Biquad calculations for custom filter design

[gedlee]

I can do some things in C, but I never got to liking it. I have looked at C# and that looks very impressive.

I had a math intensive app (I didn't write it) in C++ and whenever I ran a certain program on it it would just keep chewing up memory until it ran out. Then it would start disk swapping the memory, and finally just stop. You could run it just fine several times, but then you had to reboot the system to release the memory back to the OS. You couldn't get it back otherwise.

 

[parb]

Hmmm

I can do some things in C, but I never got to liking it. I have looked at C# and that looks very impressive.

I had a math intensive app (I didn't write it) in C++ and whenever I ran a certain program on it it would just keep chewing up memory until it ran out. Then it would start disk swapping the memory, and finally just stop. You could run it just fine several times, but then you had to reboot the system to release the memory back to the OS. You couldn't get it back otherwise.

C Is my first love. Did the others first but c was the language that defined my programming skills....

If a program his memory and it doesn't get released when it exits then it must reserve system resources and then system doesn't know how to release them safely. Probably a real bug IMHO.

 

[twest820]

I have some ideas I'd like to try out in a real system (rather than just Matlab) which require direct input of biquad coefficients rather than going through one of the standard EQ patches. However I don't have a SigmaDSP part or similar handy; does anyone know of a VST which supports coefficient entry? I've been searching but haven't found one yet. Wouldn't be that hard to write one, but before I go reinventing the wheel...

 

[boris81]

Here is a biquad coefficients calculator.

EarLevel Engineering

Try it out and let us know if it works. Keep in mind that this is a second order biquad implementation so odd order filters won't work. You could cascade multiple filters to get a higher order slopes.

Here's also a link to Q-factor look-up table for the most common filter slopes. Again this calculator only works with even orders.
http://focus.ti.com/lit/ml/sloa088/sloa088.pdf

 

[twest820]

Thanks, but I get no hits for VST at earlevel.com. I've no shortage of tools for coefficient calculation, just haven't gotten around to writing a VST which could be used to prototype with them in real systems (and am too cheap to buy a MiniDSP for that one purpose ).

 

[boris81]

Thanks, but I get no hits for VST at earlevel.com. I've no shortage of tools for coefficient calculation, just haven't gotten around to writing a VST which could be used to prototype with them in real systems (and am too cheap to buy a MiniDSP for that one purpose ).

The previous post was for the miniDSP people. I stopped using VSTs, I'm not that impressed with them.

I'm too cheap to buy miniDSP too. I use kx-project. It's a free custom driver for the Creative Audigy 2 soundcards that allows the user to re-program the DSP chip. A multitude of filtering, routing and other tasks can be realized by simply arranging function boxes in the user interface. The downsides are 16/48KHz sampling rate and the price of an old soundcard on ebay ($30). It works great for me because I have all my music on the computer anyway.

 

[twest820]

I looked at kx-project a while back but, not having a desktop, I happen to be kind of stuck; since most Extigy and Audigy 2 NX processing is in software and kx-project's planned support is ASIO based I suspect the delta to VSTs on ASIO is probably small. Anyone happen to have experience with Noise or VST.NET? My interest here is strictly for proto and I anticipate needing only a few biquads---I've another project going to DIY a board with a SigmaDSP, though ARM's Cortex M4 is rather interesting too---so I'm more concerned with rapid development than performance.