Regarding improving soundstage via acoustic design (or by reducing cabinet diffraction effects)

[Vineethkumar01]

Hi

I found this informative thread by Patrick Bateman (He posts gazillions of posts regarding waveguides and other related things on diyaudio)

https://www.diymobileaudio.com/threads/improve-your-soundstage-for-2.65061/page-32

Even though the above thread is in the context of car audio, I find the subject matter relevant to home audio as well.
In a nutshell, his opinion (seems to be the opinion of other well known people including Earl Geddes too) seems that diffraction-related effects makes themselves known at higher volumes causing listening fatigue etc.
So the amount of care we put into acoustic design of the speaker cabinet (or the cabinet shape and baffle shape and other things which influence radiation characteristics of the drivers) will really matter especially at higher playback levels (Just considering the speaker design aspect alone here and not considering effects of room treatment and other things).

(also posting link to Geddes site for anyone interested in learning about his perspectives to all things audio: http://www.gedlee.com/Papers/papers.aspx)

Thanks
Vineeth

 

[Kannan]

Patrick is highly respected in the car audio diy segment. The mids n tweets in car are usually fixed to the A Panel and sometime to the dashboard or to the doors and are in close proximity to the listener.
Added to this the problems in controlling the baffle reflections due to space constrains for building proper enclosures is another big caveat.
The other problem is acres of metal and glass.
So active DSPs have been long popular in car audio from the 80s as they offer lots of control and tweaking.

 

[keith_correa]

After reading more than necessary about all of this over the years, I came to the conclusion that this is a rabbit hole and there are varied theories and measurements on this. I then threw my hands up & gave up! There are certain "best practices" and RoT to be employed and that is enough. For me at least!

 

[Vineethkumar01]

Thanks @Kannan and @keith_correa for your perspectives
I don't know anything about Car audio, but I can imagine the much greater challenge it can present when one wants to set up a good sound system due the aspects like reflections.

Anyway, I know that the diffraction problem has been beaten to death in the past..
Just sharing my thoughts on this.
I keep posting about these things mostly for my own education and others interested about the engineering aspects behind them..
The issue with diffraction is that it is a fundamental problem to be solved mostly in the acoustic domain. So if we try to solve it completely in the electrical domain, it may result in undesired results, if we are unlucky..
Most frequently, especially for beginners, what is usually done is look at an on-axis frequency response measurement or a slightly off-axis one, identify the peaks and dips and try to flatten it into a straight line in the crossover design or at least that is how I started..
This is especially true with access to cheap DSP processors. The amount of EQ that can be done to the driver + cabinet response is just incredible these days. However, diffraction (whether it is caused by cabinet, or baffle, or something else) affects not only on-axis radiation but also off-axis behavior and that is particularly important with speaker placement in typical rooms. Often if one doesn't carefully apply EQ, it will screw up the off axis radiation or the overall directivity characteristics. With the availability of softwares like VituixCAD this can be easily seen and believed. We even have access to powerful wave front simulators these days which help us a lot in visualizing the problem in 3D rather than just one or two frequency responses in certain directions.

Regarding the solution to the diffraction problem, people have time and again showed that the typical shoe-boxed sized enclosures are not the best enclosure designs for housing the drivers. Ideally we want to hear just the driver doing its thing. Due to diffraction the shape of the cabinet the driver is housed is also seriously affects the overall response. So to deal with that ideally the first step we should take is to make the shape of the housing aid the driver do its thing and get out of the way. One of the reasons most high end companies do curved and rounded kind of cabinets for this reason (other than aesthetics). Even with cheap but good quality drivers, great speakers can be made with a great acoustic design but with bad acoustic design, even the best drivers are a little lost in terms of the level of performance they can reach. DSP can be employed to address many issues but acoustic design issues should be the last thing we should tell the DSP to handle.

I feel that in the past, one of the most common reasons why enclosure design was chosen as a typical cuboid was because of the difficulty in manufacturing good shapes. However, with great softwares becoming available for free for enclosure design, crossover design, availability of CNC and 3D printing etc, it is easier to fabricate good acoustic designs better now than ever before. And I think we should really make use of it and deal with fundamental problems like diffraction at the level that it should be solved...

Off course, no subjectivity explored here.

 

[Decadent_Spectre]

Haven't read the whole thread but Geddes as I recall was a proponent of rounded waveguides (horns?) and foam.

 

[Vineethkumar01]

Haven't read the whole thread but Geddes as I recall was a proponent of rounded waveguides (horns?) and foam.

Yes. From what I know, his speakers feature big waveguides with foam (not calling it a horn since I don't know the in-depth design considerations/differences between the two yet), generously rounded baffles, constrained layer damping-based bracing schemes etc.

 

[Decadent_Spectre]

Yes. From what I know, his speakers feature big waveguides (not calling it a horn since I don't know the in-depth design considerations/differences between the two yet), generously rounded baffles, constrained layer damping-based bracing schemes etc.

I think waveguides are horns or function as such depending upon their size/mouth. They tend to be smaller so "horn loading" is less and down to a higher frequency. AFAIK a horn provides directivity and efficiency while a waveguide may be geared more towards directivity to avoid "horn shout" and other "problems". Yes rounded baffles/corners/waveguides are a thing among some members of the DIY community.

If you like diyaudio, consider reading weltersys, he is very objective and has tons of experience in this. I reckon you may enjoy his posts with your particular train of thought.

Edit: Reach out to him, he is generally very helpful if you wish to get some advice for your work.

 

[Vineethkumar01]

I think waveguides are horns or function as such depending upon their size/mouth. They tend to be smaller so "horn loading" is less and down to a higher frequency. AFAIK a horn provides directivity and efficiency while a waveguide may be geared more towards directivity to avoid "horn shout" and other "problems". Yes rounded baffles/corners/waveguides are a thing among some members of the DIY community.

If you like diyaudio, consider reading weltersys, he is very objective and has tons of experience in this. I reckon you may enjoy his posts with your particular train of thought.

Edit: Reach out to him, he is generally very helpful if you wish to get some advice for your work.

Thank you for the info.
I have read a few posts by weltersys and his projects. But I don't think I am ready to get into the horns' world yet.
I like kimmosto's approaches to directivity and overall speaker design for now.

 

[Decadent_Spectre]

Thank you for the info.
I have read a few posts by weltersys and his projects. But I don't think I am ready to get into the horns' world yet.
I like kimmosto's approaches to directivity and overall speaker design for now.

Sure work on your own pace.

I will say this, nothing sounds like a large well designed horn. Nothing.

Love it or hate it

 

[Vineethkumar01]

Sure work on your own pace.

I will say this, nothing sounds like a large well designed horn. Nothing.

Love it or hate it

Thanks. I have heard this from many people about horns..
I like them too but just too much to accommodate at home currently due to "other" factors/people for whom looks and compactness are too much important.

 

[Decadent_Spectre]

For looks compactness I could only suggest power density and smaller waveguides, that would be my approach. If output is not required at high levels you can skip power density. DSP is mandatory if you ask me

You may also consider angling the mid/woof ala some pro audio designs in a sort of WTW configuration with slot ports on the top and bottom. This is assuming some form of CD. A regular dome might require a different approach. I am not well versed enough with home audio designs to offer more there.

We'll paint it red!

 

[Vineethkumar01]

For looks compactness I could only suggest power density and smaller waveguides, that would be my approach. If output is not required at high levels you can skip power density. DSP is mandatory if you ask me

You may also consider angling the mid/woof ala some pro audio designs in a sort of WTW configuration with slot ports on the top and bottom. This is assuming some form of CD. A regular dome might require a different approach. I am not well versed enough with home audio designs to offer more there.

We'll paint it red!

Thank you. If possible, could you point me to some link/pic which has a similar pro-audio design. If not now it will be useful for me in future.
My current project part of which I have also documented on this forum has only a small waveguided loaded tweeter-mid-woofer combination with possibly a shaped cabinet.
DSP is a must for me too.
I love DSP so much that I have got a PhD in it and work on DSP for wireless chips day in and day out. That love just grows more everyday..
Anyway, if I do a next personal project, i hope it would be a proper horn-based design.

 

[Decadent_Spectre]

Thank you. If possible, could you point me to some link/pic which has a similar pro-audio design. If not now it will be useful for me in future.
My current project part of which I have also documented on this forum has only a small waveguided loaded tweeter-mid-woofer combination with possibly a shaped cabinet.
DSP is a must for me too.
I love DSP so much that I have got a PhD in it and work on DSP for wireless chips day in and day out. That love just grows more everyday..
Anyway, if I do a next personal project, i hope it would be a proper horn-based design.

Something like this, though I can't find many more, I had seen these years ago so my apologies but I can't find the one I had thought of.

ULTRA-X40

Compact Loudspeakers

meyersound.com

I think the idea may be to get high output from a smaller enclosure, while keeping it more cohesive and also with good directivity to a lower frequency, though this would demand larger enclosures.

That's very interesting to hear. If one wishes to introduce phase distortion, "lagging" of the signal output in audio DSP how best to do this apart from a regular phase filter? All pass? Any "tricks"?

I look forward to your horn project and do hope it materializes. But remember go big or go home

 

[Vineethkumar01]

Something like this, though I can't find many more, I had seen these years ago so my apologies but I can't find the one I had thought of.

ULTRA-X40

Compact Loudspeakers

meyersound.com

I think the idea may be to get high output from a smaller enclosure, while keeping it more cohesive and also with good directivity to a lower frequency, though this would demand larger enclosures.

That's very interesting to hear. If one wishes to introduce phase distortion, "lagging" of the signal output in audio DSP how best to do this apart from a regular phase filter? All pass? Any "tricks"?

I look forward to your horn project and do hope it materializes. But remember go big or go home

Thank you. Your link looks very interesting. I will study more about this..
Usually all pass filters are used to introduce pure delay if one doesn't wish to modify the amplitude. I am sorry that I don't understand what are regular phase filters. I know about minimum phase, linear phase etc..
All my life, I have been trying to study how not to add distortion/live with just the amount of distortion the signal already has..
So your question about how to add distortion to signals took me by surprise..
I think people who do mixing in studios etc might know just the right way to create controlled distortion tailored to audio applications.
A simple google search gave this:

I haven't heard it but can try to listen and understand this over free time..

I too hope I get a chance to work on a horn project sometime in the future once I and others at home mature to the overall taste of it..

Edit: In fact thinking about it, if one has access to FIR filters with high enough taps possible in DSP (instead of regular IIR filters), one can add arbitrary phase shifts to tones/groups of tones. This way one has complete control over distortion added..
But this needs serious processing capabilities

 

[Decadent_Spectre]

By phase filters I mean filters that adjust phase in degrees.

What do you mean by tones? Frequency?

Have you investigated Accourate and Audiolense? I think those were their names.

 

[Vineethkumar01]

By phase filters I mean filters that adjust phase in degrees.

What do you mean by tones? Frequency?

Have you investigated Accourate and Audiolense? I think those were their names.

I have heard about Accourate and Audiolense. I think they have FIR processing capabilities. I haven't used them though. Atleast Acourate seems to have it as per their website. If it is PC-BASED DSP reasonable length FIRs are a piece of cake for PCs these days. I understood about phase. I was just confused by the term 'regular' phase filter. Anyway that doesn't matter. Just a matter of terminology.

By a tone, I mean a sinewave of a particular frequency. So when I say a 400Hz tone, I mean a 400 Hz sinewave.

There is this software called "Rephase". diyaudio has a thread about it. It is open source. One can define FIR filters and IIR filters in it (or simple FIRs using it was all I have played with for creating FIR filters in audio). In general people use this software for creating FIR filters for "phase correction" after all the driver passband linearization filters + crossover filters work on the audio signal in a DSP crossover implementation.
So I thought why not use the software for reverse process. That is to change phases arbitrarily as per one's desire. In general, the higher the number of taps you have with FIR filter, the higher is the resolution of the filter and the greater control one can have in adding required (amplitude and) phase shifts to different frequencies.
Once the filters are created one can use a convolver software/some plugin (I have not used these much yet) to apply the filtering on the signal. I have used only EQAPO software for creating some custom filters for EQ (and they were IIR). I think convolvers might be available in JRIVER/FOOBAR/other softwares.

Edit: By the way I have used CamillaDSP on Raspbery pi (inside Moode audio software) to create custom FIR filters for playing around with Bass reflex alignment modification using DSP. This will work with FIR filters for sure. My signal chain was moode audio streaming the audio signal after applying the custom filters i designed (in CamillaDSP) through my speakers having a passive crossover.

 

[Decadent_Spectre]

I have heard about Accourate and Audiolense. I think they have FIR processing capabilities. I haven't used them though. Atleast Acourate seems to have it as per their website. If it is PC-BASED DSP reasonable length FIRs are a piece of cake for PCs these days. I understood about phase. I was just confused by the term 'regular' phase filter. Anyway that doesn't matter. Just a matter of terminology.

By a tone, I mean a sinewave of a particular frequency. So when I say a 400Hz tone, I mean a 400 Hz sinewave.

There is this software called "Rephase". diyaudio has a thread about it. It is open source. One can define FIR filters and IIR filters in it (or simple FIRs using it was all I have played with for creating FIR filters in audio). In general people use this software for creating FIR filters for "phase correction" after all the driver passband linearization filters + crossover filters work on the audio signal in a DSP crossover implementation.
So I thought why not use the software for reverse process. That is to change phases arbitrarily as per one's desire. In general, the higher the number of taps you have with FIR filter, the higher is the resolution of the filter and the greater control one can have in adding required (amplitude and) phase shifts to different frequencies.
Once the filters are created one can use a convolver software/some plugin (I have not used these much yet) to apply the filtering on the signal. I have used only EQAPO software for creating some custom filters for EQ (and they were IIR). I think convolvers might be available in JRIVER/FOOBAR/other softwares.

Edit: By the way I have used CamillaDSP on Raspbery pi (inside Moode audio software) to create custom FIR filters for playing around with Bass reflex alignment modification using DSP. This will work with FIR filters for sure. My signal chain was moode audio streaming the audio signal after applying the custom filters i designed (in CamillaDSP) through my speakers having a passive crossover.

I think I have heard of Rephase as well somewhere. Just remembered a company called DEQX for DSP as well.

Yes JRiver works well for this, it is what I use as my music/video player. It also has a lot of it's own processing available, not sure if it does FIR.

If one wants to distort phase, I wonder if IIR is actually preferred.

 

[Vineethkumar01]

I think I have heard of Rephase as well somewhere. Just remembered a company called DEQX for DSP as well.

Yes JRiver works well for this, it is what I use as my music/video player. It also has a lot of it's own processing available, not sure if it does FIR.

If one wants to distort phase, I wonder if IIR is actually preferred.

I haven't used any dedicated DSP processors at home yet like MiniDSP or DEQx or any other ones. If it is IIR implementation, I think those will suffice.
The most I have done yet is implement a full active 2 way (or 3 way) crossover in JRIVER with driver passband linearization based on measured on horizontal polar measurements and filters designed in VituixCAD (all IIR filters). This is all done via my 6 channel ESI U86XT audio interface as the hardware.

Preference of FIR vs IIR, I have no idea. Given enough taps, one can, in theory, get close to any IIR transfer function with FIR (in magnitude and phase). Hence, maybe one has to hear and compare these two regarding if/how they sound different and what one likes. Implementationwise IIR is more familiar and widely used. FIRs are less common yet.

 

[Decadent_Spectre]

Do you mean you use JRivers internal DSP for the filters?

Horizontal? Any reason you did not use the verticals?

I'd not thought about the IIR FIR like that, interesting. I wonder if they would sound the same, also if one needs IIR then it makes more sense to just use that since its easier and accomplishes the goal.

 

[Vineethkumar01]

Do you mean you use JRivers internal DSP for the filters?

Horizontal? Any reason you did not use the verticals?

I'd not thought about the IIR FIR like that, interesting. I wonder if they would sound the same, also if one needs IIR then it makes more sense to just use that since its easier and accomplishes the goal.

Yes. I used JRIVER's DSP studio for implementing a DSP crossover + EQ. Attached pic-1 shows the filters I wanted to implement. Pic 2 shows a snapshot of JRiver's DSP studio in which I implemented it. It works well.

Regarding taking only horizontal, I was lazy to do more measurements with the prototype box shown in pic-3 . My goal was just to convice myself that Jriver can do this. In any real build I will take horizontal and vertical polars.

If one can's purpose gets fulfilled with IIRs, it is the way to go. Less processing power required and nice transfer functions can be obtained in magnitude response. However one has to live the phase response yielded by the magnitude response. WIth FIRs one can tune magnitude response and phase response independently of each other. To one's heart's content. Penalty is more processing power.