I think I see whats happening. Its performing an FFT, which results in a linear frequency result. Most spectrum analyzers show log frequency not linear. The result is something that doesn't look right. Rather than convert to log frequency a mapping function is applied to the amplitudes to try and make it look more like what people are expecting.
I think the best way to go is dump the amplitude mapping and really convert the linear frequency to log frequency. Its not that hard to do. You need to run the FFT with more frequency bands than you have on the display, and average the bands, with more bands averaged together on the top and fewer on the bottom. The code is already there to do it, but its doing a linear averaging (the same number of bands per output bar). If the number of output bars on the display is kept the same you will need to increase the number of frequency bands in the FFT. So I think the best compromise is to decrease the number of output bars (to say 8 or 10) and keep the number of bands about the same. This should give a much more accurate display without taking much more cpu cycles. Of course this is all in C so it has to get compiled into the firmware. It would probably be much faster to debug this in desktop squeezeplay where you can get much faster turn around time. John S. -- JohnSwenson ------------------------------------------------------------------------ JohnSwenson's Profile: http://forums.slimdevices.com/member.php?userid=5974 View this thread: http://forums.slimdevices.com/showthread.php?t=77783 _______________________________________________ Touch mailing list Touch@lists.slimdevices.com http://lists.slimdevices.com/mailman/listinfo/touch
