diff --git a/plugins/spu2-x/src/Mixer.cpp b/plugins/spu2-x/src/Mixer.cpp
index f826027609..8442c36a79 100644
--- a/plugins/spu2-x/src/Mixer.cpp
+++ b/plugins/spu2-x/src/Mixer.cpp
@@ -331,7 +331,7 @@ static void __forceinline UpdatePitch( uint coreidx, uint voiceidx )
 	if( (vc.Modulated==0) || (voiceidx==0) )
 		pitch = vc.Pitch;
 	else
-		pitch = (vc.Pitch*(32768 + abs(Cores[coreidx].Voices[voiceidx-1].OutX)))>>15;
+		pitch = (vc.Pitch*(32768 + Cores[coreidx].Voices[voiceidx-1].OutX))>>15;
 
 	vc.SP+=pitch;
 }
@@ -573,18 +573,28 @@ static __forceinline StereoOut32 MixVoice( uint coreidx, uint voiceidx )
 
 		CalculateADSR( thiscore, voiceidx );
 		Value	= MulShr32( Value, vc.ADSR.Value );
-		vc.OutX	= Value;		// Note: All values recorded into OutX (may be used for modulation later)
-
-		if(voiceidx<23 && Cores[coreidx].Voices[voiceidx+1].Modulated)
-			Value=0;
+		
+		// Store Value for eventual modulation later
+		// Pseudonym's Crest calculation idea. Actually calculates a crest, unlike the old code which was just peak.
+		u32 Amplitude = std::abs(Value);
+		if(Amplitude < vc.NextCrest)
+		{
+			vc.OutX = vc.NextCrest;
+			vc.NextCrest = 0;
+		}
+		if(Amplitude > vc.PrevAmp)
+		{
+			vc.NextCrest = Amplitude;
+		}
+		vc.PrevAmp = Amplitude;
 
 		if( IsDevBuild )
-			DebugCores[coreidx].Voices[voiceidx].displayPeak = std::max(DebugCores[coreidx].Voices[voiceidx].displayPeak,abs(vc.OutX));
+			DebugCores[coreidx].Voices[voiceidx].displayPeak = std::max(DebugCores[coreidx].Voices[voiceidx].displayPeak,(s32)vc.OutX);
 
 		// Write-back of raw voice data (post ADSR applied)
 
-		if (voiceidx==1)      spu2M_WriteFast( ( (0==coreidx) ? 0x400 : 0xc00 ) + OutPos, Value );
-		else if (voiceidx==3) spu2M_WriteFast( ( (0==coreidx) ? 0x600 : 0xe00 ) + OutPos, Value );
+		if (voiceidx==1)      spu2M_WriteFast( ( (0==coreidx) ? 0x400 : 0xc00 ) + OutPos, vc.OutX );
+		else if (voiceidx==3) spu2M_WriteFast( ( (0==coreidx) ? 0x600 : 0xe00 ) + OutPos, vc.OutX );
 
 		return ApplyVolume( StereoOut32( Value, Value ), vc.Volume );
 	}
diff --git a/plugins/spu2-x/src/defs.h b/plugins/spu2-x/src/defs.h
index d5ea70eda1..5705c4e96c 100644
--- a/plugins/spu2-x/src/defs.h
+++ b/plugins/spu2-x/src/defs.h
@@ -143,7 +143,7 @@ struct V_Voice
 // Envelope
 	V_ADSR ADSR;
 // Pitch (also Reg_PITCH)
-	s16 Pitch;
+	u16 Pitch;
 // Loop Start address (also Reg_LSAH/L)
 	u32 LoopStartA;
 // Sound Start address (also Reg_SSAH/L)
@@ -180,7 +180,9 @@ struct V_Voice
 	s32 PV1;
 
 // Last outputted audio value, used for voice modulation.
-	s32 OutX;
+	u32 OutX;
+	u32 NextCrest; // temp value for Crest calculation
+	u32 PrevAmp;  // temp value for Crest calculation (abs of last value)
 
 // SBuffer now points directly to an ADPCM cache entry.
 	s16 *SBuffer;
diff --git a/plugins/spu2-x/src/spu2sys.cpp b/plugins/spu2-x/src/spu2sys.cpp
index be1f97c30d..70af5fadc7 100644
--- a/plugins/spu2-x/src/spu2sys.cpp
+++ b/plugins/spu2-x/src/spu2sys.cpp
@@ -144,7 +144,7 @@ void V_Core::Reset( int index )
 		VoiceGates[v].WetL = -1;
 		VoiceGates[v].WetR = -1;
 
-		Voices[v].Volume		= V_VolumeSlideLR::Max;
+		Voices[v].Volume		= V_VolumeSlideLR(0,0); // V_VolumeSlideLR::Max;
 		Voices[v].SCurrent		= 28;
 
 		Voices[v].ADSR.Value	= 0;
@@ -263,6 +263,8 @@ void V_Voice::Start()
 
 		PV1 = PV2		= 0;
 		PV3 = PV4		= 0;
+		PrevAmp = 0;
+		NextCrest = 0;
 	}
 	else
 	{
@@ -754,7 +756,7 @@ static void __fastcall RegWrite_VoiceParams( u16 value )
 		break;
 
 		case 2:
-			thisvoice.Pitch			= value;
+			thisvoice.Pitch			= value & 0x3fff;
 		break;
 
 		case 3: // ADSR1 (Envelope)