Casting too large a value to java short

Question

I have a java program that works with audio data.

I have the following code that takes an array of float between -1 and 1 and casts to short. I have realized that this code could maybe be broken if values outside the range of -1 to 1 are provided.

out = (short) ((data[i])*32768);

In the following code if data[i] == 2.0 , it will be too large for a short to hold. In this case would the cast just take the first 16 bits, whatever they may be, and treat it like a short?

Answer 1

(data[i])*32768 will result smallest type that can hold (e.g. integer here)

Then this cast: out = (short) ((data[i])*32768) will keep least significant bits of short (16 bits here, included sign bit - most significant bit aka left most bit)

You can see that if MSB is 1. The casted value will be nagative, else value with positive.

    int i = Short.MAX_VALUE;
    short b = (short) i;
    System.out.println(i +" -> "+ b);
    System.out.println("" + Integer.toBinaryString(0xFFFF & b));

    i = Short.MAX_VALUE + 1;
    b = (short) i;
    System.out.println(i +" -> "+ b);
    System.out.println("" + Integer.toBinaryString(0xFFFF & b));

    i = Short.MAX_VALUE + Short.MAX_VALUE + Short.MAX_VALUE;
    b = (short) i;
    System.out.println(i +" -> "+ b);
    System.out.println("" + Integer.toBinaryString(0xFFFF & b));

Will print

32767 -> 32767
111111111111111
32768 -> -32768
1000000000000000
98301 -> 32765
111111111111101

See 2nd test, cast value has 1 in MSB