FatKingTubby wrote: ↑Thu May 21, 2020 2:17 pm
The Fumana v. Bark Filter quandry has been on my mind since finally indulging on a Fumana back in October.
...............
As someone's already mentioned, the major difference between the two is the tuning of each band. The Fumana is much more focused on bands below 1kHz (9/16 bands) and the Bark has way more above 1kHz (8/12).
That may be because Mark's Bark is missing some -over half- of its 'bite'! And its 'tooth' alignment is off by half.
Here's the complete Bark filter center freqs:
50, 150, 250, 350, 450, 570, 700, 840, 1k, 1k17, 1k37, 1k6, 1k8, 2k15, 2k5, 2k9, 3k4, 4k, 4k8, 5k8, 7k, 8k5, 10k5, 13k5
Here are the complete bark edges:
0,
100, 200,
300, 400,
510, 630,
770, 920,
1k08, 1k27,
1k48, 1k72,
2k, 2k32,
2k7, 3k15,
3k7, 4k4,
5k3, 6k4,
7k7, 9k5, 12k, 15k5
Here's Mark's frequencies.
<100 hz, 300, 510, 770, 1k08, 1k48, 2 k, 2k7, 3k7, 5k3, 7k7,
>10k5
I've bolded the Verbos above to make clear what Mark has chosen. And it's somewhat LESS 'verbose' than a bark filter should be. The whole point of bark filters is that they represent a mapping of our ear to the world. The different frequency edges represent points of 'banding' in our own audio perception. They are intended to be steep edges. Notice that the Verbos has not only skipped over half of these, he's also put his center freq's at the Bark
edges instead of the intended Bark center frequencies.
This means he's given us a familar name, but -possibly- missing the meaning behind it. Which is to apply filtering in a way that matches our physical audio perception.
A Bark filter's 24 bands *should* range like this:
0-
50-100
100-
150-200
200-
250-300
300-
350-400
400-
450-510
510-
570-630
630-
700-770
770-
840-920
920-
1k-1k08
1k08-
1k17-1k27
1k27-
1k37-1k48
1k48-
1k6-1k72
1k72-
1k8-2k
2k-
2k15-2k32
2k32-
2k5-2k7
2k7-
2k9-3k15
3k15-
3k4-3k7
3k7-
4k-4k4
4k4-
4k8-5k3
5k3-
5k8-6k4
6k4-
7k-7k7
7k7-
8k5-9k5
9k5-
10k5-12k
12k-
13k5-15k
Given this, you can see that the lack of filtering on the low end that you made comment on, is well founded. The true bark filter has NINE bands below 1K. Even if we eliminate the lowest as Mark has done, there are still eight. The Verbos 'Bark' filter gives us 4 or 5.
But even these do not appear to be matched to the science behind the Bark Frequencies existence. Verbos spec say they are 6 pole, and therefore should be plenty steep. What is their width? See link below and final note at bottom. With half of them missing, there are either gaps -if the width is held- or incorrect widths.
With this difference alone <between Fumana Freqs and Verbos> it seems like the audio sculpting possibilities are varied enough that both could be useful in their own applications, as well as being able to compliment each other.
<Added for clarity>
Getting to the underlying results of this is
that none of this means the Verbos 'Bark' filter isn't a good product, or isn't capable of manipulating your sounds in useful and effective ways. It surely can do that.
And to the point of your concern, it will definitely be different in practice from the Fumana.
Just don't expect you've experienced the point behind a Bark filter after using it. It's another set of FFB frequencies with a Fancy name.
I recently designed a true 24 band Bark filter -not for Eurorack- so all this is fresh and recent experience. I've never listened to the Verbos Bark Filter. All of this is going on their published specs, and an understanding of Bark Frequencies.
--------------
Freq centers and edges for this post were taken from here:
https://ccrma.stanford.edu/~jos/bbt/Bar ... Scale.html
This Wikipedia Article adds an important point. Ain't the vagaries of perception grand!
https://en.wikipedia.org/wiki/Bark_scale
Wikipedia wrote:
Since the direct measurements of the critical bands are subject to error, the values in this table have been generously rounded.[1]
In his letter "Subdivision of the Audible Frequency Range into Critical Bands", Zwicker states:
"These bands have been directly measured in experiments on the threshold for complex sounds, on masking, on the perception of phase, and most often on the loudness of complex sounds. In all these phenomena, the critical band seems to play an important role. It must be pointed out that the measurements taken so far indicate that the critical bands have a certain width, but that their position on the frequency scale is not fixed; rather, the position can be changed continuously, perhaps by the ear itself."
Thus the important attribute of the Bark scale is the width of the critical band at any given frequency, not the exact values of the edges or centers of any band.
edit: Allen Strange covers Bark Filters briefly on Pg 145-146 of his seminal book
electronic music, 2nd ed. He also emphasizes the band
widths /edit
with a trautonium filter, a triple resonance filter and that - that's a shit tonne of sculpting possibilities!
