CODEWITHSUNDEEP
iphonexcodecore-audiovdsp
psybersonic
psybersonic

Reputation: 29

Convert SIint16 to float on iPhone4

I am developing an app which uses simple pitch perception and it runs fine in the Xcode Simulator. The app loads in the iPhone 4 and I can navigate the app but no output is shown . I have started to debug and find that when I convert Sint16 to float I get an overflow error. I have tried vDSP and a simple loop.

simplified code n=1536

sampleBuffer = (SInt16*) malloc(n);
floatSamples = (float*) malloc(sizeof(float)*n);
    // Convert SInt 16 to float
for(int i = 0; i<n; i++) {
    floatSamples[i] = (float)samples[i];

}

//vDSP_vflt16(samples,1,floatSamples,1,n);

This results in 

-0.000000
 -0.000000
 -0.000000
 964957789008661674961361960960.000000
 -5629971079108504200413184.000000
 -inf
 35116851871418647838720.000000
 -inf
 0.000000
 0.000000
 0.000000
 -1233.760620
 288987875649316726325339192557568.000000
 -0.000000
 -0.000000
 -0.000000
 -7508525217561044282816045485425426432.000000
 -656399043038732589927376093184.000000
 0.000000
 -0.000053
 9470091451011432448.000000
 -24555002

similar result from vDSP
Everything is fine in the simulator on iPhone 4s all the vDSP calculations work.
iPhone is running ios7.1.2 and xCode 6 on MacBook pro.
Just looking for a clue really.

Upvotes: 0

Views: 117

Answers (1)

zaph
zaph

Reputation: 112857

The code
sampleBuffer = (SInt16*) malloc(n);
allocates n bytes, not n SInt16 values.

You need:
sampleBuffer = (SInt16*) malloc(sizeof(SInt16)*n);

Upvotes: 1

Related Questions