->sigma_x2 / p->nb_samples)));
         av_log(ctx, AV_LOG_INFO, "RMS peak dB: %f\n", LINEAR_TO_DB(sqrt(p->max_sigma_x2)));
@@ -318,12 +356,16 @@ static void print_stats(AVFilterContext *ctx)
         av_log(ctx, AV_LOG_INFO, "Crest factor: %f\n", p->sigma_x2 ? FFMAX(-p->min, p->max) / sqrt(p->sigma_x2 / p->nb_samples) : 1);
         av_log(ctx, AV_LOG_INFO, "Flat factor: %f\n", LINEAR_TO_DB((p->min_runs + p->max_runs) / (p->min_count + p->max_count)));
         av_log(ctx, AV_LOG_INFO, "Peak count: %"PRId64"\n", p->min_count + p->max_count);
+        av_log(ctx, AV_LOG_INFO, "Bit depth: %u\n", bit_depth(p->mask));
     }
 
     av_log(ctx, AV_LOG_INFO, "Overall\n");
     av_log(ctx, AV_LOG_INFO, "DC offset: %f\n", max_sigma_x / (nb_samples / s->nb_channels));
     av_log(ctx, AV_LOG_INFO, "Min level: %f\n", min);
     av_log(ctx, AV_LOG_INFO, "Max level: %f\n", max);
+    av_log(ctx, AV_LOG_INFO, "Min difference: %f\n", min_diff);
+    av_log(ctx, AV_LOG_INFO, "Max difference: %f\n", max_diff);
+    av_log(ctx, AV_LOG_INFO, "Mean difference: %f\n", diff1_sum / (nb_samples - s->nb_channels));
     av_log(ctx, AV_LOG_INFO, "Peak level dB: %f\n", LINEAR_TO_DB(FFMAX(-min, max)));
     av_log(ctx, AV_LOG_INFO, "RMS level dB: %f\n", LINEAR_TO_DB(sqrt(sigma_x2 / nb_samples)));
     av_log(ctx, AV_LOG_INFO, "RMS peak dB: %f\n", LINEAR_TO_DB(sqrt(max_sigma_x2)));
@@ -331,6 +373,7 @@ static void print_stats(AVFilterContext *ctx)
         av_log(ctx, AV_LOG_INFO, "RMS trough dB: %f\n", LINEAR_TO_DB(sqrt(min_sigma_x2)));
     av_log(ctx, AV_LOG_INFO, "Flat factor: %f\n", LINEAR_TO_DB((min_runs + max_runs) / (min_count + max_count)));
     av_log(ctx, AV_LOG_INFO, "Peak count: %f\n", (min_count + max_count) / (double)s->nb_channels);
+    av_log(ctx, AV_LOG_INFO, "Bit depth: %u\n", bit_depth(mask));
     av_log(ctx, AV_LOG_INFO, "Number of samples: %"PRId64"\n", nb_samples / s->nb_channels);
 }