Question

It is a function in vDSP in iOS. The reference said this function

Converts an array of unsigned 16-bit integers to single-precision floating-point values.

But what actually is created? For example, I have a series of 16-bit integers storing phonetic samples. What do I actually get when I call this function?

Answer 1

Nothing is created. You pass in an array of N unsigned 16 bit short ints in the A parameter and an array of N floats in the __vDSP_C parameter and the routine converts the unsigned short int values to floats. E.g. if A[0] = 42 then __vDSP_C[0] will be set to 42.0f.

void vDSP_vfltu16 (
   unsigned short *A,
   vDSP_Stride __vDSP_I,
   float *__vDSP_C,
   vDSP_Stride __vDSP_K,
   vDSP_Length __vDSP_N
);

There is reasonable documentation on developer.apple.com: https://developer.apple.com/library/mac/#documentation/Accelerate/Reference/vDSPRef/Reference/reference.html