/* Make sure this is in line with what we were expecting */
if (!ast_tveq(path->nextin, f->delivery)) {
/* The time has changed between what we expected and this
- most recent time on the new packet. Adjust our output
- time appropriately */
- path->nextout = ast_tvadd(path->nextout,
- ast_tvsub(f->delivery, path->nextin));
+ most recent time on the new packet. If we have a
+ valid prediction adjust our output time appropriately */
+ if (!ast_tvzero(path->nextout)) {
+ path->nextout = ast_tvadd(path->nextout,
+ ast_tvsub(f->delivery, path->nextin));
+ }
path->nextin = f->delivery;
}
} else {
p->next->step->framein(p->next->state, out);
else {
if (!ast_tvzero(delivery)) {
+ /* Regenerate prediction after a discontinuity */
+ if (ast_tvzero(path->nextout))
+ path->nextout = ast_tvnow();
+
/* Use next predicted outgoing timestamp */
out->delivery = path->nextout;
} else {
out->delivery = ast_tv(0, 0);
}
+ /* Invalidate prediction if we're entering a silence period */
+ if (out->frametype == AST_FRAME_CNG)
+ path->nextout = ast_tv(0, 0);
return out;
}
p = p->next;