Spectrogram that behaves like Waveform~?(gets data from buffer~ instead of signs

Jan 7, 2013 at 2:49pm

Spectrogram that behaves like Waveform~?(gets data from buffer~ instead of signs

I’d need a spectrogram/sonogram that’s not realtime like Spectroscope but instead gets its data from buffer~.

anyone knows an objects like that?

Jan 9, 2013 at 2:59pm

Don’t tell me there is none? It seems simpler than realtime analysis.

Jan 9, 2013 at 4:09pm

No one is answering you because your question does not make sense. a buffer~ represents audio that takes place in time, representing thousands of analysis frames, so it is ‘realtime.’ So no one knows what you want, and you’ll have to state your problem more clearly.


Jan 9, 2013 at 4:24pm

Oh okay.

Waveform~ is a visual representation of the whole buffer~ contents. It displays buffer~s data without the need to feed it any signal.
I’d like a spectrographic representation of buffer~.
You would assign a “spectroscope~” to a buffer and it would display the whole buffer~s content spectrographically. (as opposed to playing it to get a scrolling display)

Jan 9, 2013 at 6:22pm

FWIW, i have an external that calculates the FFT/iFFT offline. it takes data from a buffer, calculates magnitudes and phases (or simply real and imaginary parts) and puts those back into the buffer and vice versa. but it treats the buffer contents as one input frame and thus calculates one huge FFT out of it.
if that is what you want, you can find the object here: http://www.esbasel.ch/software/#vb-objects

otherwise, if you need overlapping frames, i once did an object that calculates phase-vocoder data of buffer contents, i.e. frames of amplitude and frequency data offline, which is then copied into a separate buffer for later use in a real-time phase-vocoder patch.
the external is not ready for prime time, but if there is interest i could have a look what is missing.

then, i read a few days ago that someone (i think adam siska) is planing to make an object for offline FFT.
and of course there is jit.fft which works quite nice, albeit a little slower than external solutions.

Jan 10, 2013 at 9:06am

and you should explore the hissTools IR Toolkit. There is a very, very good spectral visualizer in there that works in non-real-time on a buffer content if you want…



Jan 10, 2013 at 10:40pm

@volker böhm not exactly, but perhaps it will serve my purpose nontheless, will give it a whirl. Thanks a lot!

@PIerre Alexandre Tremblay: link doesn’t work unfortunately.

What I basically want is a view that iZotope RX offers

Jan 11, 2013 at 12:56pm

What I basically want is a view that iZotope RX offers

doesn’t work with waveform~ as it can only display one value per sample index.
you’d need a matrix for that.

check out the HISStools (link works for me). if they don’t do what you want,
here’s a sketch using jitter. it’s not terribly fast, but works ok for small files.
maybe this gets you an idea of how to patch this yourself.

– Pasted Max Patch, click to expand. –
Jan 29, 2013 at 3:31pm

@VB: Works for me now. Wow, the patch you made is perfect. I would have never ever thought that. Thank you so much. I’ve relied so much upon your work (starting with vb.cheby) I’m putting you in thank-you notes if I ever get to publish anything.

Dec 25, 2013 at 11:55am

Dear Ploki and dear Mr. Bohm,
the sketch patch that has been linked actually visualize a Waveform (what is the point of doing a FFT to visualize a waveform?) and not a spectrogram.

Am I doing something wrong?

Dec 25, 2013 at 12:07pm

Sorry, i just forgot to initialize the loadbang correctly!
Works perfectly, just very slow


You must be logged in to reply to this topic.