This repository makes fft.c from fftpack user-friendly. FFTPACK is a very high-performance fft tool even when compared to Apple's vDSP and many other libraries.

Original code at http://www.netlib.org/fftpack/

NOTE:The operation is performed in-place

• CCMATH: http://freshmeat.sourceforge.net/projects/ccmath

make example
./example

0. You can edit FFT_PRECISION in fft.h to use double or float
1. Create and manage your input signal, including memory
2. Initialize fourier transformer

// n: length of the signal
// scale_output: FFT_SCALED_OUTPUT or FFT_UNSCALED_OUTPUT
FFTTransformer* transformer = create_fft_transformer(n, FFT_SCALED_OUTPUT);

3. Transform signal

fft_forward(transformer, input);

4. Your output is here

for(int i = 0; i < transformer -> n; i+=2){
   double f = i / 2 * sample_rate / n;
   double cos_comp = input[i];
   double sin_comp = input[i+1];
   double frequency_magnitude = sqrt(cos_comp*coscomp + sin_comp*sin_comp);
}

5. After all transformation is done, release memory

free_fft_transformer(transformer);

1. FFTTransformer * create_fft_transformer(int signal_length, int scale_output);

2. void free_fft_transformer(FFTTransformer * transformer);

3. void fft_forward(FFTTransformer * transformer, double* input);

4. void fft_backward(FFTTransformer * transformer, double* input);

5. FFTCosqTransformer * create_fft_cosq_transformer(int signal_length, int scale_output);

6. void free_cosq_fft_transformer(FFTCosqTransformer * transformer);

7. void fft_cosq_forward(FFTCosqTransformer * transformer, double* input);

8. void fft_cosq_backward(FFTCosqTransformer * transformer, double* input);