[LAU] Frequency-response analyzer for Linux?

[Ken Restivo]

On Fri, Jul 17, 2009 at 05:43:09PM -0700, Justin Smith wrote:
> On Fri, Jul 17, 2009 at 4:47 PM, Ken Restivo<ken at restivo.org> wrote:
> > On Tue, Jul 14, 2009 at 04:13:13PM -0700, Justin Smith wrote:
> >> On Tue, Jul 14, 2009 at 3:56 PM, Fons Adriaensen<fons at kokkinizita.net> wrote:
> >> > On Tue, Jul 14, 2009 at 03:25:39PM -0700, Justin Smith wrote:
> >> >
> >> >> Try running a white noise source through the filter, then look at the
> >> >> spectral graph output in your favorite spectrum analysis program.
> >> >> Since white noise should have a flat spectrum, any peaks and valleys
> >> >> in the output are those of the filter.
> >> >
> >> > For a more 'musical' analysis, use Japa. It has either
> >> > a logarithmic or perceptual frequency scale. Pink noise
> >> > source is built-in, and if you set the analyser response
> >> > to 'Proportional' it will show a flat spectrum for pink
> >> > noise. It can also show the difference between two spectra,
> >> > e.g. input and output of a filter, and this produces much
> >> > more stable displays when using a noise test signal.
> >> >
> >> > For the filter, you could try on of the four Moog VCF
> >> > plugins I wrote years ago. Apart from the resonant
> >> > lowpass they will also add some juicy distortion at
> >> > high levels, as does the original Moog filter and
> >> > probably all real analog Wah processors. Frequency
> >> > control is logarithmic, which is probably what you
> >> > want, and there are separate control inputs to
> >> > set the static frequency and resonance and their
> >> > modulation. One problem could be that the modulation
> >> > inputs require audio rate signals (the filter were
> >> > designed for use in AMS) but that can be changed
> >> > easily if you want. If you want to play with them
> >> > use AMS.
> >> >
> >> > But Japa and the filters to be found at
> >> > http://www.kokkinizita.net/linuxaudio/downloads
> >> >
> >> > The filter plugins are in the MCP set.
> >> >
> >> > --
> >> > FA
> >>
> >> I used japa years ago, but forgot its name, thanks for this excellent
> >> piece of software, which I am excited to have rediscovered.
> >
> >
> > Mille grazie! I've got JAPA going on two platforms, but I've got to apologise for being a bit dense in figuring out how to use it.
> >
> > The UI looks a bit like an HP oscope, but, alas, i'm not much of an engineer, so I'm not sure to start. Is there a quick simple tutorial somewhere on how to use it? Anyone care to give me a few sentences of advice?
> >
> > I basically want to put white noise through some filters (prolly using JACK-RACK), and view the frequency response of it. Looks like JAPA already has a built-in white an pink noise generator, which is nice. What would I connect to what (using i.e. using jack_connect) and how would I set up the switches and control on JAPA to do this? A simple screenshot of the correct setup might be enough to get me going.
> >
> > Thanks again!
> >
> > -ken
> >
> 
> Connect the pink output of japa to the input of the application with
> the filter, connect the output of the app with the filter back into
> japa, the curve you see in japa will be the frequency response of the
> filter. Since japa can overlap displays, you can also try putting some
> musical material through: one input straight into japa, one through
> the filter first, and look at the compared spectrums.
> 
> Another possibility would be routing the japa pink output to an analog
> wah pedal, and comparing the frequency response of the analog pedal
> coming back in to the digital wah emulator. Or you could compare the
> spectral output of the real and digital wahs with musical input.


Thanks.

I did as you suggested, and it worked! But I'm confused by the results.

This is straigtht pink noise directly from JAPA's generator back into its input... but the frequency response appears not to be flat!
http://restivo.org/misc/pink.png

The white noise was similarly not flat:
http://restivo.org/misc/white.png

So, in terms of viewing the response of the filter, do I need to A/B or A-B to get some kind of estimate of what the filter is doing?

-ken