Author Topic: Reverb IC / impulse response dsp  (Read 3598 times)

sbirkenstock

Reverb IC / impulse response dsp
« on: November 29, 2016, 04:26:49 AM »
Hi Everybody,

I am currently tinkering around with the Spin FV-1 and the Belton BTDR-3 chips.
Both only offer limited variaty in different reverb styles.
Is there any other chip out that offers more reverbs (Spring, Room, Plate etc) with adjustable "decay" time?
Belton has some DSP, but I did not find any with the possiblilty to adjust the decay via an pot like the BTDR-3 can.

Some pedals I like use a lexicon chip to my knowledge (Digitech Polara, Hardwire Supernatural) that I like a lot.
I could not find any source to buy a lexicon chip though.
My favorite sounding reverb pedal is the G Lab DR3 Dual Reverb (only has two reverbs to select from).
Does anybody by chance know which chip it is using?


Logidy offers a pedal that processes impulse response files.
http://www.logidy.com/
I have this pedal, and the effects are very cool, but it modifies the original signal too much imo.
Does anybody know which chip is used here or any other chip that does process impulse response files?

Best regards,

Stephan


gena_p1

Re: Reverb IC / impulse response dsp
« Reply #1 on: December 02, 2016, 08:19:40 AM »
Belton BTDR contain no DSP, only simple PT2399 chips. It could not be reconfigured, except tails length.

Keith Barr  formed Spin Semiconductor and produced Spin FV-1 chip, which have more possibilities , than previous Alesis AL3201 (AL3201 is DSP).

FV-1 is simplier, need only couple of resistors and i2c eeprom with programmer to download chip with your own software.
Lot of expirience is on Spin Semi's forum: http://www.spinsemi.com/forum/index.php and official web site

But, writting programs for FV-1 is not couple of clicks.
« Last Edit: December 02, 2016, 08:26:52 AM by gena_p1 »

Digital Larry

Re: Reverb IC / impulse response dsp
« Reply #2 on: December 02, 2016, 09:48:39 AM »
I've been messing around with the FV-1 for several years now.  Most of the available reverb algorithms are variations on a specific structure, which is usually:

Cascade 4 all pass filters
Send this signal into a ring of chained comb filters (delay lines with feedback) which can also include high pass and low pass filters and extra all pass filters in each block.

Recently I got Julius Smith's book on Physical signal processing - online version here:  https://ccrma.stanford.edu/~jos/pasp/

This contains a section on reverberators which you will probably find interesting to read. This stuff is taught to PhD engineering students so expect lots of math but some of the concepts are fairly straightforward.  In any case, I made a "room reverb" following JOS's guidelines simply using 3 delay lines with feedback and low pass filters and it does not sound bad.  I had not seen this structure in any of the examples supplied by Spin.

Another interesting thing about reverb algorithms is that they have various structures (Schroeder, Feedback Delay Network, Gardner, etc.) and with the exception of the FDN, don't leap out at you with "HALL", "PLATE", etc. descriptions.  So you may have to intuit what the characteristics of pre-DSP reverberators really are sonically and then figure out how to make those things happen.

The FV-1 can deliver more reverb types than you might imagine, but in order to accomplish those things you are going to have to dedicate a sizeable chunk of your life to study and experimentation.
Digital Larry
DSP tinkerer and former transistor twister

ElectricDruid

Re: Reverb IC / impulse response dsp
« Reply #3 on: December 02, 2016, 07:28:50 PM »
Quote
The FV-1 can deliver more reverb types than you might imagine, but in order to accomplish those things you are going to have to dedicate a sizeable chunk of your life to study and experimentation.

In many ways I'm glad to hear it. I want this to remain an art that demands a certain level of effort and respect! If you can simply bash it together and get a good result, where's the fun?! For me, the challenge is squeezing a little bit more out of hardware that shouldn't really be capable of such a convincing result. I think Keith Barr thought the same way. I never met him, but all his work speaks to me of this same idea.

In this age of 32-bit chips that cost pence, the simplest solution is always to throw more computing power at it, but we forget the skills of the people that got us here if we do that. I grew up on 8-bit chips and I'd still rather do more with less, if I possibly can. "More with less" is still a valuable skill, whatever level you work at.

Tom

Digital Larry

Re: Reverb IC / impulse response dsp
« Reply #4 on: December 04, 2016, 10:37:24 AM »
One thing I noticed early on is that for utility always-on reverb sounds for guitar anyway, something like the MN3011 6-tap delay emulation I came up with is as good as some more elaborate hall reverbs since you're just trying to give it a little bit of ambience.  The difference becomes more important when your source material is more sparse or percussive, and/or you're really trying to go for that 20 second decay time, or you're sitting there with headphones listening to isolated sounds.

One thing that bothers me about the Spin examples is that all numbers such as delay line length in a comb filter ring are just presented as "magic numbers".  The JOS book elaborates that these should generally be selected to be mutually prime and even spread out over multiple decades, e.g. 13, 331, 1024.  Spin examples might or might not be mutually prime (I haven't checked) but they are mostly not spread out that far either.  I spent a bit of time just messing with FV-1 reverb parameters randomly for awhile, but since each complex algorithm has 20 or more things you could vary, it gets really challenging because an improvement over an existing patch might require you to adjust 5 or 6 things at a time.  Compare this to a flanger or chorus, where the sonic impact of changing a single parameter such as LFO speed, width, or wet/dry mix is fairly obvious and you can evolve a patch in a generally positive direction by adjusting one thing at a time.

So I have utmost respect for those who have spent the time to dive deeply into all of that since I don't think I'll be doing too much of it myself.
Digital Larry
DSP tinkerer and former transistor twister

Ice-9

Re: Reverb IC / impulse response dsp
« Reply #5 on: December 04, 2016, 02:36:29 PM »
Larry, I do remember reading that these 'magic numbers' that Keith Barr used in the FV-1 algo's were numbers he used as 'knowing what worked.' I expect for Keith, many years working with reverbs these magic delays sizes  just become second nature and embedded in his knowledge of what produces a good result. I suppose just like anyone who has a great knowledge of music knows automatically what notes will work perfectly in any key and out of key when jamming instead of just reverting to the pentatonic, it's all about accumulated knowledge.

While a lot of reverb delay lines numbers etc are in a lot of reverb papers, which you have also posted links for (above) Cheers. It would have been great to have read some information about how Mr Barr came to his preferred numbers and why they work well.

Sorry a little OT @Larry, Can I ask you if I can put up a link to your SpinCad forum/site on my website for anyone wanting to use Spincad with the Blue Nebula ?





 

www.stanleyfx.co.uk

It's fairly straight forward, if you want to start it , press start. You can work out the rest of the controls for yourself !

Digital Larry

Re: Reverb IC / impulse response dsp
« Reply #6 on: December 04, 2016, 03:51:13 PM »
Sorry a little OT @Larry, Can I ask you if I can put up a link to your SpinCad forum/site on my website for anyone wanting to use Spincad with the Blue Nebula ?

Ah sure of course!
Digital Larry
DSP tinkerer and former transistor twister