how to make a perceptual spectrum analyzer?
the magnitude output of rfft~ seems to need some tweaking, like:
- correcting amplitude of bins
- spacing the bins from equally linearly spaced to equally logarithmic
spaced (for outputting a set of frequencies like: 62 125 250 500 1000 2000 4000 8000 16000)
does anyone already did it? in part 2), apart I don't know how to do it, I'm not sure if I need to sum up all the bins belonging to a frequency band, or to average them ...?
~
have a look if you can find some constant Q filter analyser or bar
filters, it can be a good starting point.
hope that helps
loic
On 11 oct. 09, at 21:26, mescalinum@gmail.com wrote:
how to make a perceptual spectrum analyzer?
the magnitude output of rfft~ seems to need some tweaking, like:
- correcting amplitude of bins
- spacing the bins from equally linearly spaced to equally
logarithmic spaced (for outputting a set of frequencies like: 62 125 250 500 1000 2000 4000 8000 16000)
does anyone already did it? in part 2), apart I don't know how to do it, I'm not sure if I need to sum up all the bins belonging to a frequency band, or to average
them ...?~
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
well in other words, but you probably understood you can use a filter
bank and measure energy of each filter outputs using env~
maybe someone already did it...I don't know, else there are 24 bark
bands, here you can find explanation:
https://ccrma.stanford.edu/~jos/bbt/Bark_Frequency_Scale.html
loic
have a look if you can find some constant Q filter analyser or bar
filters, it can be a good starting point.
hope that helps
loic
On 11 oct. 09, at 21:26, mescalinum@gmail.com wrote:
how to make a perceptual spectrum analyzer?
the magnitude output of rfft~ seems to need some tweaking, like:
- correcting amplitude of bins
- spacing the bins from equally linearly spaced to equally
logarithmic spaced (for outputting a set of frequencies like: 62 125 250 500 1000 2000 4000 8000 16000)
does anyone already did it? in part 2), apart I don't know how to do it, I'm not sure if I need to sum up all the bins belonging to a frequency band, or to average
them ...?~
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
I've got a couple externs in my timbre analysis set that do bark- and mel-frequency cepstrum. You can set the spacing of the filterbank in barks, for instance. I was thinking about making externs for plain bark spectrum rather than cepstrum too, which might be exactly what you're looking for.
You can download the source and mac binaries on this part of my site. It's the third item down the page: http://williambrent.conflations.com/pages/research.html
Let me know if that helps, William
On Sun, Oct 11, 2009 at 12:26 PM, mescalinum@gmail.com mescalinum@gmail.com wrote:
how to make a perceptual spectrum analyzer?
the magnitude output of rfft~ seems to need some tweaking, like:
- correcting amplitude of bins
- spacing the bins from equally linearly spaced to equally logarithmic
spaced (for outputting a set of frequencies like: 62 125 250 500 1000 2000 4000 8000 16000)
does anyone already did it? in part 2), apart I don't know how to do it, I'm not sure if I need to sum up all the bins belonging to a frequency band, or to average them ...?
~
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
-
Loic Kessous -
mescalinum＠gmail.com -
William Brent