Silence warnings on _mm_storel_epi64()

No alignment requirement despite the __m128i* argument

Author: jmvalin

celt/x86/celt_lpc_sse4_1.c

@@ -80,7 +80,7 @@ void celt_fir_sse4_1(const opus_val16 *x,
       vecX = OP_CVTEPI16_EPI32_M64(x + i);
       vecSum = _mm_add_epi32(vecSum, vecX);
       vecSum = _mm_packs_epi32(vecSum, vecSum);
-      _mm_storel_epi64((__m128i *)(y + i), vecSum);
+      _mm_storel_epi64((__m128i *)(void *)(y + i), vecSum);
    }
    for (;i<N;i++)
    {

silk/fixed/x86/vector_ops_FIX_sse4_1.c

@@ -76,7 +76,7 @@ opus_int64 silk_inner_prod16_sse4_1(
     acc2 = _mm_shuffle_epi32( acc1, _MM_SHUFFLE( 0, 0, 3, 2 ) );
     acc1 = _mm_add_epi64( acc1, acc2 );
 
-    _mm_storel_epi64( (__m128i *)&sum, acc1 );
+    _mm_storel_epi64( (__m128i *)(void *)&sum, acc1 );
 
     for( ; i < len; i++ ) {
         sum = silk_SMLALBB( sum, inVec1[ i ], inVec2[ i ] );