An update on windows crossovers
Hi, i thought i'd share some more experiences after testing both AsioXO and the Aedio plugin for Foobar:
I got AsioXO to output to KX now, it turns out a generic Asio-driver that came with Cubase messed things up for me.
AsioXO selected it instead of the KX-driver, and it's only two channels, hence the "invalid channel count".
I still can't figure out how to get it to map to the correct outputs in KX though. I'm positive this can be setup to work, theres almost infinite possibilities in the signal routing in KX. Just have to find the right outputs...
The Aedio plug-in is alot easier to setup as Foobar lets you choose output device. It worked almost straight away. There are still a few bugs here and there, but i'm not sure if they come from the plug-in or Foobar itself. It crashed a couple of times when i was messing around with the settings. It didn't seem to load my system much either, i have an Athlon 2400+ and 512mb ram, the difference wasn't noticable to me.
The really great thing about the KX plugin is that it's in the driver. That way your not limited to a specific player and all audio goes through the crossover. This is a really great advantage as there's no worry about blowing your tweeters with standard windows sound or games by accident.
I have an idea (wishful thinking really😀), how about a crossover as a virtual device in windows? That is, a driver that acts as a sound card, but really just does the filtering and then outputs the soundstreams to the soundcard(s) of your choice. I guess this is similar to how BruteFIR works? Hopefully it could be made to work with just about any soundcard...
How come all my ideas always end up with windows driver programming?
Sorry about the long post, i hope at least some of it is of any interest. Well maybe it's time to stop talking and start coding...
/Andreas
Hi, i thought i'd share some more experiences after testing both AsioXO and the Aedio plugin for Foobar:
I got AsioXO to output to KX now, it turns out a generic Asio-driver that came with Cubase messed things up for me.
AsioXO selected it instead of the KX-driver, and it's only two channels, hence the "invalid channel count".
I still can't figure out how to get it to map to the correct outputs in KX though. I'm positive this can be setup to work, theres almost infinite possibilities in the signal routing in KX. Just have to find the right outputs...
The Aedio plug-in is alot easier to setup as Foobar lets you choose output device. It worked almost straight away. There are still a few bugs here and there, but i'm not sure if they come from the plug-in or Foobar itself. It crashed a couple of times when i was messing around with the settings. It didn't seem to load my system much either, i have an Athlon 2400+ and 512mb ram, the difference wasn't noticable to me.
The really great thing about the KX plugin is that it's in the driver. That way your not limited to a specific player and all audio goes through the crossover. This is a really great advantage as there's no worry about blowing your tweeters with standard windows sound or games by accident.
I have an idea (wishful thinking really😀), how about a crossover as a virtual device in windows? That is, a driver that acts as a sound card, but really just does the filtering and then outputs the soundstreams to the soundcard(s) of your choice. I guess this is similar to how BruteFIR works? Hopefully it could be made to work with just about any soundcard...
How come all my ideas always end up with windows driver programming?

Sorry about the long post, i hope at least some of it is of any interest. Well maybe it's time to stop talking and start coding...
/Andreas
Need help
I need to finish this phase of my development of the Linux music player to clear the deck for serious obligations coming up.
As soon as drc is completed, a summary of progress will be written, which should help others avoid some of the pitfalls.
However I am deep inside a pitfall of my own atm and am hoping some of you can hand down a rope.
I can generate a 2-channel drc fir filter and I already have a 2-ch to 4-ch crossover fir filter that I like. How can these 2 be combined?
1. 2 instances of brutefir, drc-brutefir -> crossover-brutefir? How?
OR
2. combine the drc filter into the existing crossover filter? How? Is it as simple as literally the dot product of [drc ir vector] * [low or highpass ir vector] ?
Thanks.
-Robert
I need to finish this phase of my development of the Linux music player to clear the deck for serious obligations coming up.
As soon as drc is completed, a summary of progress will be written, which should help others avoid some of the pitfalls.
However I am deep inside a pitfall of my own atm and am hoping some of you can hand down a rope.
I can generate a 2-channel drc fir filter and I already have a 2-ch to 4-ch crossover fir filter that I like. How can these 2 be combined?
1. 2 instances of brutefir, drc-brutefir -> crossover-brutefir? How?
OR
2. combine the drc filter into the existing crossover filter? How? Is it as simple as literally the dot product of [drc ir vector] * [low or highpass ir vector] ?
Thanks.
-Robert
Re: Need help
Dont take the dot product of the IR. What you do is run the DRC IR trough brutefir ( Low pass IR vektor) - This new IR is what you use as the new Low pass + DRC IR in Brure fir. Similar onthe high pass side.
If you want to do it directly you would need to convolve the DRC IR and the IR Low/high pass. You can look up convolution in google - it is a little to much to write in this thread
Regards
Morten
RFScheer said:
I can generate a 2-channel drc fir filter and I already have a 2-ch to 4-ch crossover fir filter that I like. How can these 2 be combined?
..
2. combine the drc filter into the existing crossover filter? How? Is it as simple as literally the dot product of [drc ir vector] * [low or highpass ir vector] ?
Thanks.
-Robert
Dont take the dot product of the IR. What you do is run the DRC IR trough brutefir ( Low pass IR vektor) - This new IR is what you use as the new Low pass + DRC IR in Brure fir. Similar onthe high pass side.
If you want to do it directly you would need to convolve the DRC IR and the IR Low/high pass. You can look up convolution in google - it is a little to much to write in this thread
Regards
Morten
You can feed filters into other filters using the one BruteFIR process.
So youll want to do somthing like this in the one brutefir config:
In left -> DRC filter -> xover filters -> left outs
In right -> DRC filter -> xover filters -> right outs
You could convolve the filter sets together, but if your unsure on FFT theory, the above is easier if youve got a bit of spare CPU time.
So youll want to do somthing like this in the one brutefir config:
In left -> DRC filter -> xover filters -> left outs
In right -> DRC filter -> xover filters -> right outs
You could convolve the filter sets together, but if your unsure on FFT theory, the above is easier if youve got a bit of spare CPU time.
Really a great thread! I've been looking to do something like this for some time now.
Personally, I think a DIY stand-alone DSP card would be ideal:
1. user-upgradeable software, and, depending on the implementation, may be much cheaper than replacing hardware like the DEQX
2. low latency -- I think it was rfbrw who said that DSP chips run circles around modern PCs. For this type of DSP-specific number-crunching, everything I've read agrees with that
3. the stand-alone is not OS-specific - you can use Windows or Linux-based programs run through the stand-alone DSP
However, the downside, and what eventually discouraged me from attempting this is the steep learning curve. I don't have any formal training with DSP, so, learning the hardware and software and DSP programming could take me years of nights and weekends, and no guarantee of satisfaction. While I love a good intellectual challenge as much as the next guy, this is too much. The industry moves too quickly -- as soon as I'd learned some piece of the puzzle, the landscape would have shifted and what I learned would either be useless, or out-dated.
I'd like to use whatever DSP processing I invest time or money in for both HT and audio, which means latency is a huge concern. Windows itself isn't built for high resolution low-latency multi-channel audio. ASIO might be or VST plug-ins might be. I don't know enough about Windows audio architecture to say whether an XO / DRC-type DSP could be implemented using ASIO, VST, ReWire, etc. In any case, the primary reason for Windows would be the surround processing capabilities of programs like WinDVD or ZoomPlayer for DVD-playback. Also, front-ends like myHTPC or the Cinemar Lobby series are years ahead of anything available on Linux. Finally, and for me, most importantly, the lack of soundcard support on Linux. I love my Lynx TwoB and would like to add in (or, *gasp* replace it with) an AES16. There are no Linux drivers for either of these cards, and, due to the NDA Lynx Studios requires, it doesn't look like there ever will be drivers for these cards. (I'm aware of the OSS driver by 4Front, but, I don't believe this works (or works well) for ALSA, and, this is for the LynxTwo, not the AES16). Maybe there's something else out there? Getting an RME setup with the requisite inputs and outputs could cost just south of $1,000, not including the computer itself and any DACs one would build or buy. Add in the time and frustration required to properly set up a low-latency kernel, ALSA, Jack and BruteFIR, take impulse responses, design filters and work through connection and latency issues and DEQX looks like a bargain, particularly if it outperforms BruteFIR....
So, I wonder about compiling Brutefir on OS X? Jack works on OS X. Lynx Studios makes OS X Core Audio drivers.
Also, what about porting BruteFIR to Windows?
Finally, has anyone done a direct comparison of the BruteFIR (XO + DRC) combination with DEQX? If so, how did BruteFIR compare?
brad
Personally, I think a DIY stand-alone DSP card would be ideal:
1. user-upgradeable software, and, depending on the implementation, may be much cheaper than replacing hardware like the DEQX
2. low latency -- I think it was rfbrw who said that DSP chips run circles around modern PCs. For this type of DSP-specific number-crunching, everything I've read agrees with that
3. the stand-alone is not OS-specific - you can use Windows or Linux-based programs run through the stand-alone DSP
However, the downside, and what eventually discouraged me from attempting this is the steep learning curve. I don't have any formal training with DSP, so, learning the hardware and software and DSP programming could take me years of nights and weekends, and no guarantee of satisfaction. While I love a good intellectual challenge as much as the next guy, this is too much. The industry moves too quickly -- as soon as I'd learned some piece of the puzzle, the landscape would have shifted and what I learned would either be useless, or out-dated.
I'd like to use whatever DSP processing I invest time or money in for both HT and audio, which means latency is a huge concern. Windows itself isn't built for high resolution low-latency multi-channel audio. ASIO might be or VST plug-ins might be. I don't know enough about Windows audio architecture to say whether an XO / DRC-type DSP could be implemented using ASIO, VST, ReWire, etc. In any case, the primary reason for Windows would be the surround processing capabilities of programs like WinDVD or ZoomPlayer for DVD-playback. Also, front-ends like myHTPC or the Cinemar Lobby series are years ahead of anything available on Linux. Finally, and for me, most importantly, the lack of soundcard support on Linux. I love my Lynx TwoB and would like to add in (or, *gasp* replace it with) an AES16. There are no Linux drivers for either of these cards, and, due to the NDA Lynx Studios requires, it doesn't look like there ever will be drivers for these cards. (I'm aware of the OSS driver by 4Front, but, I don't believe this works (or works well) for ALSA, and, this is for the LynxTwo, not the AES16). Maybe there's something else out there? Getting an RME setup with the requisite inputs and outputs could cost just south of $1,000, not including the computer itself and any DACs one would build or buy. Add in the time and frustration required to properly set up a low-latency kernel, ALSA, Jack and BruteFIR, take impulse responses, design filters and work through connection and latency issues and DEQX looks like a bargain, particularly if it outperforms BruteFIR....
So, I wonder about compiling Brutefir on OS X? Jack works on OS X. Lynx Studios makes OS X Core Audio drivers.
Also, what about porting BruteFIR to Windows?
Finally, has anyone done a direct comparison of the BruteFIR (XO + DRC) combination with DEQX? If so, how did BruteFIR compare?
brad
ttt as its still the most informative thread on this subject that I have found.
Hopefully Robert is enjoying his project, has anyone else been doing anything similar?
Hopefully Robert is enjoying his project, has anyone else been doing anything similar?
Enjoying oh yes
Hi all,
I have been enjoying my system even though it has one more step to go before it's complete as version 1.0. Still haven't finished the digital room correction, although it's very close.
The music player box is now headless (no video), keyboardless and mouseless. The box is controlled through a VNC window on my desktop screen so the music player is effectively a server.
It is running kernel 2.6.7 with con kolivas patch and realtime_lsm enabled. All audio components are launched by schedtool. There is absolutely no issue with glitches and the player will run without crashing for weeks if I want (although usually I interrupt it with some kind of futzing around).
I finally figured out how to setup Brutefir so that drc filters precede the crossover filters. This means it is cranking 6x 16k filters in real time and this occupies about 29% of cpu. For now, the drc filters are just dirac pulses, meaning they just pass the sound right through without correction, but very soon...
Many hours of experimenting have led to recording through a digital audio workstation program called Ardour, which also interfaces nicely with Jack so the test sweep can be played through the crossover filters while recording the response. Soon, I will record the system/room response and generate real drc filters. I've practiced recording and have the mic/preamp setup well optimized.
Using xmms as the player front-end, I can play just about anything, but mainly use wav for my personal collection and often listen to internet radio as well. Internet radio sounds ok but with minor artifacts that I haven't bothered with trying to improve.
After weeks of listening while working at my office desk, I still greatly enjoy the sound although it is definitely not world-class. Next year, I will likely design and build a totally new version 2.0, with improvements to low-end, visual aesthetics and room acoustics.
But first, after finishing drc filters, I'll publish the documentation on version 1.0. Promise.
-Robert
Hi all,
I have been enjoying my system even though it has one more step to go before it's complete as version 1.0. Still haven't finished the digital room correction, although it's very close.
The music player box is now headless (no video), keyboardless and mouseless. The box is controlled through a VNC window on my desktop screen so the music player is effectively a server.
It is running kernel 2.6.7 with con kolivas patch and realtime_lsm enabled. All audio components are launched by schedtool. There is absolutely no issue with glitches and the player will run without crashing for weeks if I want (although usually I interrupt it with some kind of futzing around).
I finally figured out how to setup Brutefir so that drc filters precede the crossover filters. This means it is cranking 6x 16k filters in real time and this occupies about 29% of cpu. For now, the drc filters are just dirac pulses, meaning they just pass the sound right through without correction, but very soon...
Many hours of experimenting have led to recording through a digital audio workstation program called Ardour, which also interfaces nicely with Jack so the test sweep can be played through the crossover filters while recording the response. Soon, I will record the system/room response and generate real drc filters. I've practiced recording and have the mic/preamp setup well optimized.
Using xmms as the player front-end, I can play just about anything, but mainly use wav for my personal collection and often listen to internet radio as well. Internet radio sounds ok but with minor artifacts that I haven't bothered with trying to improve.
After weeks of listening while working at my office desk, I still greatly enjoy the sound although it is definitely not world-class. Next year, I will likely design and build a totally new version 2.0, with improvements to low-end, visual aesthetics and room acoustics.
But first, after finishing drc filters, I'll publish the documentation on version 1.0. Promise.
-Robert
Robert, good to hear you are enjoying your system and thanks for the update.
I am convinced that this or something like it is what we will all be using in the not too distant future.
I am looking foward to your further info and hopefully I will be having a go myself fairly soon.
Tim.
I am convinced that this or something like it is what we will all be using in the not too distant future.
I am looking foward to your further info and hopefully I will be having a go myself fairly soon.
Tim.
wytco0 said:
<snip>
I am convinced that this or something like it is what we will all be using in the not too distant future.
<snip>
Not for this luddite. Too bloated and impractical for multi-room use.
It's not an ipod
rfbrw,
The goal of my system is to play music at my desk and could be extended to playing in a listening room situation, but has never been intended for multi-room sound.
For most people, multi-room sound is best handled by an ipod or similar "full-range" portable player, definitely the anti-bloat solution. Not interesting to me though.
Anyway, more interesting to me is whether version 2.0 can achieve the level of sound quality I'm looking for on a budget of < $5k, which is a limit I've set for my final office sound system. I'm confident in the electronics but not the electromechanics (ie the speakers). I can't figure out how to build great speakers on par with my home theater Magnepan MG20's & Rel Studio subwoofers without spending too much on drivers.
-Robert
rfbrw,
The goal of my system is to play music at my desk and could be extended to playing in a listening room situation, but has never been intended for multi-room sound.
For most people, multi-room sound is best handled by an ipod or similar "full-range" portable player, definitely the anti-bloat solution. Not interesting to me though.
Anyway, more interesting to me is whether version 2.0 can achieve the level of sound quality I'm looking for on a budget of < $5k, which is a limit I've set for my final office sound system. I'm confident in the electronics but not the electromechanics (ie the speakers). I can't figure out how to build great speakers on par with my home theater Magnepan MG20's & Rel Studio subwoofers without spending too much on drivers.
-Robert
Ipod for multi-room. Methinks not.
Anyway I would probably find it a little tricky fitting all my cds on an Ipod in an uncompressed form not to mention LP's recorded at 96k or higher but we can leave the topic of the practical application of the personal computer in a multi-room scenario for a different time.
As regards speakers in the office I would simply buy a pair of near-field monitors.
Anyway I would probably find it a little tricky fitting all my cds on an Ipod in an uncompressed form not to mention LP's recorded at 96k or higher but we can leave the topic of the practical application of the personal computer in a multi-room scenario for a different time.
As regards speakers in the office I would simply buy a pair of near-field monitors.
Near-field monitors
Right, I'm currently using Madisound Vifa Studio's, which are basically near-field monitors that were a $500 "kit". However, they don't get down low enough and are not giving what I expect in detail and life-like imagery. They sound a lot better than good computer speakers but not good enough.
Any recommendations?
The ipod was a joke of course, but seriously, what are you up to regarding multiroom sound?
-Robert
Right, I'm currently using Madisound Vifa Studio's, which are basically near-field monitors that were a $500 "kit". However, they don't get down low enough and are not giving what I expect in detail and life-like imagery. They sound a lot better than good computer speakers but not good enough.
Any recommendations?
The ipod was a joke of course, but seriously, what are you up to regarding multiroom sound?
-Robert
Haven't paid much attention to speakers recently so I can't say. I have some tentative ideas on using the PC for multi-room apps but that needs more work and I have yet to find the time.
Re: Near-field monitors
Because you can have very steep rolloff using DSP filters it's safe and easy to use metal cone speakers. I would like to recommend Seas drivers, I have Excel W21 woofers with Seas H883 tweeters in waveguide design under development (50L@35Hz). Now there's also the new W22 woofer, and also much better H1213 tweeter. Also the L18RNX/P (or Excel W18) would be a good choice, as it's easier to handle without waveguide (for tweeter) than the 8" inch one if you cross them over at 1200 - 1500 Hz - experiment is the key, if you go too low the tweeters sound stressed (you can also get waveguides but you have to order it from Finland :-, ask them from Hifitalo, info@hifitalo.fi, they cost ~ $15/piece , the 140 mm model fits to these 25-27 mm Seas tweeters). In my case there are notches @3500 and 7000Hz which you (actually me) have to correct some way, but I think you can use DRC to correct the response of wave guides or horns easily, just take care not to boost extrem highs too much, you should be able to plot the frequency response of the IR before trying it, scopeFIR can do it.
Here is one Seas example (though it's passive):
DD8-MkII
Using 24L enclosures tuned @35 Hz you could get nice and deep bass (not too much volume because of the limited cone area compared to 22 cm bass driver but better midrange). In room almost flat to tuning frequency. With DRC even better.
These famous speakers use Seas alu tweeters: http://www.totemacoustic.com/english/products/bookShelf_model01.htm
RFScheer said:Right, I'm currently using Madisound Vifa Studio's, which are basically near-field monitors that were a $500 "kit". However, they don't get down low enough and are not giving what I expect in detail and life-like imagery. They sound a lot better than good computer speakers but not good enough.
Any recommendations?
-Robert
Because you can have very steep rolloff using DSP filters it's safe and easy to use metal cone speakers. I would like to recommend Seas drivers, I have Excel W21 woofers with Seas H883 tweeters in waveguide design under development (50L@35Hz). Now there's also the new W22 woofer, and also much better H1213 tweeter. Also the L18RNX/P (or Excel W18) would be a good choice, as it's easier to handle without waveguide (for tweeter) than the 8" inch one if you cross them over at 1200 - 1500 Hz - experiment is the key, if you go too low the tweeters sound stressed (you can also get waveguides but you have to order it from Finland :-, ask them from Hifitalo, info@hifitalo.fi, they cost ~ $15/piece , the 140 mm model fits to these 25-27 mm Seas tweeters). In my case there are notches @3500 and 7000Hz which you (actually me) have to correct some way, but I think you can use DRC to correct the response of wave guides or horns easily, just take care not to boost extrem highs too much, you should be able to plot the frequency response of the IR before trying it, scopeFIR can do it.
Here is one Seas example (though it's passive):
DD8-MkII
Using 24L enclosures tuned @35 Hz you could get nice and deep bass (not too much volume because of the limited cone area compared to 22 cm bass driver but better midrange). In room almost flat to tuning frequency. With DRC even better.
These famous speakers use Seas alu tweeters: http://www.totemacoustic.com/english/products/bookShelf_model01.htm
64k filters
mhelin,
Thanks for the reference on speakers. I might look further into the 24L cabinet monitors. They look promising.
However, first things first. I did some experimenting this morning with creating various filters and testing them. Here are the results:
1. 64k is better than 16k. Now my cpu load is up to 60% though.
2. Varying xo point between 2600 and 3800 Hz did not make much difference. I'm using 2600 atm but may move that around later.
3. Adding a subsonic cutoff in the low pass filter made a HUGE difference in low end quality. HUGE! Now, there is nothing about the sound that I could complain about.
4. Adding a sound absorption panel on the wall to intercept that reflection made a nice difference in soundstage.
I will be sure to detail how to create these filters in my final writeup. Let me know if you want the code (Octave) for these filters and I'll post it right away. I am having great luck with Octave and find it calculates 2 64k filters in about 3 seconds on my Athlon 64 3400+. Never crashes.
-Robert
mhelin,
Thanks for the reference on speakers. I might look further into the 24L cabinet monitors. They look promising.
However, first things first. I did some experimenting this morning with creating various filters and testing them. Here are the results:
1. 64k is better than 16k. Now my cpu load is up to 60% though.
2. Varying xo point between 2600 and 3800 Hz did not make much difference. I'm using 2600 atm but may move that around later.
3. Adding a subsonic cutoff in the low pass filter made a HUGE difference in low end quality. HUGE! Now, there is nothing about the sound that I could complain about.
4. Adding a sound absorption panel on the wall to intercept that reflection made a nice difference in soundstage.
I will be sure to detail how to create these filters in my final writeup. Let me know if you want the code (Octave) for these filters and I'll post it right away. I am having great luck with Octave and find it calculates 2 64k filters in about 3 seconds on my Athlon 64 3400+. Never crashes.
-Robert
Wow, this thead is still going....
Well, ill update my setup info then as its now all working.
This is my new DAC+AMP that im now running:
6 Channels all up, 3 optical SPDIF in, 2 level volume attenuators.
The music source is lossless encoded audio on a Linux box.
MPlayer then plays this into JACK which then goes into BruteFIR for xovers and DRC which is then output to my modified soundcard.
I am using the Magnesium coned Seas Excel drivers for mids, and Seas aluminum domed tweeters, currently crossed over at 1.8Khz (woofer cone breakup occours at 3Khz).
Sub is a 12" carbon fiber "response" unit in 75L ported box.
Im using xover filters created with Matlab scripts... current filter length is 8192.
BruteFIR is chewing ~10% CPU time with these filters.
I was going to use Octave to create the filters, but its missing some of the nicer functions matlab has (like fircls1 that im using now).
So your putting the stereo input through a highpass before the other xover filters?
What is the -3db frequency of it?
Well, ill update my setup info then as its now all working.
This is my new DAC+AMP that im now running:
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
6 Channels all up, 3 optical SPDIF in, 2 level volume attenuators.
The music source is lossless encoded audio on a Linux box.
MPlayer then plays this into JACK which then goes into BruteFIR for xovers and DRC which is then output to my modified soundcard.
I am using the Magnesium coned Seas Excel drivers for mids, and Seas aluminum domed tweeters, currently crossed over at 1.8Khz (woofer cone breakup occours at 3Khz).
Sub is a 12" carbon fiber "response" unit in 75L ported box.
Im using xover filters created with Matlab scripts... current filter length is 8192.
BruteFIR is chewing ~10% CPU time with these filters.
I was going to use Octave to create the filters, but its missing some of the nicer functions matlab has (like fircls1 that im using now).
Adding a subsonic cutoff in the low pass filter made a HUGE difference in low end quality. HUGE! Now, there is nothing about the sound that I could complain about.
So your putting the stereo input through a highpass before the other xover filters?
What is the -3db frequency of it?
Re: 64k filters
Hi Robert,
I for one would be very interested in your octave code. Right now I was transcribing some of the Hawksford matlab code on the same subject.
Regards,
Roberto
RFScheer said:
I will be sure to detail how to create these filters in my final writeup. Let me know if you want the code (Octave) for these filters and I'll post it right away. I am having great luck with Octave and find it calculates 2 64k filters in about 3 seconds on my Athlon 64 3400+. Never crashes.
-Robert
Hi Robert,
I for one would be very interested in your octave code. Right now I was transcribing some of the Hawksford matlab code on the same subject.
Regards,
Roberto
Loudspeakers and x-over freq
Those small midwoofers can and should be crossed as high as possible as the directivity of the woofer then matches that of the tweeter also off-axis. If you have a larger woofer (6.5" or bigger) you have to cross at lower frequencies because it's there where the off-axis frequency response matches best. If there is a sudden jump or notch in off-axis response it is bad as it means that the sound waves reflected from walls have missing energy at that frequencies and in you listening position the response isn't as flat anymore; it's the power response (which is the average of the on-axis and off-axis responses) which should be made flat, not just the on-axis response.
btw. Robert, the DD8-MkII is only a 12L monitor, but the author recommends also using 24L box and lower tuning. Then you have quite a big monitor, but you can increase the depth as much as is needed, or you can build it floorstander.
Regarding the IR lenght, many suggests using very short (less than 100) samples for the filter (the LPF and HPF in midrange). For low frequencies you need longer filter. Even though it's possible to use brick-wall filters which are really steep the less extreme 48-52 db/oct filters may actually sound better (and need less CPU%). For DRC (if you want affect low frequencies) ~8k is enough for decent results (with 64k you get better, flatter response).
Those small midwoofers can and should be crossed as high as possible as the directivity of the woofer then matches that of the tweeter also off-axis. If you have a larger woofer (6.5" or bigger) you have to cross at lower frequencies because it's there where the off-axis frequency response matches best. If there is a sudden jump or notch in off-axis response it is bad as it means that the sound waves reflected from walls have missing energy at that frequencies and in you listening position the response isn't as flat anymore; it's the power response (which is the average of the on-axis and off-axis responses) which should be made flat, not just the on-axis response.
btw. Robert, the DD8-MkII is only a 12L monitor, but the author recommends also using 24L box and lower tuning. Then you have quite a big monitor, but you can increase the depth as much as is needed, or you can build it floorstander.
Regarding the IR lenght, many suggests using very short (less than 100) samples for the filter (the LPF and HPF in midrange). For low frequencies you need longer filter. Even though it's possible to use brick-wall filters which are really steep the less extreme 48-52 db/oct filters may actually sound better (and need less CPU%). For DRC (if you want affect low frequencies) ~8k is enough for decent results (with 64k you get better, flatter response).
"What the future holds is little black boxes that can enhance loudspeakers and do some DSP things that will improve the performance of the loudspeaker for both home-theater and stereo applications"
Paul Barton at Sterophile interview, October, 1997
Interesting stuff to read on speaker design.
Paul Barton at Sterophile interview, October, 1997
Interesting stuff to read on speaker design.
Re: Loudspeakers and x-over freq
Strange...
Looking at the datasheets for the W18E001 and 25TAC/D (these are what im using), wouldnt it be best to xover low (like 1.5Khz) to reduce off axis drop off that appears between 1.5Khz and 3Khz on the woofer?
W18E001: http://www.seas.no/excel_line/excel/E0018.PDF
25TAC/D: http://www.seas.no/seas_line/tweeters/H535.PDF
If i xovered at 3Khz, the woofers off-axis levels would be very different to those of the tweeter.
What you say makes sense in a way, i just cant see how setting a higher xover would reduce the woofer->tweeter off-axis step.
mhelin said:Those small midwoofers can and should be crossed as high as possible as the directivity of the woofer then matches that of the tweeter also off-axis. If you have a larger woofer (6.5" or bigger) you have to cross at lower frequencies because it's there where the off-axis frequency response matches best. If there is a sudden jump or notch in off-axis response it is bad as it means that the sound waves reflected from walls have missing energy at that frequencies and in you listening position the response isn't as flat anymore; it's the power response (which is the average of the on-axis and off-axis responses) which should be made flat, not just the on-axis response.
Strange...
Looking at the datasheets for the W18E001 and 25TAC/D (these are what im using), wouldnt it be best to xover low (like 1.5Khz) to reduce off axis drop off that appears between 1.5Khz and 3Khz on the woofer?
W18E001: http://www.seas.no/excel_line/excel/E0018.PDF
25TAC/D: http://www.seas.no/seas_line/tweeters/H535.PDF
If i xovered at 3Khz, the woofers off-axis levels would be very different to those of the tweeter.
What you say makes sense in a way, i just cant see how setting a higher xover would reduce the woofer->tweeter off-axis step.
- Status
- Not open for further replies.
- Home
- Source & Line
- Digital Source
- PC music players