Hardware for high quality software based DSP

[JeroenD]

Hello, I am new to this forum but reading already for quite some time interesting questions and answers.

Over the years I developed myself a nice audio system which I really enjoy (system components listed below). I did investigate what is the best option for me to improve my system and I think replacing the analog active filter I use by a DSP based system would be a nice next step. I want to move to DSP based crossover, phase correction, time alignment, room correction and have some questions related. I think PC software based DSP is the best way for me to go for.

I am planning to use Acourate for measurement and creating the filters.

At the moment I am using Roon and I did read Roon can handle these filters so that could be interesting for me. Otherwise JRiver is one of the best option as far as I understood.

I now try to make the right design decisions regarding the hardware and interfaces to use.

Question 1: I am thinking of using an fanless NUC with i5 or i7 and 8GB RAM. Any advice on this?

Question 2: I am planning to add 2 more Metrum DAC’s to my system to create a DSP based 3-way system. What is the best option to connect these DAC’s to the NUC? Use the USB port of the Metrum and connect 3 USB cables from NUC to the 3 Metrum DAC’s? I have the feeling this can generate timing (clock) problems. But is this more theoretical are also audible?

Or is it better to add an audio card to the NUC with for example 3x AES/EBU output. Or is it better to use something like “XMOS USB Digital Interface 32/384khz AES EBU” like the https://www.audiophonics.fr/en/digital-interfaces/xmos-usb-digital-interface-32384khz-aes-ebu-tcxo-alimentation-230v-silver-p-11507.html or https://www.audiophonics.fr/en/digital-interfaces/singxer-su-2-usb-digital-interface-32bit-768khz-dsd1024-spdif-aesebu-i2s-hdmi-lvds-p-14270.html ?

Or do you advice a different way to connect the DSP NUC to the DAC’s?

 

In short my current audio system:

·         Magnepan MG20.1 (fully modified, silver foil wiring, no internal crossover, all external with high quality components)

·         2x Velodyne DD12 (I am not using the bass panel of the MG20.1 because it has a negative influence on the mid panel and tweeter. The 2x DD12 replaces the bass panels and using the internal DSP I already improved the low level frequency domain in combination with my room.)

·         Metrum Ambre source

·         Metrum Adagio DAC (volume control normally not used)

·         Audio Research LS26 (modified with ODAM caps)

·         Magnepan active analog filter (crossover between bass and mid/high)

·         Krell KSA 300S (upgraded version) and Audio Research VS115 (I use 1 of both to power the MG20.1. The Krell is better in controlling the mid panel and the VS115 creates a nicer and bigger soundstage with the big ribbon tweeter)

·         Interlinks Transparent Balanced reference and speaker wires self made using Mundorf 44mm foil and Teflon isolator

[The Computer Audiophile]

2 minutes ago, JeroenD said:

Hello, I am new to this forum but reading already for quite some time interesting questions and answers.

Over the years I developed myself a nice audio system which I really enjoy (system components listed below). I did investigate what is the best option for me to improve my system and I think replacing the analog active filter I use by a DSP based system would be a nice next step. I want to move to DSP based crossover, phase correction, time alignment, room correction and have some questions related. I think PC software based DSP is the best way for me to go for.

I am planning to use Acourate for measurement and creating the filters.

At the moment I am using Roon and I did read Roon can handle these filters so that could be interesting for me. Otherwise JRiver is one of the best option as far as I understood.

I now try to make the right design decisions regarding the hardware and interfaces to use.

Question 1: I am thinking of using an fanless NUC with i5 or i7 and 8GB RAM. Any advice on this?

Question 2: I am planning to add 2 more Metrum DAC’s to my system to create a DSP based 3-way system. What is the best option to connect these DAC’s to the NUC? Use the USB port of the Metrum and connect 3 USB cables from NUC to the 3 Metrum DAC’s? I have the feeling this can generate timing (clock) problems. But is this more theoretical are also audible?

Or is it better to add an audio card to the NUC with for example 3x AES/EBU output. Or is it better to use something like “XMOS USB Digital Interface 32/384khz AES EBU” like the https://www.audiophonics.fr/en/digital-interfaces/xmos-usb-digital-interface-32384khz-aes-ebu-tcxo-alimentation-230v-silver-p-11507.html or https://www.audiophonics.fr/en/digital-interfaces/singxer-su-2-usb-digital-interface-32bit-768khz-dsd1024-spdif-aesebu-i2s-hdmi-lvds-p-14270.html ?

Or do you advice a different way to connect the DSP NUC to the DAC’s?

 

In short my current audio system:

·         Magnepan MG20.1 (fully modified, silver foil wiring, no internal crossover, all external with high quality components)

·         2x Velodyne DD12 (I am not using the bass panel of the MG20.1 because it has a negative influence on the mid panel and tweeter. The 2x DD12 replaces the bass panels and using the internal DSP I already improved the low level frequency domain in combination with my room.)

·         Metrum Ambre source

·         Metrum Adagio DAC (volume control normally not used)

·         Audio Research LS26 (modified with ODAM caps)

·         Magnepan active analog filter (crossover between bass and mid/high)

·         Krell KSA 300S (upgraded version) and Audio Research VS115 (I use 1 of both to power the MG20.1. The Krell is better in controlling the mid panel and the VS115 creates a nicer and bigger soundstage with the big ribbon tweeter)

·         Interlinks Transparent Balanced reference and speaker wires self made using Mundorf 44mm foil and Teflon isolator

Hi Jeroen, welkome to Audiophile Style. I like what you're trying to do. It sounds like a neat project with hopefully a good outcome. 

 

The filters I use were created using Audiolense, by @mitchco. I use them in Roon, HQPlayer, JRiver, and Hang Loose Convolver. 

 

I'm unfamiliar with a system using 3 DACs for a DSP 3 way system. Can you talk a bit more about this? It sounds really interesting but I need to wrap my head around it. 

 

You will have trouble outputting via USB to more than one DAC at a time. I'm not saying it's impossible, but I wouldn't do it. I've used Lynx Studio AES cards in the past and think the AES16 PCIe could do what you want. Or the AES card from RME. You could externally clock the card and the DACs if they have word clock inputs. 

 

 

[The Computer Audiophile]

You could do something like use 3 Mytek Manhattan DACs, an AES card, and an external clock. The clock would sync the card and 3 DACs. 

 

I've used an external clock with Lynx cards in the past, it works well. 

[ecwl]

38 minutes ago, JeroenD said:

I am planning to use Acourate for measurement and creating the filters.

At the moment I am using Roon and I did read Roon can handle these filters so that could be interesting for me. Otherwise JRiver is one of the best option as far as I understood.

I now try to make the right design decisions regarding the hardware and interfaces to use.

Question 1: I am thinking of using an fanless NUC with i5 or i7 and 8GB RAM. Any advice on this?

Question 2: I am planning to add 2 more Metrum DAC’s to my system to create a DSP based 3-way system.

My take is that if you’re going to do this yourself with Acourate, you should just start now with your current system rather than spend any money trying to build the optimal active DSP speaker system.

Sure, the crossover between your subwoofers and 20.1 won’t be optimal. But Acourate has a fairly steep learning curve. So to me, being able to create a great convolution filter for your current system by treating it as a normal 2-channel one, would allow you to more easily optimize future filters for active 2-way/3-way systems in the future. It would also give you a better idea based on your current measurements, what upgrades and which 2-way/3-way setups would give you the most benefits in the future. Alternatively, you might be sufficiently happy with a 1-way 2-channel convolution filter correction that you’d stop and just go back to enjoying the music.

[JeroenD]

Thanks for your replay and warm welcome J.

Are you using Roon, HQPlayer, JRiver, and Hang Loose all on the same computer? Are you using them all or are you now actually using one and the others in the past to test?

 

I try to achieve a very high audio quality level. The Magnepan speakers are very revealing so small changes/ upgrades in the system are audible. The Metrum Adagio DAC which I use now is very good!

I would not like to move to a lower performing DAC because of moving to DSP based system. That is why I want to use a very good DAC for each of the 3 speaker drivers I want use.
I Ideally I would like to use this DAC at least for the analog input for the VS115 driving the tweeter and the KSA 300S driving the mid panel of the MG20.1. The DAC for analog input of the two Velodynes could be from lower spec like the Metrum Onyx.

It is defiantly not a cheap solution but I already have 1 Metrum Adagio. I also can sell my LS26 and a few cables to finance J.

 

I also have the feeling that using 3x USB out of the PC as input for the DAC’s is not a good option. I think it would be very nice if I could use I2S because this signal also has its own clock and the Metrum DAC’s have a I2S input. Using a Lynx AES16e card could also be a good option (not a cheap PC card) to use AES out which are synced by the clock on the Lynx. Thx for the advice!

[The Computer Audiophile]

7 minutes ago, JeroenD said:

Thanks for your replay and warm welcome J.

Are you using Roon, HQPlayer, JRiver, and Hang Loose all on the same computer? Are you using them all or are you now actually using one and the others in the past to test?

 

I try to achieve a very high audio quality level. The Magnepan speakers are very revealing so small changes/ upgrades in the system are audible. The Metrum Adagio DAC which I use now is very good!

I would not like to move to a lower performing DAC because of moving to DSP based system. That is why I want to use a very good DAC for each of the 3 speaker drivers I want use.
I Ideally I would like to use this DAC at least for the analog input for the VS115 driving the tweeter and the KSA 300S driving the mid panel of the MG20.1. The DAC for analog input of the two Velodynes could be from lower spec like the Metrum Onyx.

It is defiantly not a cheap solution but I already have 1 Metrum Adagio. I also can sell my LS26 and a few cables to finance J.

 

I also have the feeling that using 3x USB out of the PC as input for the DAC’s is not a good option. I think it would be very nice if I could use I2S because this signal also has its own clock and the Metrum DAC’s have a I2S input. Using a Lynx AES16e card could also be a good option (not a cheap PC card) to use AES out which are synced by the clock on the Lynx. Thx for the advice!

 

 

Here's my system and what I'm doing for DSP - 

 

 

 

I use Roon with HQPlayer frequently, or Audirvana with Hang Loose convolver. Sometimes just HQPlayer on its own. 

 

I love what you want to do and keep it as high quality as possible. I'd do the same thing. I am worried about using 3 DACs without a master clock through and the DAC you have doesn't allow for word clock input. You can always go the dCS route and use a dCS master clock, but that ups the price significantly. 

 

Ideally you would have one master clock that sends clock to the 3 DACs and the PC card. 

 

[JeroenD]

39 minutes ago, The Computer Audiophile said:

You could do something like use 3 Mytek Manhattan DACs, an AES card, and an external clock. The clock would sync the card and 3 DACs. 

 

I've used an external clock with Lynx cards in the past, it works well. 

This is more or less what I am trying to do but than using the Metrum NOS DAC's instead of the Mytek DAC’s. Of course interesting that you could use the same good clock signal for the DAC’s as well the Lynx PCIe cards. As far as I understood from a Metrum tech it could already be enough if the 3 AES or I2S signals used as input for the DAC’s are clocked with the same signal.

[JeroenD]

8 minutes ago, The Computer Audiophile said:

 

 

Here's my system and what I'm doing for DSP - 

 

 

 

I use Roon with HQPlayer frequently, or Audirvana with Hang Loose convolver. Sometimes just HQPlayer on its own. 

 

I love what you want to do and keep it as high quality as possible. I'd do the same thing. I am worried about using 3 DACs without a master clock through and the DAC you have doesn't allow for word clock input. You can always go the dCS route and use a dCS master clock, but that ups the price significantly. 

 

Ideally you would have one master clock that sends clock to the 3 DACs and the PC card. 

 

I will read your article about what you did regarding DSP and how. Looks very interesting!

I agree that having 1 good master clock connected to the PC card and DAC would be ideal.

I just also asked Metrum for advice. They mentioned that assuming that the 3 AES or I2S signals coming from DSP (or PCIe card) are synced regarding timing the jitter between the 3 DAC’s will be so limited that it far from audible. But defiantly something I want to be sure of.

[JeroenD]

37 minutes ago, ecwl said:

My take is that if you’re going to do this yourself with Acourate, you should just start now with your current system rather than spend any money trying to build the optimal active DSP speaker system.

Sure, the crossover between your subwoofers and 20.1 won’t be optimal. But Acourate has a fairly steep learning curve. So to me, being able to create a great convolution filter for your current system by treating it as a normal 2-channel one, would allow you to more easily optimize future filters for active 2-way/3-way systems in the future. It would also give you a better idea based on your current measurements, what upgrades and which 2-way/3-way setups would give you the most benefits in the future. Alternatively, you might be sufficiently happy with a 1-way 2-channel convolution filter correction that you’d stop and just go back to enjoying the music.

Hello ecwl, thx for your response.

I am aware it will not be easy to produce the right filters for the DSP I want to setup. The good thing is that I have a background in embedded software development (although now for a long time as leader and not actually developing code anymore) and are eager to understand creating the DSP filter settings. I defeitly will need help but I first will try to collect information and start reading about it. I already printed the whitepaper from Dr. Ulrich Bruggemann who is behind Acourate software.

 

So you suggest using Acourate for my current system and just create a room correction filter to start with. And if I succeed doing that I could purchase a PC and create a DSP crossover and per channel perform the timing correction of all the drivers.
Knowing myself I probably will not stop optimizing my system after creating a convolution filter for my current system but I like the ideay to start already now with my current system to start learning before investing money in more hardware and by that adding complexity.

[ecwl]

8 minutes ago, JeroenD said:

So you suggest using Acourate for my current system and just create a room correction filter to start with. And if I succeed doing that I could purchase a PC and create a DSP crossover and per channel perform the timing correction of all the drivers.
Knowing myself I probably will not stop optimizing my system after creating a convolution filter for my current system but I like the ideay to start already now with my current system to start learning before investing money in more hardware and by that adding complexity.

Right. Because to use Acourate to optimize for your current system, you have to:

1) Pick your preferred target frequency response

2) Learn how to optimize the time-domain filtering, aka. Optimize the excessive phase windowing parameters

 

And then you can even do more within Acourate because

3) if your speakers/subwoofers are asymmetrically setup in a room, you can do phase correction to avoid cancellations when sound is coming from the middle and your left & right channels are cancelling each other at specific frequencies

4) if your dominant room mode is the frequency generated by resonance between your front wall and back wall, you can add virtual bass array filtering into your convolution filter to remove that resonance first before doing the general room correction filter

 

Once you’ve optimized all these things, you can go back and decide if you want to start building 2-way or 3-way active crossover systems with multiple DACs and probably a Lynx interface feeding the DACs.

[JeroenD]

4 minutes ago, ecwl said:

Right. Because to use Acourate to optimize for your current system, you have to:

1) Pick your preferred target frequency response

2) Learn how to optimize the time-domain filtering, aka. Optimize the excessive phase windowing parameters

 

And then you can even do more within Acourate because

3) if your speakers/subwoofers are asymmetrically setup in a room, you can do phase correction to avoid cancellations when sound is coming from the middle and your left & right channels are cancelling each other at specific frequencies

4) if your dominant room mode is the frequency generated by resonance between your front wall and back wall, you can add virtual bass array filtering into your convolution filter to remove that resonance first before doing the general room correction filter

 

Once you’ve optimized all these things, you can go back and decide if you want to start building 2-way or 3-way active crossover systems with multiple DACs and probably a Lynx interface feeding the DACs.

Thx ecwl for the advice!

I understand mostly what you mentioned and you are right, I can already do quite some things using my current system regarding filter correction.

Just the go through the list you mentioned:

1.       What do you mean with “Pick your preferred target frequency response”?

2.       So perform time optimization by applying for delay to certain frequencies so the output of the drivers arrives more or less at the same time at the listening position? So this has to do with the frequency range a driver has (based on current analog crossover).

3.       Subwoofers are now just in front of the MG20 positioned at the side wall and drivers are facing towards each other. I did measure the subwoofers and corrected them as good as possible with the internal DSP. But tuning 2 Velodynes in 1 room is not that easy and took me many sweeps to adjust. But now sounds much better than using the bass panels of the MG20.1.

4.       With REW I measured my room before using the Velodynes. I have some room modes which were very annoying. With the Velodynes this is much, much better now but still some annoying frequencies in the bass area which blur the bass sound. So maybe using Acourate software I could improve the bass which comes from the subs to a next level.

[ecwl]

21 minutes ago, JeroenD said:

Thx ecwl for the advice!

I understand mostly what you mentioned and you are right, I can already do quite some things using my current system regarding filter correction.

Just the go through the list you mentioned:

1.       What do you mean with “Pick your preferred target frequency response”?

2.       So perform time optimization by applying for delay to certain frequencies so the output of the drivers arrives more or less at the same time at the listening position? So this has to do with the frequency range a driver has (based on current analog crossover).

3.       Subwoofers are now just in front of the MG20 positioned at the side wall and drivers are facing towards each other. I did measure the subwoofers and corrected them as good as possible with the internal DSP. But tuning 2 Velodynes in 1 room is not that easy and took me many sweeps to adjust. But now sounds much better than using the bass panels of the MG20.1.

4.       With REW I measured my room before using the Velodynes. I have some room modes which were very annoying. With the Velodynes this is much, much better now but still some annoying frequencies in the bass area which blur the bass sound. So maybe using Acourate software I could improve the bass which comes from the subs to a next level.

1. Different people have different preferences on the ideal frequency response curve:

https://accuratesound.ca/standards

Some people like a few dB higher in the bass. Some people want 20kHz to be at -10dB lower while others prefer only -4dB lower than the rest of the frequencies. When you create the convolution filter in Acourate, you can choose what target you like. It is impossible to guess your preference so you have to basically create a few optimized filters with different target curves and just listen to what you like best. I do find that knowing what your current system’s frequency response usually helps guess what target response you’ll prefer.

2. So when we talk about “time-domain correction” in Acourate or Audiolense for 1-way systems, we are mostly not talking about aligning the tweeter, midrange, woofer and subwoofer to become time coincident which is what a 2-way or 3-way system can do. The advantage of the “time-domain correction” in Acourate for a generic filter is that at different bass frequencies, you’re going to have different degrees of resonances leading to different phase shifts. By correcting these phase shifts in addition to an overall frequency response correction, you’re going to get much better and more coherent bass rather than just a straightforward frequency response correction.

3. I call your system “1-way” because to Acourate right now, it is treating your left MG20.1 & left subwoofer as one left speaker and your right MG20.1 & right subwoofer as one right speaker. That’s actually the setup I used to have. Of course, you’ve already made additional adjustments to your subwoofers so they’re optimally integrated to the MG20.1. So essentially, you’re trying to get Acourate to correct what you can’t correct just from tweaking the subwoofer DSP.

4. By having a target frequency response, any bass peaks in your current system would be corrected. However, this is a generic correction where it just lowers the volume at the peaks. But for certain types of resonances like say in my system on the right channel, I have peaks at 40Hz, 51.5Hz, 63Hz, 75.4Hz & 90.8Hz but because my front to back wall is 3.33m, the 51.5Hz resonance is specifically due to that lengthwise resonance. So I can either have a convolution filter to correct all 5 peaks generically, or I can generate a virtual bass array filter to correct the 51.5Hz resonance first and then run a convolution filter on top of the 51.5Hz correction to correct the other 4 peaks.

 

The bottomline is that until you actually start using Acourate, all of this is very technical and confusing. Sometimes, planning is just talking mumbo jumbo that don’t make any sense until you actually start doing. That’s why I think your next step really is to start using Acourate.

 

Once you have everything that we talked about here sorted out, that’s when you’re ready to start thinking about whether you want to embarking on building and creating active crossover filters for 2-way or 3-way digital setups. Basically, you should learn how to swim first before trying to do the butterfly stroke.

 

With all that said, as a total aside… If you already have REW measurements of your current system (MG20.1 + Velodyne; left channel and right channel), have you tried just using parametric EQs in Roon to tune out the residual bass? Maybe that’s the better, easier and cheaper way to go for you. Maybe you’ll be sufficiently happy with that result and you wouldn’t want to pay for Acourate and embark in this complicated process of creating convolution filters.

[JeroenD]

8 minutes ago, ecwl said:

1. Different people have different preferences on the ideal frequency response curve:

https://accuratesound.ca/standards

Some people like a few dB higher in the bass. Some people want 20kHz to be at -10dB lower while others prefer only -4dB lower than the rest of the frequencies. When you create the convolution filter in Acourate, you can choose what target you like. It is impossible to guess your preference so you have to basically create a few optimized filters with different target curves and just listen to what you like best. I do find that knowing what your current system’s frequency response usually helps guess what target response you’ll prefer.

2. So when we talk about “time-domain correction” in Acourate or Audiolense for 1-way systems, we are mostly not talking about aligning the tweeter, midrange, woofer and subwoofer to become time coincident which is what a 2-way or 3-way system can do. The advantage of the “time-domain correction” in Acourate for a generic filter is that at different bass frequencies, you’re going to have different degrees of resonances leading to different phase shifts. By correcting these phase shifts in addition to an overall frequency response correction, you’re going to get much better and more coherent bass rather than just a straightforward frequency response correction.

3. I call your system “1-way” because to Acourate right now, it is treating your left MG20.1 & left subwoofer as one left speaker and your right MG20.1 & right subwoofer as one right speaker. That’s actually the setup I used to have. Of course, you’ve already made additional adjustments to your subwoofers so they’re optimally integrated to the MG20.1. So essentially, you’re trying to get Acourate to correct what you can’t correct just from tweaking the subwoofer DSP.

4. By having a target frequency response, any bass peaks in your current system would be corrected. However, this is a generic correction where it just lowers the volume at the peaks. But for certain types of resonances like say in my system on the right channel, I have peaks at 40Hz, 51.5Hz, 63Hz, 75.4Hz & 90.8Hz but because my front to back wall is 3.33m, the 51.5Hz resonance is specifically due to that lengthwise resonance. So I can either have a convolution filter to correct all 5 peaks generically, or I can generate a virtual bass array filter to correct the 51.5Hz resonance first and then run a convolution filter on top of the 51.5Hz correction to correct the other 4 peaks.

 

The bottomline is that until you actually start using Acourate, all of this is very technical and confusing. Sometimes, planning is just talking mumbo jumbo that don’t make any sense until you actually start doing. That’s why I think your next step really is to start using Acourate.

 

Once you have everything that we talked about here sorted out, that’s when you’re ready to start thinking about whether you want to embarking on building and creating active crossover filters for 2-way or 3-way digital setups. Basically, you should learn how to swim first before trying to do the butterfly stroke.

 

With all that said, as a total aside… If you already have REW measurements of your current system (MG20.1 + Velodyne; left channel and right channel), have you tried just using parametric EQs in Roon to tune out the residual bass? Maybe that’s the better, easier and cheaper way to go for you. Maybe you’ll be sufficiently happy with that result and you wouldn’t want to pay for Acourate and embark in this complicated process of creating convolution filters.

Thx ecwl for the clear explanation!

That really helps me to understand the basics to start with.

I will follow your advice and first start to use Acourate with my current setup. My system sounds already quit amazing but I know there is much more potential in it. I would like to use the Audio Research VS115 just to drive the tweeters. The Krell works great for the mid panel. I also want to remove the analog crossover components like the active crossover (very old and not the best I guess) and the crossover behind the speakers. My upgraded LS26 is very nice (outperforms a REF 5) and does something magical to the sound. I hope I will not miss it when I move to software DSP.

Is controlling the volume with Roon not a drawback or do you control the volume diffrenetly?

 

 I am lucky I have a separate room for my audio hobby in the basement of my new build home. The room is 5m wide, 8 meter deep and 2.6m high. The MG20.1 is about 1.8m from the back wall tweeters at the outside.

I have a acoustic treatment at the sealing which is a black paper glue mixture that is sprayed to the sealing and helps to even frequencies above 300 Hz a bit. I also have some room treatment at the back corners.

At the moment my Metrum Ambre is my Roon endpoint. So I am not sure if I can use the Acourate filters with this setup. Or do I need to run Roon on a PC and than control the Ambre from that PC?

 

For the REW measurements I used the measurement microphone which comes with the Velodyne which is a Dayton Audio EMM-6. Is that Microphone good enough to use with Acourate or should I go for something better?

I found this article to start learning to use Acourate: https://archimago.blogspot.com/2019/10/2019-update-basic-acourate-dsp-room.html

What do you think about this to start with?

What kind of system do you have?

 

[ecwl]

20 minutes ago, JeroenD said:

At the moment my Metrum Ambre is my Roon endpoint. So I am not sure if I can use the Acourate filters with this setup. Or do I need to run Roon on a PC and than control the Ambre from that PC?

 

For the REW measurements I used the measurement microphone which comes with the Velodyne which is a Dayton Audio EMM-6. Is that Microphone good enough to use with Acourate or should I go for something better?

I found this article to start learning to use Acourate: https://archimago.blogspot.com/2019/10/2019-update-basic-acourate-dsp-room.html

What do you think about this to start with?

1. If Metrum Ambre is your Roon endpoint, what is your Roon Core? I really don’t understand your question. If you’re already running Roon, you have to have a PC or another computer running Roon Core. You run the convolution filter on the Roon Core. How are you currently using Metrum Ambre?

2. Dayton Audio EMM-6 is fine. Just make sure you have the calibration file

3. I actually read through Mitch Barrett’s book Accurate Sound Reproduction Using DSP to learn how to use Acourate. I just bought the Kindle version. But I guess you can start with Archimago’s instructions. I personally don’t think Archimago addressed the issues of pre-ringing and group delays sufficiently but you gotta start somewhere I guess.

[JeroenD]

1 hour ago, ecwl said:

1. If Metrum Ambre is your Roon endpoint, what is your Roon Core? I really don’t understand your question. If you’re already running Roon, you have to have a PC or another computer running Roon Core. You run the convolution filter on the Roon Core. How are you currently using Metrum Ambre?

2. Dayton Audio EMM-6 is fine. Just make sure you have the calibration file

3. I actually read through Mitch Barrett’s book Accurate Sound Reproduction Using DSP to learn how to use Acourate. I just bought the Kindle version. But I guess you can start with Archimago’s instructions. I personally don’t think Archimago addressed the issues of pre-ringing and group delays sufficiently but you gotta start somewhere I guess.

1. My Roon core is running on my Qnap NAS. I normale use my tablet to control Roon. So I am not sure where in my setup the Acourate filters wil run. On the Qnap Roon core software is installed which I know is not officially developed by Roon but Roon refers to this Qnap software to use the Qnap as a Roon server. I don't know if I upload the files for the filters if they than are loaded in the Roon core.

2. On the Dayton Audio EMM-6 microphones shipped by Velodyne there is no serial number. Normally a serial nr is on the bottom that you need for downloading the calibration file. When it is important to have the calibration file than I am happy to invest in a new measurement microphone.

3. I did Google but couldn't find anything about the book of Mitch Barrett. Do you know where to buy the book?

[ecwl]

44 minutes ago, JeroenD said:

1. My Roon core is running on my Qnap NAS. I normale use my tablet to control Roon. So I am not sure where in my setup the Acourate filters wil run. On the Qnap Roon core software is installed which I know is not officially developed by Roon but Roon refers to this Qnap software to use the Qnap as a Roon server. I don't know if I upload the files for the filters if they than are loaded in the Roon core.

2. On the Dayton Audio EMM-6 microphones shipped by Velodyne there is no serial number. Normally a serial nr is on the bottom that you need for downloading the calibration file. When it is important to have the calibration file than I am happy to invest in a new measurement microphone.

3. I did Google but couldn't find anything about the book of Mitch Barrett. Do you know where to buy the book?

1. You should just be able to load the convolution files onto your Qnap NAS. You can check it in Roon DSP section for your zone. The bigger question is whether your Qnap NAS can run the convolution. That would depend on what CPU your Qnap uses. No way to know until you try it? I don’t think convolution is that processor intensive.

2. I don’t know how accurate your EMM-6 microphone is. I guess ideally you want a calibrated microphone. I personally use the miniDSP UMIK-1 microphone. The other issue is that if your computer is connected to a USB microphone and a USB DAC, sometimes (or maybe more often than not), you may have clocking syncing issues during the recording of the LogSweep in Acourate. Acourate recommends using like an Uber expensive USB interface from Lynx or RME and a compatible microphone (non-USB/analog) and feeding your DAC with the digital output to ensure optimal clocking. I got lucky and didn’t run into much of an issue. But I can also use my laptop’s headphone jack to feed my amp. You can probably do that to your LS26 for the purposes of creating the convolution filter. And then there is LSR3 recorder which can compensate for some clock drift but not a lot.

3. https://www.amazon.ca/Accurate-Sound-Reproduction-Using-DSP-ebook/dp/B01FURPS40/ref=sr_1_1?dchild=1&keywords=accurate+sound+reproduction&qid=1630112388&s=digital-text&sr=1-1

I got the kindle version. I just realized it’s Mitch Barnett not Barrett.

[JeroenD]

5 hours ago, ecwl said:

1. You should just be able to load the convolution files onto your Qnap NAS. You can check it in Roon DSP section for your zone. The bigger question is whether your Qnap NAS can run the convolution. That would depend on what CPU your Qnap uses. No way to know until you try it? I don’t think convolution is that processor intensive.

2. I don’t know how accurate your EMM-6 microphone is. I guess ideally you want a calibrated microphone. I personally use the miniDSP UMIK-1 microphone. The other issue is that if your computer is connected to a USB microphone and a USB DAC, sometimes (or maybe more often than not), you may have clocking syncing issues during the recording of the LogSweep in Acourate. Acourate recommends using like an Uber expensive USB interface from Lynx or RME and a compatible microphone (non-USB/analog) and feeding your DAC with the digital output to ensure optimal clocking. I got lucky and didn’t run into much of an issue. But I can also use my laptop’s headphone jack to feed my amp. You can probably do that to your LS26 for the purposes of creating the convolution filter. And then there is LSR3 recorder which can compensate for some clock drift but not a lot.

3. https://www.amazon.ca/Accurate-Sound-Reproduction-Using-DSP-ebook/dp/B01FURPS40/ref=sr_1_1?dchild=1&keywords=accurate+sound+reproduction&qid=1630112388&s=digital-text&sr=1-1

I got the kindle version. I just realized it’s Mitch Barnett not Barrett.

1.       I use the Qnap TS-673-8G which has a AMD RX-421ND quad-core 2.1 GHz CPU (burst up to 3.4 GHz), and 8GB dual channel DDR4. I have plenty of storage on harddisks and SSD. I also use it for running a virtual machine that is not using much CPU and memory.

2.       I will first try with the microphone I have. I have a Behringer cheap USB interface that I connect with my laptop and a line out to my LS26 input. https://www.behringer.com/product.html?modelCode=P0AVV

It is probably not the best solution like you mention but to start with maybe ok. When needed I can upgrade to something more close to what Acourate advices.

3.       Many thanks for the link to the book. I will soon download it and start reading.

Still curious what kind of audio system you are using including your DSP setup.

[ecwl]

4 hours ago, JeroenD said:

Still curious what kind of audio system you are using including your DSP setup.

Hmmm. It’s in my signature?

I use my ancient PC (i7-3770k) as my Roon core with my music files stored in an ancient DS211j NAS playing into UltraRendu that is connected to Peachtree X1 to convert to Toslink feeding my Chord M-Scaler then Chord DAVE then Chord Etude into the Dynaudio C1 speakers. I used to have Sunfire subwoofers integrated like you do with your Dynaudio with the DAVE feeding them but I took them out as I may get a new pair of speakers soon. So I’ve made a few convolution filters for the old setup and for my current no subwoofers setup.

My living room is untreated and asymmetric so the front to back wall distance for the left speaker is slightly different than the right. My ancient laptop is a Lenovo X40 which I use to run Acourate. 

[JeroenD]

Initially I had to plan to use a PC with PC card like the RME HDSPe AES to output 3 AES signals (low, medium and high frequency) out of the PC and use 3 stereo Metrum DAC’s to perform the DA conversion.

I did read people advising to use a multi DAC with 6 or 8 channels (like the Merging Hapi) to perform the DAC because this would lead to better performance.

For me at this moment it is not clear why this would be better.

The Metrum Adagio I have now is a really good very analog sounding DAC which I expect will perform better than the Hapi regarding DA conversion.

I just want to understand why a “lower” performing 8 channel DAC like the Hapi would be a better choice. Is anybody able to explain this or has anybody experienced this in practice?

[ecwl]

I think the best way to think about this is that all stereo systems are about compromises. And different people hear different things so they tend to emphasize on different things in their stereo system. Our ears are very good at tuning out sonic anomalies. Otherwise, we wouldn't be able to pick out a specific conversation in a noisy room.

For some people DACs matter a lot sonically and for others, if the DAC measures great, they all sound the same. For some people, active crossovers that you prefer matter a lot more and for others like myself, I think there are so many sonic issues that passive crossovers are acceptable if the other sonic issues are addressed, e.g. DACs, digital room correction. You may think there is a no compromise solution but that doesn't exist.

 

So in answer to your question why Hapi vs your solution

Hapi

Benefits: one DAC with easy volume control.

Downside: DAC might not give you the sound you like

Your solution

Benefits: you get the DAC sound you think you want

Downside: other than the issue of the cost of 3 DACs, the primary issue is one of volume control. You can control the volume to all 3 DACs digitally but the problem with ladder DACs is that they have poor low-level linearity so while your DAC sounds great with analog volume control, it might not once you have to lower the volume digitally and feed through your DAC. Moreover, running a DAC straight into the amp gives you a higher chance of blowing the amp or the speaker drivers. And then there is the issue of whether all 3 DACs would be time-synced through your RME. In theory they should be but there is no 100% guarantee.