audio: add a mode to insert silence on severe A/V desync

This is probably a stupid idea, but it can't be denied that this
actually allows playing video without larger desync, even if video is
too slow.

DOCS/man/options.rst

@@ -967,6 +967,11 @@ Audio
     Set the step size of mixer volume changes in percent of the full range
     (default: 3).
 
+``--audiodrop``
+    If video is by more than 500ms behind, insert 500ms of silence, to make up
+    for audio getting ahead. Might help if video decoding is too slow beyond
+    help.
+
 ``--volume-restore-data=<string>``
     Used internally for use by playback resume (e.g. with ``quit_watch_later``).
     Restoring value has to be done carefully, because different AOs as well as

options/options.c

@@ -462,6 +462,8 @@ const m_option_t mp_opts[] = {
                 {"decoder", 2},
                 {"decoder+vo", 3})),
 
+    OPT_FLAG("audiodrop", insert_silence, 0),
+
     OPT_FLAG("untimed", untimed, M_OPT_FIXED),
 
     OPT_STRING("stream-capture", stream_capture, M_OPT_FIXED),

options/options.h

@@ -137,6 +137,7 @@ typedef struct MPOpts {
     float default_max_pts_correction;
     int autosync;
     int frame_dropping;
+    int insert_silence;
     int term_osd;
     int term_osd_bar;
     char *term_osd_bar_chars;

player/audio.c

@@ -384,6 +384,10 @@ void fill_audio_out_buffers(struct MPContext *mpctx, double endpts)
         return; // try again next iteration
     }
 
+    struct mp_audio out_format = {0};
+    ao_get_format(mpctx->ao, &out_format);
+    double play_samplerate = out_format.rate / opts->playback_speed;
+
     // If audio is infinitely fast, somehow try keeping approximate A/V sync.
     if (mpctx->audio_status == STATUS_PLAYING && ao_untimed(mpctx->ao) &&
         mpctx->video_status != STATUS_EOF && mpctx->delay > 0)
@@ -402,6 +406,19 @@ void fill_audio_out_buffers(struct MPContext *mpctx, double endpts)
         playsize = MPMAX(1, playsize + skip); // silence will be prepended
     }
 
+    if (opts->insert_silence) {
+        float S = 0.5;
+        if (!mpctx->paused && mpctx->audio_status == STATUS_PLAYING &&
+            mpctx->last_av_difference - mpctx->insert_silence > S)
+            mpctx->insert_silence += S;
+
+        if (mpctx->insert_silence > 0) {
+            int samples = MPMIN(playsize, play_samplerate * mpctx->insert_silence);
+            mp_audio_buffer_prepend_silence(mpctx->ao_buffer, samples);
+            mpctx->insert_silence -= samples / play_samplerate;
+        }
+    }
+
     int status = AD_OK;
     if (playsize > mp_audio_buffer_samples(mpctx->ao_buffer)) {
         status = audio_decode(d_audio, mpctx->ao_buffer, playsize);
@@ -470,10 +487,6 @@ void fill_audio_out_buffers(struct MPContext *mpctx, double endpts)
     bool partial_fill = false;
     int playflags = 0;
 
-    struct mp_audio out_format = {0};
-    ao_get_format(mpctx->ao, &out_format);
-    double play_samplerate = out_format.rate / opts->playback_speed;
-
     if (endpts != MP_NOPTS_VALUE) {
         double samples = (endpts - written_audio_pts(mpctx) - mpctx->audio_delay)
                          * play_samplerate;

player/core.h

@@ -260,6 +260,8 @@ typedef struct MPContext {
     double delay;
     // AV sync: time until next frame should be shown
     double time_frame;
+    // Optional/additional AV sync compensation if video is too slow.
+    double insert_silence;
     // Set to true some time after a new frame has been shown, and it turns out
     // that this frame was the last one before video ends.
     bool playing_last_frame;