I'm doing a ground up build and was wondering what would be a better solution for DSP in an active speaker? Hypex has an all-in-one option, but it doesn't allow for much interchangeability in gear. Sigmastudio+ , while it does a lot more, is confined to a certain platform. Camilla DSP is a very powerful solution, but that means I need to load it onto a SBC like RaspberryPI. What's the better tradeoff?
Moved to Multi-WayI chose to use minidsp product with Dirac for the same need.
CamillaDSP is interesting but as you mention, it requires a computer with all the maintenance associated and the risk of error and bugs (related to camilladsp or not). It's not user friendly, you don't want to be delving with ssh and command line options when you turn on the music and it doesn't play.
CamillaDSP is interesting but as you mention, it requires a computer with all the maintenance associated and the risk of error and bugs (related to camilladsp or not). It's not user friendly, you don't want to be delving with ssh and command line options when you turn on the music and it doesn't play.
What do you mean with this?
I personally find Sigmastudio a wonderful platform to work with.
That being said, the choices in general are very limited between either pricey (Hypex etc), or cheap Chinese boards that just perform suboptimal. (Often for stupid reasons).
There seems nothing in between
Thanks for the tip, that looks like it could do everything that I need and more, and could be a good upgrade for me
Sorry for the hijack BKrOn, Brian
I adopted the software DSP platform about 10 years ago for the following reasons:
1. much greater capabilities in terms of number of filters, number FIR taps, etc. compared to hardware based DSP
2. much greater flexibility in the filter path
3. flexibility of and control over the ADC end of the chain
4. ability to create new features and filters that are not always available via hardware DSP
Back then, miniDSP was a nice hardware option but it had very mediocre DACs. The DSP processing path was fixed - you got N filters on the input and M filters for each output. This was chosen by miniDSP and baked into their product via its SigmaStudio flowpath. Certain filters that I Wanted to implement were not available (later they came out with the advanced biquad, which mostly removed this limitation). Sample rates were fixed at whatever the plugin used. etc.
All of these limitations were removed once I moved over to DSP processing in/on a computer. I used the Pi platform for a few years but got sick of it and now I use small Celeron class mini PCs. Everything is just more reliable and you can use the standard kernel. Software updates are available sooner, etc. Fewer headaches IMO.
The downside of software DSP is that you can't easily build it, and amps, into a speaker. A plate amp with DSP like Hypex offers beats that in terms of ease of installation and design. But I have heard many bad things about their DSP interface, so caveat emptor. I typically have the computer and a stack of amps, and use multi-wire speaker cabling. No need to take up room in the actual speaker with these parts.
The computer based DSP is very flexible and powerful, and can easily be re-used for any number of projects if you find that your first build doesn't quite do it for you. You can easily upgrade the DAC based on your budget and "need" for some particular level of sound quality, low noise, or whatever. At the same time the rest of the chain (computer, software, etc) can stay the same, or you can also swap those out very easily too. From the DIY perspective I like this flexibility and it has served me well for loudspeaker building.
That's my 2 cents.
1. much greater capabilities in terms of number of filters, number FIR taps, etc. compared to hardware based DSP
2. much greater flexibility in the filter path
3. flexibility of and control over the ADC end of the chain
4. ability to create new features and filters that are not always available via hardware DSP
Back then, miniDSP was a nice hardware option but it had very mediocre DACs. The DSP processing path was fixed - you got N filters on the input and M filters for each output. This was chosen by miniDSP and baked into their product via its SigmaStudio flowpath. Certain filters that I Wanted to implement were not available (later they came out with the advanced biquad, which mostly removed this limitation). Sample rates were fixed at whatever the plugin used. etc.
All of these limitations were removed once I moved over to DSP processing in/on a computer. I used the Pi platform for a few years but got sick of it and now I use small Celeron class mini PCs. Everything is just more reliable and you can use the standard kernel. Software updates are available sooner, etc. Fewer headaches IMO.
The downside of software DSP is that you can't easily build it, and amps, into a speaker. A plate amp with DSP like Hypex offers beats that in terms of ease of installation and design. But I have heard many bad things about their DSP interface, so caveat emptor. I typically have the computer and a stack of amps, and use multi-wire speaker cabling. No need to take up room in the actual speaker with these parts.
The computer based DSP is very flexible and powerful, and can easily be re-used for any number of projects if you find that your first build doesn't quite do it for you. You can easily upgrade the DAC based on your budget and "need" for some particular level of sound quality, low noise, or whatever. At the same time the rest of the chain (computer, software, etc) can stay the same, or you can also swap those out very easily too. From the DIY perspective I like this flexibility and it has served me well for loudspeaker building.
That's my 2 cents.
Me also adopted software based DSP about 15years ago.It was a desktop PC back then. Now i am using ARM SBC(single board computer). They are credit card size these days. Advantages are the flexibility. It fits all, don’t need to redesign anything just change things on software side. Can do it remotely in other continent. Can build any filter or feature you want. Or run as a streamer or anything you wish.
The down side could be that they are less reliable than hw DSP at least theoretically. But how much reliability do we need? Linux runs on boards going into space and everything ok with reliability. So i am not worrying about that much.
The problems I have with everything computer based is the potential for failure. I had many occasions of volume surge, 100% volume in your speakers when you turn something on is a very displeasing surprise!
If you have some reliable indépendant hardware from the sbc to mitigate this, it is definitely a plus in my book.
If you have some reliable indépendant hardware from the sbc to mitigate this, it is definitely a plus in my book.
I have an Intel NUC with Linux and CamillaDSP in my drawer, I installed it almost 2 years ago. It is waiting for its time when the final pair of speakers is built. My plan to solve potential problems at boot is to use Extron amplifiers which have a mute input and volume input that can be controlled for multiple amps from a single place. I would only have to sort out a board controlled by the PC to unmute the amps and turn on a LED when all is booted and ready.
Thank you everyone for your input!
So dedicated hardware definitely seems to be a recurring theme here. That's not a terribly hard issue to figure out considering how many types of SBCs and processor boards there are. I'm building the amps and a carrier board for the DSP (whichever that may be) as well which would include whatever intermediary circuitry would be required like DAC/ADC so that further gives me flexibility on choice. The main hurdle will be the size thing since I still need to cram everything into the speaker.
Sigmastudio is confined to the SigmaDSP and Sharc processor ecosystem. If memory serves you can run it on a PI, but that takes a little fiddling around with to do.
Yes I've been noticing that. Seems like aliexpress is the only place to get a reasonably priced board.
I'm more than likely going to have an external DAC as well. While yes that may seem a little counterintuitive, the only processing I want to do for the speaker is mostly things like noise shaping and crossover. Anything else like room correction would be done by my PC to which my setup will more than likely be hooked up to. Now I COULD develop a USB controller so I could just plug directly into the speakers, but writing a driver is a bear and tbh I just don't want to deal with it. Additionally, it would be preferable to have a set it and forget it solution for just the speakers themselves. Booting into a PC environment every time I want to use the speaker is a little much imo.
This is probably my biggest concern. While software DSP has come a long way, it's still finicky. I'm using RAAL tweeters so a volume surge could kill them and that's not a fun part to fix lol
So dedicated hardware definitely seems to be a recurring theme here. That's not a terribly hard issue to figure out considering how many types of SBCs and processor boards there are. I'm building the amps and a carrier board for the DSP (whichever that may be) as well which would include whatever intermediary circuitry would be required like DAC/ADC so that further gives me flexibility on choice. The main hurdle will be the size thing since I still need to cram everything into the speaker.
Maybe only to program sigma DSP's, but not running it as a DSP on a rasp pi.
Don't see the downside of the ecosystem?
It's just a result of?
But I totally agree that most DSP implementations lack good GUI that's easy to understand for most people.
So true.
And another thing i like about q-sys, is that the open-architecture-processor, schematic building GUI, is very straightforward with great help files.
And, after the design schematic is built, you can them build ANY GUI remote you like....easily, with what you want in it.
This is where my hopes are:
https://www.diyaudio.com/community/threads/2x150w-amp-module-for-sale.388868/post-7553765
SOTA DAC plus DSP options that are not theft are inexistent. The pricing is unacceptable. So let us hope that Eric will help us. I want to go active cardioid FIR.
https://www.diyaudio.com/community/threads/2x150w-amp-module-for-sale.388868/post-7553765
SOTA DAC plus DSP options that are not theft are inexistent. The pricing is unacceptable. So let us hope that Eric will help us. I want to go active cardioid FIR.
While I use multiple PCs in my studio/lab, I'm increasingly uninterested in having a PC be essential for my LR speakers, or to mate a PC to every speaker/audio system I build. This is where miniDSP is head and shoulders above the rest. Starting with the 2x4 HD, the interface is intuitive enough, logical enough, and pretty complete. With the Flex series, DAC/audio performance is excellent, good enough to be a non-factor -- 99% of us can't hear the difference between these and "state of the art" DACs. Flex 8 & its variants really have no peers especially at the asking prices.
@sheeple -- I really like Eric's amps & the great pricing, but DSP is a completely different realm than amplifiers. If Hypex with its presumably extensive resources can't develop user-friendly crossover DSP for its plate amps, I don't have high hopes for DSP from an independent DIYer with no evident skill in software development.
Last edited: