How to fill in fft buffer for linear phase FIR response calculation?

General JUCE discussion
1

Hi,
I’m using the inverse FFT to create an impulse response for a FIR filter.
Given a frequency response I need to fill in the fft buffer. I want a linear phase response. I’m not sure if I filled in the phase correctly because I do not get a symmetrical impulse response. And to my knowledge a linear phase filter should have a symmetrical impulse response. Right?

This is the code to fill in the fft input buffer:

		float *fftBuf = new float[fftSize*2];
		float *bPtr = fftBuf;
		// set DC bin
		*bPtr++ = *freqResponse++; // cos
		*bPtr++ = 0.0f; // sin
		// fill the remaining bins
		int numBins = fftSize/2;
		for (int b=1 ; b<numBins ; b++) {
			// linear phase (not sure how exactly)
			float phase = (float)b * MathConstants<float>::twoPi / (float)(numBins);
			// get amplitude
			float amp = *freqResponse++;
			// convert so sin and cos
			*bPtr++ = amp * cosf(phase);
			*bPtr++ = amp * sinf(phase);
		}
		// the nyquist bin should also be 0?
		*bPtr++ = 0.0f;
		*bPtr++ = 0.0f;