diff --git a/papers/export-tech.md b/papers/export-tech.md
index 506c97b00..c8882b601 100644
--- a/papers/export-tech.md
+++ b/papers/export-tech.md
@@ -109,7 +109,7 @@ hex | description
  f3 | loop (negative offset and count follow... both are 8-bit)
  f4 | UNUSED - call symbol (32-bit index follows; only used internally)
  f5 | call sub-block (32-bit address follows)
- f6 | UNUSED
+ f6 | note off + wait one tick
  f7 | full command (command and data follows)
  f8 | call sub-block (16-bit address follows)
  f9 | return from sub-block
diff --git a/src/engine/cmdStream.cpp b/src/engine/cmdStream.cpp
index f790d6d1a..d9a056177 100644
--- a/src/engine/cmdStream.cpp
+++ b/src/engine/cmdStream.cpp
@@ -210,6 +210,11 @@ bool DivCSPlayer::tick() {
           }
           break;
         }
+        case 0xf6: // note off + wait 1
+          e->dispatchCmd(DivCommand(DIV_CMD_NOTE_OFF,i));
+          chan[i].waitTicks=1;
+          chan[i].lastWaitLen=chan[i].waitTicks;
+          break;
         case 0xf7:
           command=stream.readC();
           break;
diff --git a/src/engine/cmdStream.h b/src/engine/cmdStream.h
index 047bf1c74..6bf7914a2 100644
--- a/src/engine/cmdStream.h
+++ b/src/engine/cmdStream.h
@@ -113,10 +113,13 @@ class DivCSPlayer {
 
 struct DivCSProgress {
   int stage, count, total;
+  int optCurrent, optTotal;
   DivCSProgress():
     stage(0),
     count(0),
-    total(0) {}
+    total(0),
+    optCurrent(0),
+    optTotal(0) {}
 };
 
 struct DivCSOptions {
diff --git a/src/engine/cmdStreamOps.cpp b/src/engine/cmdStreamOps.cpp
index 39c9acb9d..18e2cf157 100644
--- a/src/engine/cmdStreamOps.cpp
+++ b/src/engine/cmdStreamOps.cpp
@@ -836,8 +836,6 @@ struct MatchBenefit {
 
 #define MIN_MATCH_SIZE 32
 
-// TODO:
-// - see if we can optimize even more
 SafeWriter* findSubBlocks(SafeWriter* stream, std::vector<SafeWriter*>& subBlocks, unsigned char* speedDial, DivCSProgress* progress) {
   unsigned char* buf=stream->getFinalBuf();
   size_t matchSize=MIN_MATCH_SIZE;
@@ -869,6 +867,11 @@ SafeWriter* findSubBlocks(SafeWriter* stream, std::vector<SafeWriter*>& subBlock
   logD("%d candidates",(int)matches.size());
   logD("%d origs",(int)origs.size());
 
+  if (progress!=NULL) {
+    if ((int)matches.size()>progress->optTotal) progress->optTotal=matches.size();
+    progress->optCurrent=matches.size();
+  }
+
   // quit if there isn't anything
   if (matches.empty()) return stream;
 
@@ -1446,7 +1449,6 @@ SafeWriter* DivEngine::saveCommand(DivCSProgress* progress, DivCSOptions options
       sortPos++;
     }
 
-
     // condense delays
     for (int h=0; h<chans; h++) {
       unsigned char* buf=chanStream[h]->getFinalBuf();
@@ -1500,13 +1502,34 @@ SafeWriter* DivEngine::saveCommand(DivCSProgress* progress, DivCSOptions options
     }
   }
 
-  // PASS 3: remove nop's
+  // PASS 3: note off + one-tick wait
+  // optimize one-tick gaps sometimes used in songs
+  for (int h=0; h<chans; h++) {
+    unsigned char* buf=chanStream[h]->getFinalBuf();
+    if (chanStream[h]->size()<8) continue;
+    for (size_t i=0; i<chanStream[h]->size()-8; i+=8) {
+      // find note off
+      if (buf[i]==0xb5) {
+        // check for contiguous wait 1
+        if (buf[i+8]==0xfe) {
+          // turn it into 0xf6 (note off + wait 1) and change the next one to nop
+          buf[i]=0xf6;
+          buf[i+8]=0xf1;
+
+          // skip the next instruction
+          i+=8;
+        }
+      }
+    }
+  }
+
+  // PASS 4: remove nop's
   // this includes modifying call addresses to compensate
   for (int h=0; h<chans; h++) {
     chanStream[h]=stripNops(chanStream[h]);
   }
 
-  // PASS 4: put all channels together
+  // PASS 5: put all channels together
   for (int i=0; i<chans; i++) {
     chanStreamOff[i]=globalStream->tell();
     logI("- %d: off %x size %ld",i,chanStreamOff[i],chanStream[i]->size());
@@ -1516,7 +1539,7 @@ SafeWriter* DivEngine::saveCommand(DivCSProgress* progress, DivCSOptions options
     delete chanStream[i];
   }
 
-  // PASS 5: find sub-blocks and isolate them
+  // PASS 6: find sub-blocks and isolate them
   if (!options.noSubBlock) {
     std::vector<SafeWriter*> subBlocks;
     size_t beforeSize=globalStream->size();
@@ -1579,13 +1602,13 @@ SafeWriter* DivEngine::saveCommand(DivCSProgress* progress, DivCSOptions options
     assert(!(globalStream->size()&7));
   }
 
-  // PASS 6: pack stream
+  // PASS 7: pack stream
   globalStream=packStream(globalStream,sortedCmd);
 
-  // PASS 7: remove nop's which may be produced by 32-bit call conversion
+  // PASS 8: remove nop's which may be produced by 32-bit call conversion
   globalStream=stripNopsPacked(globalStream,sortedCmd);
 
-  // PASS 8: find new offsets
+  // PASS 9: find new offsets
   {
     unsigned char* buf=globalStream->getFinalBuf();
     for (size_t i=0; i<globalStream->size();) {
diff --git a/src/gui/csPlayer.cpp b/src/gui/csPlayer.cpp
index 18922b079..66a17702a 100644
--- a/src/gui/csPlayer.cpp
+++ b/src/gui/csPlayer.cpp
@@ -142,6 +142,9 @@ String disasmCmd(unsigned char* buf, size_t bufLen, unsigned int addr, unsigned
       if (addr+4>=bufLen) return "???";
       return fmt::sprintf("call $%.8x",(unsigned int)(buf[addr+1]|(buf[addr+2]<<8)|(buf[addr+3]<<16)|(buf[addr+4]<<24)));
       break;
+    case 0xf6:
+      return "offwait";
+      break;
     case 0xf7: {
       if (addr+1>=bufLen) return "???";
       int cmdLen=DivCS::getCmdLength(buf[addr+1]);
diff --git a/src/gui/gui.cpp b/src/gui/gui.cpp
index e070c3f33..76c044735 100644
--- a/src/gui/gui.cpp
+++ b/src/gui/gui.cpp
@@ -6064,7 +6064,8 @@ bool FurnaceGUI::loop() {
         }
       } else {
         WAKE_UP;
-        ImGui::Text("Exporting... %d",csProgress.count);
+        ImGui::Text("Exporting...");
+        ImGui::Text("%d/%d",csProgress.optCurrent,csProgress.optTotal);
 
         // check whether we're done
         if (csExportDone) {