Further clarify a comment about float precision

1 files changed, 3 insertions, 3 deletions

diff --git a/Alc/ALu.c b/Alc/ALu.c
index 9d7fcdd5..3c36cbc6 100644
--- a/Alc/ALu.c
+++ b/Alc/ALu.c
@@ -1674,9 +1674,9 @@ static inline ALfloat Conv_ALfloat(ALfloat val)
 { return val; }
 static inline ALint Conv_ALint(ALfloat val)
 {
-    /* Floats have a 23-bit mantissa. A bit of the exponent helps out along
-     * with the sign bit, giving 25 bits. So [-16777216, +16777216] is the max
-     * integer range normalized floats can be converted to before losing
+    /* Floats have a 23-bit mantissa. There is an implied 1 bit in the mantissa
+     * along with the sign bit, giving 25 bits total, so [-16777216, +16777216]
+     * is the max value a normalized float can be scaled to before losing
      * precision.
      */
     return fastf2i(clampf(val*16777216.0f, -16777216.0f, 16777215.0f))<<7;