• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

New low-cost DSP platform in development

bachatero

Senior Member
Forum Donor
Joined
Dec 25, 2023
Messages
420
Likes
420
I'm getting fed up with how there are absolutely zero (0) options in the market for a cheap and simple DSP device that can just apply whatever effect I want and be done with it. So, I'm starting development of the software base of what may in the future become a complete DSP system for audio, and also other things related to that.

Here are the gory details:
  • It will run on cheap and newly available RISC-V platforms like this one: https://milkv.io/duo-s
  • The software will support any DSP effect in the special language derived from RISC-V assembly, not just the baby ones mandated for things like miniDSP nor the bespoke graphical ones for Sigma boards
  • The software will be free and open-source and the hardware cheap, like really cheap (compared to the competition like miniDSP)
  • The overall experience might be similar to HiFiBerry, but with less emphasis on music playback and more on raw DSP
  • The first available hardware could be the Duo S board above just connected to a couple 3.5mm jacks using the built in ADC/DAC
  • It currently only barely exists in a GitHub repo that I won't share right now, but is out there if you know what to search for
  • Did I mention that it'll be cheap?
Tell me your thoughts on this idea! I'm not sure what everybody else wants or needs, and right now is the best time to figure that stuff out.
 
Last edited:
The new boards are now powerful enough to do proper DSP at higher sample rates and bit depth but so is a 5th gen i3. If your code is solid then your biggest obstacle will be a GUI across multiple platforms. It doesn't matter if it's a perfect environment for someone who knows how to code or follow a wiki page, simplicity of implementation will win in the end which is why platforms like miniDSP and WiiM do so well.

I've been doing this for years, regularly beta test other companies DSP apps and used to edit audio code for Macs a decade ago so this is just my perspective of it. If you are an active geek/maker type what you propose is exciting but to the average Joe CamillaDSP or the tools within jRiver are mind numbing.
 
That GUI part is already far ahead of the competition since it'll be a simple web app accessible everywhere with the server hosted on each device, while miniDSP needs a bespoke thingie that only runs on Windows/Mac, and all the bare DSP boards are even worse with more bespoke thingies that run on Windows only.
 
I'm all ears then! Perfect approach for a local network DSP as long as your GUI is at least as good as miniDSP.

A quick look at your attached example only shows 2ch I/O which is an immediate limitation unless you can overcome it in software.

More info please, happy to help if I can but mostly Mac based these days.

edit: I used the miniDSP gear back when it was Flash based bulls**t so I totally get where you are coming from.
 
We are locked and loaded for some seriously low-cost R&D!

A quick look at your attached example only shows 2ch I/O which is an immediate limitation unless you can overcome it in software.
Hoping that this will only be true for the cheapest/simplest option but it looks like adding a passable DAC for 2 more channels would only cost a dollar more.
 

Attachments

  • Screenshot from 2024-11-11 10-28-44.png
    Screenshot from 2024-11-11 10-28-44.png
    235.1 KB · Views: 297
Hoping that this will only be true for the cheapest/simplest option but it looks like adding a passable DAC for 2 more channels would only cost a dollar more.
Just note that the onboard codec is only 16 bits, probably very poor bits at that. You will need an external codec for sure. I'm not sure if it can support more than 2 additional channels. The datasheet says it also supports TDM, so any multichannel codec with TDM might just work. It will not be easy to find cheap boards for one of those I think.
 
Just note that the onboard codec is only 16 bits, probably very poor bits at that. You will need an external codec for sure. I'm not sure if it can support more than 2 additional channels. The datasheet says it also supports TDM, so any multichannel codec with TDM might just work. It will not be easy to find cheap boards for one of those I think.
I'll do some testing to see just how bad it is. There is still the situation where you gain the benefits of low cost EQ/IR at the cost of some distortion/noise, which might be a great value proposition depending on your setup (I have one in particular).
 
I'm getting fed up with how there are absolutely zero (0) options in the market for a cheap and simple DSP device that can just apply whatever effect I want and be done with it. So, I'm starting development of the software base of what may in the future become a complete DSP system for audio, and also other things related to that.

Here are the gory details:
  • It will run on cheap and newly available RISC-V platforms like this one: https://milkv.io/duo-s
  • The software will support any DSP effect in the special language derived from RISC-V assembly, not just the baby ones mandated for things like miniDSP nor the bespoke graphical ones for Sigma boards
  • The software will be free and open-source and the hardware cheap, like really cheap (compared to the competition like miniDSP)
  • The overall experience might be similar to HiFiBerry, but with less emphasis on music playback and more on raw DSP
  • The first available hardware could be the Duo S board above just connected to a couple 3.5mm jacks using the built in ADC/DAC
  • It currently only barely exists in a GitHub repo that I won't share right now, but is out there if you know what to search for
Tell me your thoughts on this idea! I'm not sure what everybody else wants or needs, and right now is the best time to figure that stuff out.
Some ideas, you need a long term cybersecurity architecture. These will be IOT connected to the Internet. It would be nice to have the option of using commercial modular plugins. You would probably have to license the VST/AAX/AU interface. Then at home, we could use FabFilter, UA compressors, etc. That means a user interface. Keep us posted.
 
It would be nice to have the option of using commercial modular plugins. You would probably have to license the VST/AAX/AU interface. Then at home, we could use FabFilter, UA compressors, etc.
I hate to disappoint you now but that doesn't appear to be on the table for now. Running x86/ARM plugins on RISC-V would need emulation which would make everything extremely slow, and combine that with the Linux/FreeRTOS platform, and now you get a near zero chance of it being a good experience.

But on the plus side, such a setup could be practical once RISC-V hardware becomes powerful enough to create a higher end model capable of doing emulation both ways.
 
Oh interesting, I'm all ears!
Will this be node based? Cause that way it can both be very simple or extremely powerful :)
 
If you're talking about node.js, then definitely not, since node.js is bloated as heck and would bog down all but the fastest hardware. If you're talking about nodal filter designs, then maybe not, since the filter code itself will look a little like this:
Code:
fa0 *= 2
if fa0 = 100:
  addi t0, zero, 5
  fa0 = t0
fcvt.s.w ft0, a0
The way you string these filters together could be entirely nodal, though, but I haven't gotten to that point yet.
 
Can I suggest a Toslink in/out interface?
In my opinion there are several use cases where it is convenient to simply add a pure DSP on an existing system, without having to go from analog to digital and vice versa.
Many already have an excellent performance DAC and just need a processor in the digital domain, without having to resort to a PC or Mac, or other expensive hardware.
For example, I would find such a DSP convenient to put between the streamer and the DAC. Or even between the TV and the soundbar.
 
Love that you are doing this project, it's sorely needed as you say.

Would also like to see digital in/out on something like this. I am planning on using a WiiM Pro for this functionality (just 10 bands of EQ) but it's overkill for that and even used is hard to find under $100.
 
If you're talking about node.js, then definitely not, since node.js is bloated as heck and would bog down all but the fastest hardware. If you're talking about nodal filter designs, then maybe not, since the filter code itself will look a little like this:
Code:
fa0 *= 2
if fa0 = 100:
  addi t0, zero, 5
  fa0 = t0
fcvt.s.w ft0, a0
The way you string these filters together could be entirely nodal, though, but I haven't gotten to that point yet.
No, I mean as in node based programming, where you can do node networks to program whatever you want to program, something like Sigma Studio. Haven't used this one in particular, but I've been doing lots of graphic stuff with nodes and it really is so nice to work with!

vivaldi_241111-210246.png
 
Would also like to see digital in/out on something like this. I am planning on using a WiiM Pro for this functionality (just 10 bands of EQ) but it's overkill for that and even used is hard to find under $100.
What's interesting is how hard it is to find chips that would allow for this. There is the WM8805 which can do in/out, but it's discontinued but still has tons of boards and stock available. Then there's the DIX4192/PCM9211/DIX9211 which is still in production, but I can't find any dev boards for those outside of the $200 one direct from TI.

But for just digital in, there's a few more options like the DIR9001 which I ordered on Aliexpress so we'll have to see how well that works.
 
If a stereo channel pair is so cheap, there may be a simple way to generate multichannel audio: just use more boards ;) You can feed them with the same SPDIF source, meaning everything will be kept in sync.
 
Stuck in dev and never completly finished. Mine still works like a charm, software side a bit buggy. USB audio via ASIO on Windoof 10 too.

I'd love to see a more modular system with actual sw support. A one-person-show is really hard work.
 
Last edited:
Back
Top Bottom