Notification email reminded me to get back to this. Since I've been toggling between Linear and Log modes, I've noticed another anomaly: Sometimes the Peak Finding algorithm finds a different set of peaks in the two modes. Not completely different, just misses different ones in the two modes. Could be they're just so close together that some are being skipped due to my pointing device resolution. A "next peak fwd/bkwd" assignment for the arrow keys might be the solution.DickN wrote:I note that for frequencies above around 5 KHz (for 44,100 SPS), switching to Linear frequency axis somewhat ameliorates these effects.
Irrelevant details, but more appropriate to this forum than to the original:
I'm sifting through air conditioner noise picked up by a wireless mic, identifying which frequencies are important enough to notch out (If I could notch them all out there wouldn't be much audio left from 1500-4000 Hz, and it would sound comb-filtered!). They're not necessarily the highest peaks either - depends on the relative amplitudes of their neighbors. They're all modulation sidebands from the reed valves, and spaced apart by the compressor frequency (58-59 Hz, varies with mains voltage). Isolated peaks can often be left alone because it's the modulation (perceivable only when neighboring sidebands are audible) that makes them so distracting and keeps the ear from just losing interest in them after a while. Extra sidebands occur at 44 Hz spacing around some (whew!) of these, I believe due to modulation of the whole valve "signal" by the fan. Why only some? I don't know, guess I'm just lucky
. I think the fan is reflecting the sound rather than blocking it, and these fan sidebands are easily missed unless the direct path attenuates the associated valve sideband quite a lot. The range of interest is generally from 1.5 - 4 KHz, so the peaks are pretty close together on the screen in both modes.This is one use that could really benefit from zooming and panning.
- DickN