CamillaDSP - How are you using it?

[Snoopy8]

26 minutes ago, Grizaudio said:

I’m curious how I would implement this for Roon bridge, Camilla, Ubuntu on my Pi? Any helpful ideas. 

Use this SUSE Linux doc to learn and experiment

https://documentation.suse.com/sle-rt/12-SP5/html/SLE-RT-all/book-slert-shielding.html

Given your heavy CamillaDSP usage, you may want to dedicate 2 CPUs to it and maybe Roon Bridge with the other?

 

Configure TOP to show processes CPU usage and on which CPU.

 

[Grizaudio]

10 minutes ago, Snoopy8 said:

Use this SUSE Linux doc to learn and experiment

https://documentation.suse.com/sle-rt/12-SP5/html/SLE-RT-all/book-slert-shielding.html

Given your heavy CamillaDSP usage, you may want to dedicate 2 CPUs to it and maybe Roon Bridge with the other?

 

Configure TOP to show processes CPU usage and on which CPU.

 

Thanks Snoopy. 
I’ve calmed my tapcount down a lot using CamillaDSP. 
Just running crossover filters now, so CPU usage is quite modest. 
 

I’ll take a look, thanks. 
steve 

[Grizaudio]

Hi all, 

 

I’m considering grabbing a new Khadas Edge v2 to use as my DSP and streamer endpoint. if this is not a great idea the new N100 from Intel looks perfect. 
 

Does anyone have any experience with Armbian and CamillaDSP….

https://www.armbian.com/khadas-edge-2/
The Edge is a beast and looks great for this application. 

EDIT……..

Looks great, but the Armbian kernel is not recent enough for my Motu…. Bummer. Looking into the n100 further. 

Edited July 14, 2023 by Grizaudio

[tripitaka]

On 12/07/2023 at 5:42 PM, Grizaudio said:

I’m out atm, so a short and sweet message…

 

Your FIRs need to match the sampling rate set in CamillaDSP….

 

 

 

 

Thanks Griz, I don't recall ever setting the Camilla sampling rate, I just assumed it was 48, maybe I missed a trick somewhere in there.

For that matter I cant recall an option to do so either, so maybe there are different levels of functionality for PC VS minicomputer setups.

[Grizaudio]

5 minutes ago, tripitaka said:

 

Thanks Griz, I don't recall ever setting the Camilla sampling rate, I just assumed it was 48, maybe I missed a trick somewhere in there.

For that matter I cant recall an option to do so either, so maybe there are different levels of functionality for PC VS minicomputer setups.

 

You need to select a sample rate for operation. Unless you are using scribble, sampling rate is fixed.

So you will either need to feed your loopback into CamillaDSP @ the nominated sampling rate, or resample loopback using CamillaDSP. 

[tripitaka]

1 minute ago, Grizaudio said:

 

You need to select a sample rate for operation. Unless you are using scribble, sampling rate is fixed.

So you will either need to feed your loopback into CamillaDSP @ the nominated sampling rate, or resample loopback using CamillaDSP. 

Advice much appreciated, Griz

 

I'll get back and try again sometime, there are just so many ways to fail in this game!

 

[Grizaudio]

17 minutes ago, tripitaka said:

Advice much appreciated, Griz

 

I'll get back and try again sometime, there are just so many ways to fail in this game!

 

If you heard music - you couldn't have been that far wrong. 

I suspect you had resampling selected, and were resampling all content to 48khz? 

 

In my setup, I resample/upsample to 192khz using Roon, and have CamillaDSP setup for 192khz in USB streamer mode. 

[tripitaka]

1 minute ago, Grizaudio said:

If you heard music - you couldn't have been that far wrong. 

I suspect you had resampling selected, and were resampling all content to 48khz? 

 

In my setup, I resample/upsample to 192khz using Roon, and have CamillaDSP setup for 192khz in USB streamer mode. 

I dare say you are right on that.

 

Anyway, a good thread you have started here,  useful for those who are interested in getting their hands dirty with software tweaks or just looking out for economical paths forward.

[tuyen]

Thanks for the info so far about this solution. 

 

I must give it a go as I'm only using Roon on a windows PC at present as my only source.  Also I have an unused usb multichannel DAC that has volume control (exasound e18) .    Curious to see if brings any sonic benefits over traditional hardware based DSP solutions I have tried (ie. MiniDSP/DEQX/Najda)

[Grizaudio]

1 hour ago, tuyen said:

Thanks for the info so far about this solution. 

 

I must give it a go as I'm only using Roon on a windows PC at present as my only source.  Also I have an unused usb multichannel DAC that has volume control (exasound e18) .    Curious to see if brings any sonic benefits over traditional hardware based DSP solutions I have tried (ie. MiniDSP/DEQX/Najda)

Roon will also do multichannel IIR and FIR. 
something to explore. 
 

If you do try CamillDSP please post your experience. 
 

Edited July 21, 2023 by Grizaudio

[Snoopy8]

On 12/07/2023 at 9:56 AM, tripitaka said:

I had a brief flirtation with CamillaDSP on an RPi4, but must have been something wrong since it didn't sound great, I'll revisit it at some stage. But can I ask the brains trust, how you are formatting the FIR impulses you are uploading, eg, 32bit 48kHzWAV file?

If you load moOde, enabling CamillaDSP is a simple as uploading the Wav file and selecting "quick convolution".  The developer has done a nice job of integrating CamillaDSP and simplifying access.

[tuyen]

Is there anyone in Perth who is a bit of a whiz with setup of CamillaDSP on a windows os and would be willing to give me a hand?

 

I attempted to start creating a config file, but got a bit lost and being so time restricted, have put it on hold. 

[Snoopy8]

3 hours ago, tuyen said:

Is there anyone in Perth who is a bit of a whiz with setup of CamillaDSP on a windows os and would be willing to give me a hand?

 

I attempted to start creating a config file, but got a bit lost and being so time restricted, have put it on hold. 

If you have already set it up and got it running, the config file is relatively easy.  Suggest you post your config questions here...

[tuyen]

Don’t know if I’ve even got it setup properly to be honest. Can’t access the web interface either. Got confused at the websocket/using anaconda… 

 

Hoping also to pick the brain of someone who cluey with doing a REW sweep and then entering in a target curve and generating some appropriate filters into CamillaDSP too.     With Deqx, it’s all wizard driven so I could follow it.  But my hope is to move away from Deqx and just use Nuc PC as DSP into the Exasound 8ch dac. 

Edited August 12, 2023 by tuyen

[Snoopy8]

1 hour ago, tuyen said:

Don’t know if I’ve even got it setup properly to be honest. Can’t access the web interface either. Got confused at the websocket/using anaconda… 

 

Hoping also to pick the brain of someone who cluey with doing a REW sweep and then entering in a target curve and generating some appropriate filters into CamillaDSP too.     With Deqx, it’s all wizard driven so I could follow it.  But my hope is to move away from Deqx and just use Nuc PC as DSP into the Exasound 8ch dac. 

What are hoping to achieve with REW and CamillaDSP that cannot be delivered by DEQX ? 

 

With a 8 Ch DAC, you will be in for a very steep learning curve and a lot of pain.  And likely, REW cannot do the job and you will need software like Acourate or Audiolense.

 

I suspect you will be better off staying with DEQX ?

[Grizaudio]

2 hours ago, Snoopy8 said:

What are hoping to achieve with REW and CamillaDSP that cannot be delivered by DEQX ? 

 

With a 8 Ch DAC, you will be in for a very steep learning curve and a lot of pain.  And likely, REW cannot do the job and you will need software like Acourate or Audiolense.

 

I suspect you will be better off staying with DEQX ?

 

I would agree with @Snoopy8 Sorry I can't help with a windows setup. 

 

Alternatively, if you can pivot to using a raspberry pi, the guide I posted at the start of this thread will get you working. 

I suspect you would hear a difference between USB/DAC's and DEQX, if you have the s88 that's a cracking DAC/streamer. 

 

Camilla and REW is a great combo IMO. There's plenty that can be achieved with some know how. 

 

If you have a moment refer to the Pi guide. The instructor has created three configs for different DACs. 

The Motu Ultralite and Okto configs would be helpful config guides for you.

Edited August 12, 2023 by Grizaudio

[tuyen]

Gotta stick to windows as the Exasound e18 8ch dac only supports windows/macOS.  Else i would go down the rpi4 path.     
 

Was just hoping to get some sq improvements by simplifying/ removing the Deqx from the Audio path as my only source is digital via roon which is running on the intel Nuc as well.   The Exasound dac has built in volume controller and had a thought it’s overall sq would be of higher performance (also supports 32/384) than the dacs in the deqx hdp3 (and 24/96 max input). 
 

I’ll keep playing around and hopefully figure it out or perhaps someone might eventually post a windows guide for CamillaDSP online. All fun!

[Keith_W]

Going from a DEQX at 24/96 to an Exasound 32/384 will only give you a marginal SQ improvement. Verrrry marginal compared to the difference that can be made by say, a better designed filter. 

 

The real advantage of going to Exasound + RPi + Camilla (or some other convolver) is that it opens the door for more powerful DSP software. This will allow you to do more corrections at a finer resolution. You are no longer tied to DEQX software, you can use anything that you like. There are free options like REW + Rephase, or OpenDRC. Then there are paid options like Acourate, Audiolense, and Dirac. And you don't have to use CamillaDSP if you find it difficult, you could also use EqualizerAPO (which is also free). Then there are all the other paid options for convolvers - too many to list. If you want to discuss options for third party convolvers, just ask the question. You could even use HQPlayer and upsample everything to whatever is the maximum your DAC will support, including DSD. The only limit is how much CPU you have and how much heat and fan noise you are willing to tolerate. 

 

Don't get me wrong, the DEQX does a fantastic job. But I am not a fan of hardware convolvers / processors like the DEQX (and MiniDSP) because of said software limitations, low resolution correction, and because they are not modular. If you choose DEQX, you are stuck with their hardware and software. If you choose to go separates, you can bring along any DAC you like and any software you like. The advantage of DEQX is convenience and their fantastic support, really great for helping people who are unfamiliar with DSP. The advantage of going separates is processing power, resolution, and flexibility - at the cost of a high learning curve. 

[Grizaudio]

I came across this script, and thought it was worth sharing...

https://github.com/JWahle/piCoreCDSP

 

"piCoreCDSP

The goal of this project is to provide an easy way to install CamillaDSP 1.0.3 including GUI and automatic samplerate switching on a piCorePlayer installation."

 

Seems like a great solution for those using squeezelite and wanting a player/DSP solution that auto sample rate changes. 

 

Edited October 21, 2023 by Grizaudio

[oltos]

On 12/08/2023 at 12:44 AM, Snoopy8 said:

What are hoping to achieve with REW and CamillaDSP that cannot be delivered by DEQX ? 

 

With a 8 Ch DAC, you will be in for a very steep learning curve and a lot of pain.  And likely, REW cannot do the job and you will need software like Acourate or Audiolense.

 

I suspect you will be better off staying with DEQX ?

 

Oh boy, this doesn't look so invitingly easy. I have zero hands on experience and knowledge of any room correcting software (eg. DIRAC Live, REW, RePhase, Audiolense, Acourate). Like you mentioned earlier, I would be using Camilla for creating convolution filters. My main speakers (~ 97 db/watt/meter) are passively crossed and I have four Rythmik F12 powered subs. And I am a Windows only user. My plan was to use Camilla with an 8 channel AD/DA interface like Merging Hapi or Prism Titan. But how else would room correction be best, if not only achievable, with my hardware without using an 8 channel interface?

 

In any case, rather than break my small newbie brain over Camilla, what might be the next best room correction software for me? 

 

[davewantsmoore]

21 minutes ago, oltos said:

In any case, rather than break my small newbie brain over Camilla, what might be the next best room correction software for me? 

Dirac.

It helps you through the process of correction.

The others are either not strictly "room correction", or are tools which "provide all the options with no guidance" (and your result will be determined by your knowledge/skill).

 

 

If you are happy with the crossover between your mains and subs now, then you can run the correction in stereo (ie. "full range").... and not have to worry about na interface with > 2 channels.

Edited November 29, 2023 by davewantsmoore

[davewantsmoore]

On 12/8/2023 at 6:26 PM, Keith_W said:

difference that can be made by say, a better designed filter. 

This.

 

but it is not so much "better filters" (ie. more "technically powerful modern advanced, etc") .... but just the "right" filter. ie . the filters that gets you the frequency response you want and a correct time response (which doesn't have to be anything "special").

 

It is as easy as falling of a log to have crossovers which don't sum, or phase (impulse, step, etc) responses which are whack.   The paradox being they may all sound different, and different => preference .... but not necessary a well implemented speaker or correction (ie. not actually the best sound).

[Keith_W]

6 hours ago, oltos said:

 

Oh boy, this doesn't look so invitingly easy. I have zero hands on experience and knowledge of any room correcting software (eg. DIRAC Live, REW, RePhase, Audiolense, Acourate). Like you mentioned earlier, I would be using Camilla for creating convolution filters. My main speakers (~ 97 db/watt/meter) are passively crossed and I have four Rythmik F12 powered subs. And I am a Windows only user. My plan was to use Camilla with an 8 channel AD/DA interface like Merging Hapi or Prism Titan. But how else would room correction be best, if not only achievable, with my hardware without using an 8 channel interface?

 

In any case, rather than break my small newbie brain over Camilla, what might be the next best room correction software for me? 

 

 

I should point out that Camilla can not be used to create convolution filters. Camilla itself is a convolver, i.e. it convolves the convolution filters with music signal for output. You need a separate program to create the filters, and I listed them in my previous post: REW (free), RePhase (free), OpenDRC (free), and the paid options - Acourate, Audiolense, and Dirac. If you are not planning to convert your speakers to an active, then there is also MSO (free). 

 

If you want to control 2 speakers with 4 subwoofers, you will need 6 DAC channels with amplifiers for each speaker/sub (I assume that you already have these amps). There is no way around this, no matter what solution you choose, you will need 6 DAC channels as a minimum. 

 

Alternative convolvers to Camilla: JRiver, Roon, Acourate Convolver, Hang Loose Convolver, EKIO, Audiolense Convolver, and EqualizerAPO (free). 

Room correction software: Acourate, Audiolense, Dirac, RePhase, OpenDRC. Or if using subs only, MSO (Multi-Sub Optimizer). 

 

Acourate is the most powerful and most versatile, and it is also the cheapest of the paid options. But it is difficult to use because everything is manual and requires interpretation. Frankly I think Acourate is overkill for what you want to do and I do not recommend it. 

 

Audiolense is probably the easiest to use and the fastest way to get results because of the high degree of automation. It does not offer the same degree of control or the flexibility of Acourate, but I do not think that you need it. 

 

Dirac, RePhase, and OpenDRC - I haven't used them and I do not have direct experience with any of them. I can't comment on ease of use and flexibility, but I understand that Dirac is easy to use. The downside of Dirac is that it outputs proprietary filters which must be hosted on Dirac's own convolver. I have a thing against proprietary anything, which is why I wrote it off without trying it. The other downside is that it is rather expensive. 

 

MSO only works for integrating multiple subs with speakers and can not do any of the other advanced DSP functions, but that is OK because it looks as if you are not planning to incorporate it. It is free and must be used with REW. You take measurements with REW, then import the measurements into MSO. MSO then outputs a filter which you host on a convolver of your choice (I listed some options above). MSO has an extensive manual, an extensive walkthrough, and a community of people happy to help (like our own @Snoopy8). I think this is probably the best solution for you. 

BTW, I think that you should consider a Merging Hapi IF you are planning to bypass the passive XO's on your speakers and convert them to active. Merging Hapi can be optioned up to 16 DAC channels. 

[aris]

Camilla does support a range of biquad based XO options for active setups. 
 

You can use CamillaDSP very nicely without FIR for active, inc gain, delay, and allpass filters etc. 
 

And you can adjust bass without using FIR if you want as well. Some room nodes that may be sufficient. 

Edited November 30, 2023 by aris

[oltos]

On 12/08/2023 at 3:53 AM, tuyen said:

Gotta stick to windows as the Exasound e18 8ch dac only supports windows/macOS.  Else i would go down the rpi4 path.     
 

Was just hoping to get some sq improvements by simplifying/ removing the Deqx from the Audio path as my only source is digital via roon which is running on the intel Nuc as well.   The Exasound dac has built in volume controller and had a thought it’s overall sq would be of higher performance (also supports 32/384) than the dacs in the deqx hdp3 (and 24/96 max input). 
 

I’ll keep playing around and hopefully figure it out or perhaps someone might eventually post a windows guide for CamillaDSP online. All fun!

Don't you mean the Exasound e38? And is yours the MK2 with balanced outputs?