diff --git a/README.html b/README.html
index 8cdb998..281145e 100644
--- a/README.html
+++ b/README.html
@@ -747,6 +747,7 @@ submitted bugfixes since SoundTouch v1.3.1: </p>
   <li> Sandro Cumerlato</li>
   <li> Justin Frankel </li>
   <li> Jason Garland </li>
+  <li> Masa H. </li>
   <li> Takashi Iwai </li>
   <li> Mathias Möhl</li>
   <li> Yuval Naveh </li>
diff --git a/source/SoundTouch/TDStretch.cpp b/source/SoundTouch/TDStretch.cpp
index 3f215dc..8ea11d7 100644
--- a/source/SoundTouch/TDStretch.cpp
+++ b/source/SoundTouch/TDStretch.cpp
@@ -742,12 +742,12 @@ double TDStretch::calcCrossCorr(const short *mixingPos, const short *compare) co
     for (i = 0; i < channels * overlapLength; i += 4) 
     {
         corr += (mixingPos[i] * compare[i] + 
-                 mixingPos[i + 1] * compare[i + 1] +
-                 mixingPos[i + 2] * compare[i + 2] + 
+                 mixingPos[i + 1] * compare[i + 1]) >> overlapDividerBits;  // notice: do intermediate division here to avoid integer overflow
+        corr += (mixingPos[i + 2] * compare[i + 2] + 
                  mixingPos[i + 3] * compare[i + 3]) >> overlapDividerBits;
         norm += (mixingPos[i] * mixingPos[i] + 
-                 mixingPos[i + 1] * mixingPos[i + 1] +
-                 mixingPos[i + 2] * mixingPos[i + 2] + 
+                 mixingPos[i + 1] * mixingPos[i + 1]) >> overlapDividerBits; // notice: do intermediate division here to avoid integer overflow
+        norm += (mixingPos[i + 2] * mixingPos[i + 2] + 
                  mixingPos[i + 3] * mixingPos[i + 3]) >> overlapDividerBits;
     }
 
diff --git a/source/SoundTouch/mmx_optimized.cpp b/source/SoundTouch/mmx_optimized.cpp
index a201b14..03d1805 100644
--- a/source/SoundTouch/mmx_optimized.cpp
+++ b/source/SoundTouch/mmx_optimized.cpp
@@ -93,19 +93,19 @@ double TDStretchMMX::calcCrossCorr(const short *pV1, const short *pV2) const
         // _mm_add_pi32 : 2*32bit add
         // _m_psrad     : 32bit right-shift
 
-        temp = _mm_add_pi32(_mm_madd_pi16(pVec1[0], pVec2[0]),
-                            _mm_madd_pi16(pVec1[1], pVec2[1]));
-        temp2 = _mm_add_pi32(_mm_madd_pi16(pVec1[0], pVec1[0]),
-                             _mm_madd_pi16(pVec1[1], pVec1[1]));
-        accu = _mm_add_pi32(accu, _mm_sra_pi32(temp, shifter));
-        normaccu = _mm_add_pi32(normaccu, _mm_sra_pi32(temp2, shifter));
+        temp = _mm_add_pi32(_mm_sra_pi32(_mm_madd_pi16(pVec1[0], pVec2[0]), shifter),
+                            _mm_sra_pi32(_mm_madd_pi16(pVec1[1], pVec2[1]), shifter));
+        temp2 = _mm_add_pi32(_mm_sra_pi32(_mm_madd_pi16(pVec1[0], pVec1[0]), shifter),
+                            _mm_sra_pi32(_mm_madd_pi16(pVec1[1], pVec1[1]), shifter));
+        accu = _mm_add_pi32(accu, temp);
+        normaccu = _mm_add_pi32(normaccu, temp2);
 
-        temp = _mm_add_pi32(_mm_madd_pi16(pVec1[2], pVec2[2]),
-                            _mm_madd_pi16(pVec1[3], pVec2[3]));
-        temp2 = _mm_add_pi32(_mm_madd_pi16(pVec1[2], pVec1[2]),
-                             _mm_madd_pi16(pVec1[3], pVec1[3]));
-        accu = _mm_add_pi32(accu, _mm_sra_pi32(temp, shifter));
-        normaccu = _mm_add_pi32(normaccu, _mm_sra_pi32(temp2, shifter));
+        temp = _mm_add_pi32(_mm_sra_pi32(_mm_madd_pi16(pVec1[2], pVec2[2]), shifter),
+                            _mm_sra_pi32(_mm_madd_pi16(pVec1[3], pVec2[3]), shifter));
+        temp2 = _mm_add_pi32(_mm_sra_pi32(_mm_madd_pi16(pVec1[2], pVec1[2]), shifter),
+                            _mm_sra_pi32(_mm_madd_pi16(pVec1[3], pVec1[3]), shifter));
+        accu = _mm_add_pi32(accu, temp);
+        normaccu = _mm_add_pi32(normaccu, temp2);
 
         pVec1 += 4;
         pVec2 += 4;