Looking for latest & greatest CamillaDSP hardware options, inc non-USB input & output, DSP, DAC, etc.

[jeffme]

I'm curious what the latest & greatest hardware options people are using for CamillaDSP, in particular, non-USB input and non-USB output options.

Here's the use case:
- Goal: Multizone / multiroom house-wide audio with per-room room correction
- Plan: WiiM Pro streamers (maybe a 4x Pro in a 1u form factor) -> FIR-capable DSP -> DAC -> Amp channels.
- Constraints:
-- Rack-mount installation, so preferably something that fits in a 1u profile - I can figure out the racking / shelving
-- WiiM because of ethernet, AirPlay 2 (for Apple houseguests), Chromecast (for Android residents & guests), Roon, Spotify Direct, etc. - open to alternatives that meet those needs
-- No USB input for the DSP, as WiiM doesn't have it

Less desirable alternatives for DSP:
- Roon - this would be a great option for the FIR filter engine, but I'd like for any visitor - and for myself - to be able to get the same quality playback from any old app on their phone rather than being dependent on Roon.
- MiniDSP Flex - 4 of these on a 1u shelf would be powerful, but also pretty expensive way to get DSP + DAC
- DSP-based amps - e.g. NAD CI 8-120 DSP, Sonance DSP 8-130 MK II, AudioControl Director D2800, Lexicon DD8+, etc., but they only have PEQ and not FIR options - still, pretty awesome, but I'd like to split out the DSP from the amp
- Oktodac 8 - seems like an awesome piece of kit & has some FIR capabilities, but I couldn't see an obvious way to upload FIR files from REW or HouseCurve

So, the big questions: do people have recommendations for the following:
- Optical / Digital Coax input HATs? (all I can find is: HiFiBerry DAC / DSP - $85) - I think DSP is required here for auto-sensing digital input bitrate - is this correct?
- Any possibility for balanced or even unbalanced output (HiFiBerry DAC2 PRO XLR - $65, with redundant DAC) - if not, can accomplish this with USB output & SMSL-D6 or Topping D10

I've looked through a variety of threads of CamillaDSP, etc., here, but haven't found any focused on a reference platform for non-USB input and output that doesn't involve some sort of audio interface that doesn't scale well space-wise.

 

[MCH]

I like your project. Won't be able to help, but will follow with attention to see how it develops.
Just a couple of comments/questions:
1. Obviously, you want different music playing in the different rooms simultaneously, right? Otherwise a single CamillaDSP + multichannel DAC can take care of the different DSP for different rooms.
2. I believe there is a hifiberry I/O hat that has optical input and output. I recommend you to confirm with someone that owns one if both input and output can work simultaneously well enough
3. I am not sure you can use two i2s hats on the same raspberry pi simultaneously as I understand you are planning to do. I believe a USB DAC will be easier and open more possibilities.
4. The new raspberry pi 5 has 2 multichannel i2s ports. What will be possible to do with that, I don't know, but we might see some new audio hats in the near future.
Hope it helps and good luck!

 

[voodooless]

I plan to use a cheap PCI-e soundcard. The Pi 5 will support PCI-e, and so do some other SBCs, like the OrangePi. That will give me 8 analog channels of ~100 SINAD output, 2 channels of 100+ SINAD input, and stereo SPDIF in/out. It cost me €20 for the card

- Optical / Digital Coax input HATs? (all I can find is: HiFiBerry DAC / DSP - $85) - I think DSP is required here for auto-sensing digital input bitrate - is this correct?

Yes, the data always goes through the DSP, and I don't think that you can remove the ASRC. You can route the data to the pi for processing and back out to the Toslink output. However, you can do a lot with the DSP itself (even program it with SigmaStudio). I've used it and it works pretty well, just don't put it in a tight enclosure with a pi4.. it will burn!

 

[mdsimon2]

Okto dac8 pro could be a good option as it has 4 stereo AES inputs. Okto itself has no FIR capability but in USB / AES mode you can add processing to the digital inputs on a computer (or RPi) running CamillaDSP and output the DSP processed signal via the Okto analog outputs. CamillaDSP accepts a variety of FIR formats so I have no doubt you will be able to use it with whatever FIR generator you want.

The issue with all CamillaDSP solutions using AES / SPDIF inputs is clocking. It is not easy to detect rate changes via AES / SPDIF inputs and even if you are able to detect them you need to restart CamillaDSP at a new sample rate every time the rate changes. It is much simpler if your source produces a constant sample rate so you can run CamillaDSP at a constant sample rate.

I do not use Wiim but it seems like part of the appeal is "bit perfect streaming" that automatically switches sample rates, unfortunately that causes issues with CamillaDSP for the reasons mentioned above. I glanced at the Wiim Pro manual and it looks like you can set a maximum sample rate so you might be able to restrict to single sample rate which would be ideal for a CamillaDSP application.

With the Okto in particular, although it has 4 AES inputs it needs them all to be clock sync'd, so multiple Wiim Pros won't work. The new WiiM CI MOD S4 has 4 digital outputs in a single chassis, but from this video it almost sounds like it may just be 4 Wiim Pros in a single chassis which implies independent clocks. If all of the digital outputs are clock sync'd it would be a good fit for the Okto + CamillaDSP.

HifiBerry DAC+ DSP is nice in that it has an ASRC which can keep a constant sample rate for CamillaDSP, even if the sample rate from the Wiim was changing. Of course, you would need 4 RPis with 4 DAC+ DSP HATs each running CamillaDSP. You would then equip each RPi with a stereo DAC. If you could restrict the sample rate from Wiim you could also use something like the Hifime UR23 instead of the DAC+ DSP as a CamillaDSP capture device, which would be a bit cheaper.

Another ASRC option is the miniDSP U-DIO8, it can act as a multichannel ASRC and has 4 stereo digital inputs and 4 stereo digital outputs. You would use this as your CamillaDSP capture device on a single RPi and could use multiple independent streamers. You could also use this as a CamillaDSP playback device if you wanted to use multiple stereo AES / SPDIF input DACs or could use another 8 channel USB input DAC (Topping DM7, Okto dac8 pro, MOTU Ultralite Mk5, etc) as your CamillaDSP playback device.

Michael

 

[mdsimon2]

As the OP wants to use a toslink to i2s card, would it be possible to have a script that counts the sample rate and changes the Camilladsp config accordingly? Still needs that the spdif card auto detects and adapts to different incoming sample rates, but maybe such cards exist....

I think this is possible but I haven't explored it much. With something like the Hifime UR23 or S2, I've found that CamillaDSP can detect SPDIF rate changes as long as the CamillaDSP sample rate is greater than the incoming sample rate. I've thought about making a script such that when CamillaDSP goes into the STALLED state, it resets to a configuration using 192 kHz and then switches to a configuration of the detected sample rate.

I haven't had much motivation to work on the script as all of my listening these days is done through an AppleTV running at a constant 48 kHz. I also have some concern about the robustness of this approach. Maybe I'll give it a try when I have some free time over the holidays.

Michael

 

[MCH]

I think this is possible but I haven't explored it much. With something like the Hifime UR23 or S2, I've found that CamillaDSP can detect SPDIF rate changes as long as the CamillaDSP sample rate is greater than the incoming sample rate. I've thought about making a script such that when CamillaDSP goes into the STALLED state, it resets to a configuration using 192 kHz and then switches to a configuration of the detected sample rate.

I haven't had much motivation to work on the script as all of my listening these days is done through an AppleTV running at a constant 48 kHz. I also have some concern about the robustness of this approach. Maybe I'll give it a try when I have some free time over the holidays.

Michael

Sorry, deleted my post because thought it was a bit off topic for the OP, but thanks for the insight

 

[voodooless]

I think this is possible but I haven't explored it much.

GitHub - marcoevang/camilladsp-setrate: Automatic sample rate switcher for CamillaDSP

Automatic sample rate switcher for CamillaDSP. Contribute to marcoevang/camilladsp-setrate development by creating an account on GitHub.

github.com

 

[jeffme]

I like your project. Won't be able to help, but will follow with attention to see how it develops.

Great questions! Thanks for helping clarify what I'm after.

Just a couple of comments/questions:
1. Obviously, you want different music playing in the different rooms simultaneously, right? Otherwise a single CamillaDSP + multichannel DAC can take care of the different DSP for different rooms.

Correct - we'd need to be ready for different music in all of the rooms. Kind of an edge case to support every room simultaneously, but we routinely have 3-4 different streams going around the house, either in individual rooms or groups of rooms & which rooms those are in is always changing.

2. I believe there is a hifiberry I/O hat that has optical input and output. I recommend you to confirm with someone that owns one if both input and output can work simultaneously well enough

Yes, there is - the HiFiBerry Digi+ I/O, but that lacks the ability to handle different bit rates of input, according to the Digital Input section of HiFiBerry's note Comparison of HiFiBerry cards for audio recording.

3. I am not sure you can use two i2s hats on the same raspberry pi simultaneously as I understand you are planning to do. I believe a USB DAC will be easier and open more possibilities.

Ah - I was actually imagining using a separate rpi for every room. I'm pretty sure we could fit 4 per 1u rack, to match the 4 WiiMs in the 1u rack.

4. The new raspberry pi 5 has 2 multichannel i2s ports. What will be possible to do with that, I don't know, but we might see some new audio hats in the near future.
Hope it helps and good luck!

Rpi5 is definitely interesting & I'm open to all options.

 

[mdsimon2]

GitHub - marcoevang/camilladsp-setrate: Automatic sample rate switcher for CamillaDSP

Automatic sample rate switcher for CamillaDSP. Contribute to marcoevang/camilladsp-setrate development by creating an account on GitHub.

github.com

From previous discussion with @phofman, it doesn't seem like most SPDIF input devices report rate changes via ALSA. There was discussion on DIY Audio related to this project about using the UR23 -> https://www.diyaudio.com/community/threads/sample-rate-switcher-for-camilladsp.403054/ and it sounded like a no-go.

Although it seems like Speaky started working on a separate project for the UR23.

Michael

 

[jeffme]

I plan to use a cheap PCI-e soundcard. The Pi 5 will support PCI-e, and so do some other SBCs, like the OrangePi. That will give me 8 analog channels of ~100 SINAD output, 2 channels of 100+ SINAD input, and stereo SPDIF in/out. It cost me €20 for the card

Now that's cool. No balanced output, but pretty slick. I only need 2 channels out per room (L/R) and a max of 4 channels of output for 1, maybe 2 rooms (L/R/sub1/sub2), but still, very cool. No external DAC.

Yes, the data always goes through the DSP, and I don't think that you can remove the ASRC. You can route the data to the pi for processing and back out to the Toslink output. However, you can do a lot with the DSP itself (even program it with SigmaStudio). I've used it and it works pretty well, just don't put it in a tight enclosure with a pi4.. it will burn!

Yes... the thermals would need some investigation & I do want to try to keep it in 1u. Fans may be in order, but I'm hoping a single RPi handling a single input stream won't get too too hot... is that a fool's hope?

 

[voodooless]

I'm hoping a single RPi handling a single input stream won't get too too hot... is that a fool's hope?

I think it should work. In my case, I had the pi4 + DSP in a simple enclosure made for the combo, and no heatsink on the CPU. The whole thing was in a cupboard with other electronics generating heat. I think with some airflow and a heatsink disaster could have been avoided...

 

[MCH]

LYes, there is - the HiFiBerry Digi+ I/O, but that lacks the ability to handle different bit rates of input, according to the Digital Input section of HiFiBerry's note Comparison of HiFiBerry cards for audio recording.

Ah ok good to know

Ah - I was actually imagining using a separate rpi for every room. I'm pretty sure we could fit 4 per 1u rack, to match the 4 WiiMs in the 1u rack.

I was referring to one spdif to i2s input hat + one i2s DAC hat per RPI. I understood this is what you wanted to do with each of the 4 RPI.

Now that's cool. No balanced output, but pretty slick. I only need 2 channels out per room (L/R) and a max of 4 channels of output for 1, maybe 2 rooms (L/R/sub1/sub2), but still, very cool. No external DAC.

I don't see how you can output 4 different stereo streams to one single multichannel DAC using CamillaDSP, hence my question 1. But if it is possible, I would love to know how

 

[voodooless]

I don't see how you can output 4 different stereo streams to one single multichannel DAC using CamillaDSP, hence my question 1. But if it is possible, I would love to know how

I think you should be able to route the various inputs to the various outputs. You’ll just need to resample the inputs to the same rate.

 

[jeffme]

I do not use Wiim but it seems like part of the appeal is "bit perfect streaming" that automatically switches sample rates, unfortunately that causes issues with CamillaDSP for the reasons mentioned above. I glanced at the Wiim Pro manual and it looks like you can set a maximum sample rate so you might be able to restrict to single sample rate which would be ideal for a CamillaDSP application.

Michael - first... wow! Your Camilla + Rpi thread is fantastic. I was hoping you might stop by & provide just the type of feedback you have. Many thanks,.

I had thought about the standardized bit rate thing, but that would end up being a reasonably high lowest common denominator thing and I'm wondering what the various transports delivering data to WiiMs might do below that threshold. Aren't some of those transports variable rates? I don't think the WiiM would then upsample to that higher, but lowest common denominator. I need to do some homework on this front.

With the Okto in particular, although it has 4 AES inputs it needs them all to be clock sync'd, so multiple Wiim Pros won't work.

Plus, I'd need to convert the digital output of the Wiims to AES, which seems like another hoop to jump through.

The new WiiM CI MOD S4 has 4 digital outputs in a single chassis, but from this video it almost sounds like it may just be 4 Wiim Pros in a single chassis which implies independent clocks. If all of the digital outputs are clock sync'd it would be a good fit for the Okto + CamillaDSP.

So, I've traded email with a non-tech type at Origin Acoustics. They claim to be upgrading some parts of the WiiM setup, like the DAC (beyond the Pro Plus DAC??), but I'd be super surprised if they are putting them all on the same clock. We'll see. They suggested tech specs would be available in Jan. Maybe.

HifiBerry DAC+ DSP is nice in that it has an ASRC which can keep a constant sample rate for CamillaDSP, even if the sample rate from the Wiim was changing. Of course, you would need 4 RPis with 4 DAC+ DSP HATs each running CamillaDSP. You would then equip each RPi with a stereo DAC. If you could restrict the sample rate from Wiim you could also use something like the Hifime UR23 instead of the DAC+ DSP as a CamillaDSP capture device, which would be a bit cheaper.

Thanks - that's very helpful - I'd forgotten the Hifime angle. Very cool.

Another ASRC option is the miniDSP U-DIO8, it can act as a multichannel ASRC and has 4 stereo digital inputs and 4 stereo digital outputs. You would use this as your CamillaDSP capture device on a single RPi and could use multiple independent streamers. You could also use this as a CamillaDSP playback device if you wanted to use multiple stereo AES / SPDIF input DACs or could use another 8 channel USB input DAC (Topping DM7, Okto dac8 pro, MOTU Ultralite Mk5, etc) as your CamillaDSP playback device.

Michael

The U-DIO8 looks very interesting, but I'm not sure I'm following here. How would the streamers get their audio to the U-DIO8? It looks like this would still require some sort of digital coax or optical converter to AES or XLR. The actual cabling part of many of these options is where my imagination fails me.

Again, many thanks!!

 

[MCH]

I think you should be able to route the various inputs to the various outputs. You’ll just need to resample the inputs to the same rate.

Well, then he will need an udio8 kind of thing for the inputs. Yes probably still cheaper than quadrupling a full stereo setup

 

[jeffme]

Seems to me we just need to convince WiiM to provide USB output or add convolution support to the Pro Plus or something...

 

[mdsimon2]

Plus, I'd need to convert the digital output of the Wiims to AES, which seems like another hoop to jump through.

Majority of the time simple RCA to XLR cables work fine, something like this -> https://www.monoprice.com/product?p_id=4777.

If you want to be anal about it, you can use an RCA to BNC cable like this -> https://www.bhphotovideo.com/c/product/1053962-REG/kopul_vrbc_406_bnc_male_to_rca.html into a BNC to AES transformer like this -> https://www.bhphotovideo.com/c/product/367655-REG/Canare_BCJXPTRB_BCJ_XP_TRB_75_Ohm_to.html. Note, these are available used for a fraction of the price, I purchased four of them for $30 shipped on ebay.

The U-DIO8 looks very interesting, but I'm not sure I'm following here. How would the streamers get their audio to the U-DIO8? It looks like this would still require some sort of digital coax or optical converter to AES or XLR. The actual cabling part of many of these options is where my imagination fails me.

It has four digital stereo inputs and four digital outputs, you can either get it with RCA, BNC or XLR cabling, again converting between each is easy. Each streamer would be connected to a digital input. It then acts as an 8 channel USB input device for CamillaDSP. Because it has an ASRC it does not matter if all of the streamers have independent clocks, it will convert the inputs to a common clock. You then can process the audio in single instance of CamillaDSP.

Michael

 

[jeffme]

Ok... *now* we're getting somewhere... :fire:

So, let's see about the BOM:
- 4x RCA->BNC (4x $8 (1.5ft)) = $32
- 4x BNC->AES (ebay) = $30
- UDIO for clock syncing = $400
- 1x RPi = $45 (ish)
- OktoDAC Pro 8 = $1400

Total = ~$1900 + case or ~ $475 per channel & there'd still be at least 2u for the OktoDAC

This would be awesome & totally top notch.

But... why not skip the OktoDAC & get 4x SMSL D6 for about $800 less?

Or, just do 4x Rpi + HiFiBerry DAC/DSP + SMSL D6?

Is it a problem if the separate channels / rooms are on separate clocks?

(and apologies for these questions, which might seem a little circular - this thread has been super helpful in both making me smarter & also confirming that there really aren't any magic hw solutions I've been missing)

 

[ppataki]

Or, just do 4x Rpi + HiFiBerry DAC/DSP + SMSL D6?

If you need 4 different signals in 4 different rooms then this is what I would do if I were you
and instead of the D6 you could use the new D-6S:

SMSL D-6S / D6S ES9039Q2M DAC MQA Bluetooth 5.1 High-Res Audio Decoder

Buy SMSL D-6S / D6S ES9039Q2M DAC MQA Bluetooth 5.1 High-Res Audio Decoder with Free shipping on HiFiGo, you can enjoy a year warranty and the best after-sales service.

hifigo.com

 

[AlfaNovember]

Forgive me if I've misunderstood the design brief, but it seems to me that Roon driving a fleet of RPi/HifiBerry I2C DAC units running Ropieee gets you everything you've asked for except DSP correction of Airplay / Spotify guests. Ropieee offers the streaming protocols alongside being a Roon endpoint, but can't do the DSP. Admittedly, the hardware is not especially luxe looking and the Hifiberry sinad numbers aren't especially bragworthy, but it works quite nicely.

That's basically what I do; the only bit of friction is handing off volume levels between Roon and Shairplay; the underlying Alsa process retains the last volume setting in roon as the top of the scale for Shairplay. So in practice, I stop the music in Roon, bump the zone volume up, quit Roon, and then connect to the Airplay endpoint with iphone volume low, and adjust to taste.

With regard to clocking the whole system - It doesn't always get the marketing attention, but Roon's RAAT protocol handles network sync between zones quietly and capably. That puts the music rendering endpoints on the end of Cat6 cables (or wifi) rather than running digital or analog audio around the building on dedicated wiring. Each device adjust its digital clocking independently and imperceptibly. I gather there are non-IP standards for passing digital audio over cat6 in the professional studio world, but that's out of scope for residential use.

Finally - Be mindful of the latency for the DSP processing for each room. I'm imagining 4 different DSP correction filters might end up having different lengths / latencies, and that might be tricky to reconcile. I mention it because I've got a bunch of pi/hat stacks scattered about, but my main rig runs CamillaDSP for crossover & correction. I can't group it to other zones in Roon because the Camilla process is 'after' Roon, and so can't be automatically kept in sync. It sounds just delayed enough to be unlistenable. All the other basic realtime DAC devices group together just fine.