tags:

views:

431

answers:

2

Has anybody used the apple FFT for an iPhone app yet or know where I might find a sample application as to how to use it? I know that apple has some sample code posted, but I'm not really sure how to implement it into an actual project.

+4  A: 

Here's a real-world example: A snippet of c++ that uses Accelerate's vDSP fft routines to do autocorrelation on the Remote IO audio unit's input. Using this framework is pretty complicated, but the documentation isn't too bad.

OSStatus DSPCore::initialize (double _sampleRate, uint16_t _bufferSize) {
    sampleRate = _sampleRate;
    bufferSize = _bufferSize;
    peakIndex = 0;
    frequency = 0.f;
    uint32_t maxFrames = getMaxFramesPerSlice();
    displayData = (float*)malloc(maxFrames*sizeof(float));
    bzero(displayData, maxFrames*sizeof(float));
    log2n = log2f(maxFrames);
    n = 1 << log2n;
    assert(n == maxFrames);
    nOver2 = maxFrames/2;
    A.realp = (float*)malloc(nOver2 * sizeof(float));
    A.imagp = (float*)malloc(nOver2 * sizeof(float));
    FFTSetup fftSetup = vDSP_create_fftsetup(log2n, FFT_RADIX2);

    return noErr;
}

void DSPCore::Render(uint32_t numFrames, AudioBufferList *ioData) {

    bufferSize = numFrames;
    float ln = log2f(numFrames);

    //vDSP autocorrelation

    //convert real input to even-odd
    vDSP_ctoz((COMPLEX*)ioData->mBuffers[0].mData, 2, &A, 1, numFrames/2);
    memset(ioData->mBuffers[0].mData, 0, ioData->mBuffers[0].mDataByteSize);
    //fft
    vDSP_fft_zrip(fftSetup, &A, 1, ln, FFT_FORWARD);

    // Absolute square (equivalent to mag^2)
    vDSP_zvmags(&A, 1, A.realp, 1, numFrames/2);
    bzero(A.imagp, (numFrames/2) * sizeof(float));    

    // Inverse FFT
    vDSP_fft_zrip(fftSetup, &A, 1, ln, FFT_INVERSE);

    //convert complex split to real
    vDSP_ztoc(&A, 1, (COMPLEX*)displayData, 2, numFrames/2);

    // Normalize
    float scale = 1.f/displayData[0];
    vDSP_vsmul(displayData, 1, &scale, displayData, 1, numFrames);

    // Naive peak-pick: find the first local maximum
    peakIndex = 0;
    for (size_t ii=1; ii < numFrames-1; ++ii) {
        if ((displayData[ii] > displayData[ii-1]) && (displayData[ii] > displayData[ii+1])) {
            peakIndex = ii;
            break;
        }
    }

    // Calculate frequency
    frequency = sampleRate / peakIndex + quadInterpolate(&displayData[peakIndex-1]);

    bufferSize = numFrames;

    for (int ii=0; ii<ioData->mNumberBuffers; ++ii) {
        bzero(ioData->mBuffers[ii].mData, ioData->mBuffers[ii].mDataByteSize);
    }
}
Art Gillespie
Great example, but can you point me in the direction of implementations for these two functions: getMaxFramesPerSlice() and quadInterpolate() ?
CJ Hanson
Sorry, one more question... since my audio is 16bit lpcm I get back integer data in my buffers, how would I change it efficiently to float for use with the fft code?
CJ Hanson