DIY DSP via PC Questions (4~8 channels)

Hi all,

I'm very green to the idea of a PC based DSP. I currently run a MiniDSP HD2x4 but if you've seen what they're doing due to tariffs, it's crazy priced now to the point of extinction. Most other commercial options don't seem to come clone to this, REW importing, filtering to 10hz, etc. All this is pretty big for me as I use this for a large subwoofer network (16 of them). I'm looking to expand and instead getting more MiniDSP I'm looking to see if I can just do it myself with a PC instead. I've searched a bit, seeing Hifiberry, CamillaDSP, etc. But I'm not sure what would be ideal for my purpose. I went to YouTube thinking there would be videos on these to showcase their use, but sadly, didn't see much.

I'm looking to do a PC based DSP that is fairly easy to setup with the context of just doing typical delay, phase, EQ with several channels of subwoofers. Ideally able to filter down to 10hz or lower. I really appreciate features like importing REW, infrasonic filters and a good compressor or limiter. Those are most important to me. I've never built one or know what all goes into it. I'm happy to build a PC for it, or get a little PC or Pi or whatever. It's mostly to explore DIY and have something modular that I can scale to larger amounts of outputs.

Any suggestions for hardware and software for this is greatly appreciated!

Very best,
 
Something relatively easy to set up and use... you got it:
https://github.com/charlielaub/GSASysCon

You get:
IIR filtering and EQ, as much as you want.
You can get a compressor-limiter LADSPA plugin and use it here (see "Fast Lookahead Limiter"): https://github.com/swh/ladspa/blob/master/fast_lookahead_limiter_1913.xml
No fancy direct import of REW, but you can probably easily whip up a converter in Excel, etc.
FWIW, any DSP software will be able to do "filter down to 10hz or lower" when you choose the right sample rate

For GSASysCon, you can use whatever computer hardware you want but the computer should run a LinuxOS. Raspberry Pi is great, an old laptop or miniPC running Ubuntu, etc. Then get a USB audio interface (e.g. pro recording interface). With that you can start with a basic unit like a Behringer UMC1820 and then upgrade to a more expensive unit like a MOTU Ultralite mk5 or Focursite Clarett+ 8Pre.

This is the way I have been doing DSP for 10+ years.
 
  • Thank You
Reactions: MalVeauX
Something relatively easy to set up and use... you got it:
https://github.com/charlielaub/GSASysCon

You get:
IIR filtering and EQ, as much as you want.
You can get a compressor-limiter LADSPA plugin and use it here (see "Fast Lookahead Limiter"): https://docs.kdenlive.org/en/effects_and_filters/audio_effects/swh_plugins/index.html
No fancy direct import of REW, but you can probably easily whip up a converter in Excel, etc.
FWIW, any DSP software will be able to do "filter down to 10hz or lower" when you choose the right sample rate

For GSASysCon, you can use whatever computer hardware you want but teh computer should run a LinuxOS. Raspberry Pi is great, an old laptop or miniPC running Ubuntu, etc. Then get a USB audio interface (e.g. pro recording interface). With that you can start with a basic unit like a Behringer UMC1820 and then upgrade to a more expensive unit like a MOTU Ultralite mk5 or Focursite Clarett+ 8Pre.

This is the way I have been doing DSP for 10+ years.

Thanks Charlie, I'll begin reading this stuff and look at these options. Appreciate your input and efforts!

Any experience or suggestions in terms of being able to run this on a distro of Ubuntu in a virtual desktop? I spin up Ubuntu for various tasks in a VMWare workstation on one of my Windows based PCs for work I do with Linux there. I could use that possibly to learn how to do this with possibly and it's just a matter of learning the software right away without having to have a different PC to work with at first. I'll look at interfaces. I assume it's ok to explore more than those options. I could live with a 4 channel, but it would be ideal to have 8 probably for future expansion.

Very best,
 
CamillaDSP is great, I wrote a guide showing how to use it with a raspberry pi and variety of DACs, also includes remote / display integration.

https://github.com/mdsimon2/RPi-CamillaDSP

Michael

Thanks Michael, I shall read it. Appreciate your effort!

Do you know if this has the ability to do compressor/limiter based on signal level?
And a big one, can it handle REW import files?

Very best,
 
What will be the source? Player apps on the same pc? Or another pc, or a TV/PlayStation/streamer/.. ?

LFE output from an AVR. As explained above its for a subwoofer network. This is for a home theater. 16 subs in 4 groups of 4 so need 4 channels total. Only needs 1 input. It's a single bass array so all the phase and time is the same. So just looking to have RCA or XLR input, DSP processing, output RCA or XLR. Currently do this with a MiniDSP HD 2x4.

Very best,
 
Ok, so you want low latency. Camilladsp on a dedicated machine can reach a few milliseconds. Someone sho has tried and actually measured can maybe chip in here? It doesn't need to be very fast machine, a pi 5 would likely do.
A miniDSP will reach lower latency, and with the advantage that the latency is fixed. With a software solution the latency may vary a little. Camilladsp tries to time things during startup to get the same latency every time but it's not perfect. Again I don't have any hard numbers, but I would say if can vary by a millisecond or two.
 
Ok, so you want low latency. Camilladsp on a dedicated machine can reach a few milliseconds. Someone sho has tried and actually measured can maybe chip in here? It doesn't need to be very fast machine, a pi 5 would likely do.
A miniDSP will reach lower latency, and with the advantage that the latency is fixed. With a software solution the latency may vary a little. Camilladsp tries to time things during startup to get the same latency every time but it's not perfect. Again I don't have any hard numbers, but I would say if can vary by a millisecond or two.

Thanks, no worries, if the latency is a known value and isn't super variable (like more than 1ms each time different) that's fine, I can use that latency deducted from my delay to the network and still have everything in time without much worry.

Very best,
 
I didn't catch you were not running all channels through CamillaDSP. I don't think this is the greatest idea as the latency will be variable. If you keep CamillaDSP running, it should be constant but if you ever restart it will change. I would expect about 1-2 ms of variability based on latency testing.

How many total channels do you have out of the AVR? You might be able to run them all through an interface and avoid any issues with differential latency.

Michael
 
  • Like
Reactions: MalVeauX
I didn't catch you were not running all channels through CamillaDSP. I don't think this is the greatest idea as the latency will be variable. If you keep CamillaDSP running, it should be constant but if you ever restart it will change. I would expect about 1-2 ms of variability based on latency testing.

How many total channels do you have out of the AVR? You might be able to run them all through an interface and avoid any issues with differential latency.

Michael

Hi,

Thanks, yea, just running the LFE from the AVR through it. My bed layer speakers are very efficient and don't need hardly any amplification and the AVR has DSP built in. Just not below 20hz. I use a MiniDSP to do that for the LFE. Bed layer is 7 channels. Height layer is 2 channels. Then LFE. I only need 1 channel input and 4 channel output for LFE DSP work for my SBA wall.

Very best,