[FFmpeg-devel,v2] avfilter/avf_aphasemeter: Add out-of-phase and mono detection

Submitted by Romane Lafon on May 20, 2019, 12:40 p.m.

Details

Message ID CAMe7PWRA7Mh6zXyjXcSoiehbdLf+S04j+fUtzGCEaOfwjwuwPg@mail.gmail.com
State New
Headers show

Commit Message

Romane Lafon May 20, 2019, 12:40 p.m.
I have updated the patch that extends aphasemeter filter.
It allows to get metadata for out-of-phase or mono sequences of stereo
streams.
It displays start, end and duration as for silencedetect filter.

Comments

Paul B Mahol May 27, 2019, 4:51 p.m.
On 5/20/19, Romane Lafon <romane@nomalab.com> wrote:
> I have updated the patch that extends aphasemeter filter.
> It allows to get metadata for out-of-phase or mono sequences of stereo
> streams.
> It displays start, end and duration as for silencedetect filter.
>

Please use M_PI, and do not invent own constants.
Moritz Barsnick May 27, 2019, 7:06 p.m.
On Mon, May 20, 2019 at 14:40:24 +0200, Romane Lafon wrote:
> +    float tolerance = 1.0 - s->tolerance;

Strictly speaking 1.0f (or 1.f).

> +    float angle = cos(s->angle/180.0*PI);

If you want a float as result, use cosf(s->angle / 180.f * M_PI). (I'm
aware PI and M_PI aren't explicitly marked as float, but you can't have
it all.)

> +        if (!s->is_mono && ((tolerance - fphase) < FLT_EPSILON)) {
[...]
> +        if (s->is_mono && ((tolerance - fphase) < FLT_EPSILON) && s->start_mono_presence) {
[...]
> +        if (s->is_mono && ((tolerance - fphase) > FLT_EPSILON)) {
[...]

As tolerance and fphase are constant throughout this block of code, you
could do this floating point comparison once, and reuse the boolean
result.

BTW, I reckon the third of those comparisons should be ">=", to
properly complement the "<".

> +        if (!s->is_out_phase && (angle - fphase) > FLT_EPSILON) {
[...]
> +        if (s->is_out_phase && ((angle - fphase) > FLT_EPSILON) && s->start_out_phase_presence) {
[...]
> +        if (s->is_out_phase && (angle - fphase) < FLT_EPSILON) {
[...]

Same here.

> +        float tolerance = 1.0 - s->tolerance;
> +        float angle = cos(s->angle/180.0*PI);

Same as above.

Cheers,
Moritz
Romane Lafon June 17, 2019, 1 p.m.
Hello,
Thank you for your feedback.
I made a third version of my patch.
Regards,
Romane

Le lun. 27 mai 2019 à 21:07, Moritz Barsnick <barsnick@gmx.net> a écrit :

> On Mon, May 20, 2019 at 14:40:24 +0200, Romane Lafon wrote:
> > +    float tolerance = 1.0 - s->tolerance;
>
> Strictly speaking 1.0f (or 1.f).
>
> > +    float angle = cos(s->angle/180.0*PI);
>
> If you want a float as result, use cosf(s->angle / 180.f * M_PI). (I'm
> aware PI and M_PI aren't explicitly marked as float, but you can't have
> it all.)
>
> > +        if (!s->is_mono && ((tolerance - fphase) < FLT_EPSILON)) {
> [...]
> > +        if (s->is_mono && ((tolerance - fphase) < FLT_EPSILON) &&
> s->start_mono_presence) {
> [...]
> > +        if (s->is_mono && ((tolerance - fphase) > FLT_EPSILON)) {
> [...]
>
> As tolerance and fphase are constant throughout this block of code, you
> could do this floating point comparison once, and reuse the boolean
> result.
>
> BTW, I reckon the third of those comparisons should be ">=", to
> properly complement the "<".
>
> > +        if (!s->is_out_phase && (angle - fphase) > FLT_EPSILON) {
> [...]
> > +        if (s->is_out_phase && ((angle - fphase) > FLT_EPSILON) &&
> s->start_out_phase_presence) {
> [...]
> > +        if (s->is_out_phase && (angle - fphase) < FLT_EPSILON) {
> [...]
>
> Same here.
>
> > +        float tolerance = 1.0 - s->tolerance;
> > +        float angle = cos(s->angle/180.0*PI);
>
> Same as above.
>
> Cheers,
> Moritz
> _______________________________________________
> ffmpeg-devel mailing list
> ffmpeg-devel@ffmpeg.org
> https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
>
> To unsubscribe, visit link above, or email
> ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

Patch hide | download patch | download mbox

From 08daf8a0394da0d482ee09f29d7cefa14b5c5440 Mon Sep 17 00:00:00 2001
From: Romane Lafon <romane@nomalab.com>
Date: Mon, 20 May 2019 13:28:21 +0200
Subject: [PATCH] avfilter/avf_aphasemeter: Add out of phase and mono detection

Signed-off-by: Romane Lafon <romane@nomalab.com>
---
 libavfilter/avf_aphasemeter.c | 124 ++++++++++++++++++++++++++++++++++++++++--
 1 file changed, 120 insertions(+), 4 deletions(-)

diff --git a/libavfilter/avf_aphasemeter.c b/libavfilter/avf_aphasemeter.c
index ed837059ea..daa15b46a9 100644
--- a/libavfilter/avf_aphasemeter.c
+++ b/libavfilter/avf_aphasemeter.c
@@ -28,26 +28,42 @@ 
 #include "libavutil/intreadwrite.h"
 #include "libavutil/opt.h"
 #include "libavutil/parseutils.h"
+#include "libavutil/timestamp.h"
 #include "avfilter.h"
 #include "formats.h"
 #include "audio.h"
 #include "video.h"
 #include "internal.h"
+#include "stdbool.h"
+#include "float.h"
 
 typedef struct AudioPhaseMeterContext {
     const AVClass *class;
     AVFrame *out;
     int do_video;
+    int do_phasing_detection;
     int w, h;
     AVRational frame_rate;
     int contrast[4];
     uint8_t *mpc_str;
     uint8_t mpc[4];
     int draw_median_phase;
+    int is_mono;
+    int is_out_phase;
+    int start_mono_presence;
+    int start_out_phase_presence;
+    float tolerance;
+    float angle;
+    float phase;
+    float mono_idx[2];
+    float out_phase_idx[2];
+    double duration;
 } AudioPhaseMeterContext;
 
 #define OFFSET(x) offsetof(AudioPhaseMeterContext, x)
 #define FLAGS AV_OPT_FLAG_FILTERING_PARAM|AV_OPT_FLAG_VIDEO_PARAM
+#define PI 3.14159265359
+#define get_duration(index) (index[1] - index[0])
 
 static const AVOption aphasemeter_options[] = {
     { "rate", "set video rate", OFFSET(frame_rate), AV_OPT_TYPE_VIDEO_RATE, {.str="25"}, 0, INT_MAX, FLAGS },
@@ -59,6 +75,10 @@  static const AVOption aphasemeter_options[] = {
     { "bc", "set blue contrast",  OFFSET(contrast[2]), AV_OPT_TYPE_INT, {.i64=1}, 0, 255, FLAGS },
     { "mpc", "set median phase color", OFFSET(mpc_str), AV_OPT_TYPE_STRING, {.str = "none"}, 0, 0, FLAGS },
     { "video", "set video output", OFFSET(do_video), AV_OPT_TYPE_BOOL, {.i64 = 1}, 0, 1, FLAGS },
+    { "phasing", "set mono and out-of-phase detection output", OFFSET(do_phasing_detection), AV_OPT_TYPE_BOOL, {.i64 = 0}, 0, 1, FLAGS },
+    { "tolerance", "set phase tolerance for mono detection", OFFSET(tolerance), AV_OPT_TYPE_FLOAT, {.dbl = 0.}, 0, 1, FLAGS },
+    { "angle", "set angle threshold for out-of-phase detection", OFFSET(angle), AV_OPT_TYPE_FLOAT, {.dbl = 170.}, 90, 180, FLAGS },
+    { "duration", "set minimum mono or out-of-phase duration in seconds", OFFSET(duration), AV_OPT_TYPE_DOUBLE, {.dbl=2.}, 0, 24*60*60, FLAGS },
     { NULL }
 };
 
@@ -140,6 +160,22 @@  static inline int get_x(float phase, int w)
   return (phase + 1.) / 2. * (w - 1);
 }
 
+static inline float get_index(AVFilterLink *inlink, AVFrame *in)
+{
+    char *index_str = av_ts2timestr(in->pts, &inlink->time_base);
+    return atof(index_str);
+}
+
+static inline void add_metadata(AVFrame *insamples, const char *key, float value)
+{
+    char buf[128];
+    char str[128];
+
+    snprintf(str, sizeof(str), "%f", value);
+    snprintf(buf, sizeof(buf), "lavfi.aphasemeter.%s", key);
+    av_dict_set(&insamples->metadata, buf, str, 0);
+}
+
 static int filter_frame(AVFilterLink *inlink, AVFrame *in)
 {
     AVFilterContext *ctx = inlink->dst;
@@ -154,6 +190,8 @@  static int filter_frame(AVFilterLink *inlink, AVFrame *in)
     AVFrame *out;
     uint8_t *dst;
     int i;
+    float tolerance = 1.0 - s->tolerance;
+    float angle = cos(s->angle/180.0*PI);
 
     if (s->do_video && (!s->out || s->out->width  != outlink->w ||
                                    s->out->height != outlink->h)) {
@@ -193,7 +231,7 @@  static int filter_frame(AVFilterLink *inlink, AVFrame *in)
         fphase += phase;
     }
     fphase /= in->nb_samples;
-
+    s->phase = fphase;
     if (s->do_video) {
         if (s->draw_median_phase) {
             dst = out->data[0] + get_x(fphase, s->w) * 4;
@@ -206,10 +244,62 @@  static int filter_frame(AVFilterLink *inlink, AVFrame *in)
 
     metadata = &in->metadata;
     if (metadata) {
-        uint8_t value[128];
+        add_metadata(in, "phase", fphase);
+    }
 
-        snprintf(value, sizeof(value), "%f", fphase);
-        av_dict_set(metadata, "lavfi.aphasemeter.phase", value, 0);
+    if (s->do_phasing_detection) {
+        if (!s->is_mono && ((tolerance - fphase) < FLT_EPSILON)) {
+            s->is_mono = 1;
+            s->start_mono_presence = 1;
+            s->mono_idx[0] = get_index(inlink, in);
+        }
+        if (s->is_mono && ((tolerance - fphase) < FLT_EPSILON) && s->start_mono_presence) {
+            float mono_duration;
+            s->mono_idx[1] = get_index(inlink, in);
+            mono_duration = get_duration(s->mono_idx);
+            if (mono_duration >= s->duration) {
+                add_metadata(in, "mono_start", s->mono_idx[0]);
+                av_log(s, AV_LOG_INFO, "mono_start: %f\n", s->mono_idx[0]);
+                s->start_mono_presence = 0;
+            }
+        }
+        if (s->is_mono && ((tolerance - fphase) > FLT_EPSILON)) {
+            float mono_duration;
+            s->mono_idx[1] = get_index(inlink, in);
+            mono_duration = get_duration(s->mono_idx);
+            if (mono_duration > s->duration) {
+                add_metadata(in, "mono_end", s->mono_idx[1]);
+                add_metadata(in, "mono_duration", mono_duration);
+                av_log(s, AV_LOG_INFO, "mono_end: %f | mono_duration: %f\n", s->mono_idx[1], mono_duration);
+            }
+            s->is_mono = 0;
+        }
+        if (!s->is_out_phase && (angle - fphase) > FLT_EPSILON) {
+            s->out_phase_idx[0] = get_index(inlink, in);
+            s->is_out_phase = 1;
+            s->start_out_phase_presence = 1;
+        }
+        if (s->is_out_phase && ((angle - fphase) > FLT_EPSILON) && s->start_out_phase_presence) {
+            float out_phase_duration;
+            s->out_phase_idx[1] = get_index(inlink, in);
+            out_phase_duration = get_duration(s->out_phase_idx);
+            if (out_phase_duration >= s->duration) {
+                add_metadata(in, "out_phase_start", s->out_phase_idx[0]);
+                av_log(s, AV_LOG_INFO, "out_phase_start: %f\n", s->out_phase_idx[0]);
+                s->start_out_phase_presence = 0;
+            }
+        }
+        if (s->is_out_phase && (angle - fphase) < FLT_EPSILON) {
+            float out_phase_duration;
+            s->out_phase_idx[1] = get_index(inlink, in);
+            out_phase_duration = get_duration(s->out_phase_idx);
+            if (out_phase_duration > s->duration) {
+                add_metadata(in, "out_phase_end", s->out_phase_idx[1]);
+                add_metadata(in, "out_phase_duration", out_phase_duration);
+                av_log(s, AV_LOG_INFO, "out_phase_end: %f | out_phase_duration: %f\n", s->out_phase_idx[1], out_phase_duration);
+            }
+            s->is_out_phase = 0;
+        }
     }
 
     if (s->do_video) {
@@ -224,6 +314,32 @@  static av_cold void uninit(AVFilterContext *ctx)
     AudioPhaseMeterContext *s = ctx->priv;
     int i;
 
+    if (s->do_phasing_detection) {
+        float tolerance = 1.0 - s->tolerance;
+        float angle = cos(s->angle/180.0*PI);
+        float mono_duration;
+        float out_phase_duration;
+        AVFilterLink *inlink = ctx->inputs[0];
+
+        if ((s->is_mono) && ((tolerance - s->phase) < FLT_EPSILON)) {
+            char *index_str = av_ts2timestr(inlink->current_pts, &inlink->time_base);
+            s->mono_idx[1] = atof(index_str);
+            mono_duration = get_duration(s->mono_idx);
+            if (mono_duration > s->duration) {
+                av_log(s, AV_LOG_INFO, "mono_end: %f | mono_duration: %f\n", s->mono_idx[1], mono_duration);
+            }
+            s->is_mono = 0;
+        }
+        if ((s->is_out_phase) && ((angle - s->phase) > FLT_EPSILON)) {
+            char *index_str = av_ts2timestr(inlink->current_pts, &inlink->time_base);
+            s->out_phase_idx[1] = atof(index_str);
+            out_phase_duration = get_duration(s->out_phase_idx);
+            if (out_phase_duration > s->duration) {
+                av_log(s, AV_LOG_INFO, "out_phase_end: %f | out_phase_duration: %f\n", s->out_phase_idx[1], out_phase_duration);
+            }
+            s->is_out_phase = 0;
+        }
+    }
     av_frame_free(&s->out);
     for (i = 0; i < ctx->nb_outputs; i++)
         av_freep(&ctx->output_pads[i].name);
-- 
2.11.0