Linux/BruteFIR support coming in LspCAD

[tcpip]

Please check out this thread about the background. Can any LspCAD and Linux users out there please give feedback about how you use BruteFIR today, what you'll need from LspCAD, etc? Please post your responses here... my friend will inform Ingemar about this thread.

[cdwitmer]

FWIW here is a Windows (not Linux) application for using a computer as an electronic (digital) crossover, and for doing DRC:

http://pcazeles.perso.cegetel.net/acxo.htm

If that is of zero relevance to what you are after, please forgive me.

[tcpip]

FWIW here is a Windows (not Linux) application for using a computer as an electronic (digital) crossover, and for doing DRC...

This is quite relevant, and I didn't know about it, so, thanks for the pointer.

However, BruteFIR continues to be tempting because the OS on which it runs is royalty free, allowing some savings in licence fees and unlimited copying, modifying, and customising. I wouldn't know how to fit Windows into a small Flash RAM chip, for instance, but I (and other Linux/FreeBSD types) know how to reduce both the kernel and the user-mode programs to fit into 40MB or less without trouble. This makes for good embedded applications.

But thanks for the pointer... I've bookmarked that page.

Aren't there any BruteFIR veterans on this forum?

[cdwitmer]

If my (limited) understanding is correct, ACXO utilizes BruteFIR. I'm not sure how ACXO was put together, but depending on how it was done, porting it to another OS may not be difficult. If you have any burning Qs I can pass them on to the developer, with whom I correspond occasionally.

[ThomasW]

BruteFIR has a fair number of users (none that I'm aware of are HT-Guide members). There have been long posts on other forums, the DIYloudspeakers list (aka Basslist) and other such email lists about using it.

At this point I don't think there is one central repository for BruteFIR information, unless Ed Wildgoose's site in London is up and running. (unfortunately it appears I've lost his URL)

[JoshK]

I have a linux box that I just setup to futz around with learning enough linux to start toying with such an application. It is going to be a long way down the road before I get anything working with BruteFIR.

[capslock]

If my (limited) understanding is correct, ACXO utilizes BruteFIR. I'm not sure how ACXO was put together, but depending on how it was done, porting it to another OS may not be difficult. If you have any burning Qs I can pass them on to the developer, with whom I correspond occasionally.

If it does, then only indirectly through the room equalizer module. The XO's it offers are restricted to IIR Butterworths.

[dwk]

Since nobody has (yet?) ported BruteFIR to windows, it's clear that ACXO doesn't use it. I believe that ACXO uses the mkfilter package for the crossovers and adds a single convolution stage in front of the xover to run the DRC filters. I looked at ACXO a while back, but I wasn't thrilled with the decision to embed the functionality in the player itself - I really prefer a modular approach.

I've played around with dsp xovers quite a bit over the years, including BruteFIR. IMHO providing some level of an export for FIR filter coefficients from LspCad/SoundEasy should be an 'expected' feature. It's certainly possible to grab an impulse from them the way they work now, but it's inconvenient to say the least. If I were to make a suggestion as to how the integration should work, I'd say something like this:
- assume Jack is being used, as it simplifies things greatly. more burden on the user, though, so advertising it as 'plug-and-play' would be a stretch
- support only a couple stock configs (ie 2-way, 2.1, 3, 3.1) and use standard port names (ie right-high, left-bass etc)
- leave connection logic out, and rely on the user to do connection management via Jack.

BruteFIR is by far the most capable convolution engine out there, to the point that it may in fact be overkill for a 'simple' crossover application. I've drifted away from it recently for a couple reasons:
- my Emu 1820M is noticably better than my Delta 1010, but isn't supported under Linux
- needed a Win box to run SoundEasy anyway
- I'm looking more at subtractive xover topologies at the moment - seem to offer more bang for the computational buck, although it'll require some custom coding on my part.

Having said all that, BruteFIR really lends itself to 'appliance' use. Boot of a CF disk, take input via spdif, and you'd almost never know it was there.

[tcpip]

Good to hear from you. The first time I'd received any inputs on BruteFIR was from you, on the diyaudio forum.

I've played around with dsp xovers quite a bit over the years, including BruteFIR. IMHO providing some level of an export for FIR filter coefficients from LspCad/SoundEasy should be an 'expected' feature.

Yes, I too thought this is a very natural fit. After all, BruteFIR, for better or worse, seems to be the only freely redistributable engine on any OS for doing digital xo.

Having said all that, BruteFIR really lends itself to 'appliance' use. Boot of a CF disk, take input via spdif, and you'd almost never know it was there.

My friend has passed on the URL of this thread to Ingemar, so we hope Ingemar is tracking it. It would have made things much easier if Ingemar had interacted on these threads, but he doesn't seem to have the time.

[tcpip]

... It would have made things much easier if Ingemar had interacted on these threads, but he doesn't seem to have the time.

It appears Ingemar is serious enough about this thing to actually interact on these forums. He's posted on diyaudio.

I hope, dwk, you and others can help him along.

[ergo]

Ingemar has just posted an update for ver 6

LspCAD 6.20 (2005-10-11)
+ G(s) G(z) component, Digital phase lineal FIR filter component added, possible to simulate and
export filter coefficients for FIR filters with filter orders 129..4097 with various window options.

Ergo