I'm in the process of finalizing a VST Spectrum Analyzer much similar to Bram's S(m)exoscope.

here are some screenshots





I'm open to feature request,
but keep in mind that I'm not working on it full time and that I won't write a single plugin that can do everything.
however I've added several features I wanted a spectroscope to have, and here is your chance to ask for your own.
I'll see what I can do.

already there:
- lin/log/semitones/third octave scale resolutions
- window size selection
- frequency zooming
- freeze
- cursor information (amplitude+frequency)

planned:
- Bark/mel/Erb frequency scales
- spectrum averaging
- basic informations like spectral centroid, dominant pitch...

Bless you - I've been using Whitehurst's FFT plugin, but never really trusted it, as it tended to kill my old WinME machine. Is the amplitude scale adjustable?

not yet but I can do it,
do you find 100dB is too much?
BTW I use 10log10() dB definition

seeing the note-scale on one of the shots, i wonder can this be used to tune instruments?

of course, you'll see the fundamental and it's overtones

Occasionally. Actually, in a perfect world you could adjust both ends of the range. Sometimes you're looking for "good enough for rock-n-roll" so you don't care about anything below 60dB down, sometimes your input signal is very low amplitude to begin with and you don't want to waste the screen space between 0dB and 20dB down, for example.

Awesome addition to smexoscope! Great for us devs

Hurm, so 1/65536 = 16 bit LSB = -48dB? I'm used to that being called -96dB.

actually I was using 20*log10() but then I changed to 10*log10 because i saw it somewhere in a specifications, but maybe it was supposed to be applied on a squared signal.

I have another doubt concerning the FFT scaling since there's no aggreement on it.
Should I divide by N or sqrt(N).

I'm not too regarding on absolute values when it comes to spectrum, but I know some people might be more demanding than me on that.
So tell me what do you prefer.

cheers

rémy

cool, i think it's a nice progr' !!
but i was searching for a similar progr' thats do the same, and transform the analyze in midi notes or midi message (in realtime) ==> so you have a realtime audio to midi interface who was never done before (may i'am wrong but i don't think so!)

in fact what you ask for is something I've been thinking about for a while, but the task is much more complex than how it looks like at the first glance.

maybe one day...

There aren't much things left that haven't been done before some way or another these days
- http://www.tazman-audio.co.uk/products/TheExtractor.htm
- http://www.knzaudio.com/index.php
- http://www.digital-ear.com/digital-ear/info.htm
- there are others: google for "audio to midi"
As Remy said: it's not simple to write a good pitch-to-midi converter with low latency working in real-time. But that's exactly the point where research and development meet, right?

BTW:

I highly recommend reading this paper before doing anything related to pitch:
http://recherche.ircam.fr/equipes/pcm/cheveign/sh/ps/pitch_A dC_030131.pdf

mdsp, this looks good and useful.

But, why the installer?, can we please have a "plain" dll version?

Thanks mdsp, this is terrific!

This was also the first time I heard about Brams oscilloscope program, which is another great little plugin. I really like the way both programs have the same look.

As for feature requests, what about being able to not only average spectrums, but to also subtract one spectrum from another, add two spectrums together, and also the multiplication of two spectrums?

For the math operations above, you could compare the left versus the right input signals in realtime. This should be really handy in modular hosts, because it would let you compare the input spectrum and output spectrum of a VST plugin, and see what changes were being made to the spectrum inside the VST plugin.

You could also add the ability to save a spectrum to a file. Then, use that saved file as one of the spectrums in the math operations mentioned above. This would probably eat up less CPU, since one of the spectrums would be fixed, and wouldn't have to be updated in realtime.

Ideally, you should also be able to import two saved spectrums and perform the math operations on them as well. There should be no reason to limit the math functions to realtime use only.

I would also like to be able to manually input the frequency response curve of microphones or speakers, and apply a corrective curve to the displayed spectrum. It should be handy for analyzing mic and speaker response patterns. It should be possible to tune the frequency response of your studio monitors, for example. Naturally, any such corrective curves should be saveable to a file, for future use.

It would also be very cool if you could provide a file export feature, which saves all the spectral data to a human readable text file. There is a nice feature like this in Audacity, but Audacity doesn't work inside a VST environment. I would really like this capability in the VST realm, via your plugin.


thanks,
McLilith