I'm analyzing audio at the rate of 400 sample frames which has a 200-sample overlap at 16 KHz. I would be grateful if you could provide me a setup on how to do that?

You need to resample the incoming audio in the time domain first to 16 kHz and then use our frequency domain class. You can get a 2:1 overlap ratio with the advance() method, and setting up the fft size to 512 can get you near 400 (fft can be power of two only). But really, this is too complex and reduces audio quality without any significant performance benefit over the Superpowered frequency domain example.