Hello everyone,
I’ve recently started a company focused on high-impact house parties and drum & bass rave events. With a background in software engineering and a love for system design, I’ve been diving deep into live sound, especially subwoofers, and I’ve come up with a novel solution I’d love to share and get feedback on.
For my last event, I built and deployed a 20 Hz tuned tapped horn to deliver deep sub energy from 20 to 50 Hz. As you probably know, tapped horns are monsters when it comes to efficiency and SPL per watt. But they come with a tradeoff: they lack transient clarity and can sound sluggish or boomy compared to sealed or front-loaded designs.
That’s where my idea comes in.
I’ve thought about a modular correction system that adds a small secondary driver into the horn path, which is fed a predictive, DSP-shaped signal designed to:
The correction driver acts more like a subwoofer tweeter injecting carefully timed bursts or phase-inverted pulses to clean up decay and enhance punch. Think: servo sub tightness inside a horn-loaded monster. The driver will be smaller with a smaller MMS.
It will be DSP-driven, with a signal split and tuned either manually or with a convolution-based predictive model. Long term, I want to open this up to the community, make it modular, and maybe even build AI-driven tuning tools to match different horn alignments.
Would love to hear your thoughts — has anyone tried anything similar? Interested in collaborating on testing or refining the idea?
Thanks!
I’ve recently started a company focused on high-impact house parties and drum & bass rave events. With a background in software engineering and a love for system design, I’ve been diving deep into live sound, especially subwoofers, and I’ve come up with a novel solution I’d love to share and get feedback on.
For my last event, I built and deployed a 20 Hz tuned tapped horn to deliver deep sub energy from 20 to 50 Hz. As you probably know, tapped horns are monsters when it comes to efficiency and SPL per watt. But they come with a tradeoff: they lack transient clarity and can sound sluggish or boomy compared to sealed or front-loaded designs.
That’s where my idea comes in.
I’ve thought about a modular correction system that adds a small secondary driver into the horn path, which is fed a predictive, DSP-shaped signal designed to:
- Sharpen transients
- Cancel low-end ringing
- And tighten up timing below 60 Hz — without sacrificing the tapped horn’s natural output
The correction driver acts more like a subwoofer tweeter injecting carefully timed bursts or phase-inverted pulses to clean up decay and enhance punch. Think: servo sub tightness inside a horn-loaded monster. The driver will be smaller with a smaller MMS.
It will be DSP-driven, with a signal split and tuned either manually or with a convolution-based predictive model. Long term, I want to open this up to the community, make it modular, and maybe even build AI-driven tuning tools to match different horn alignments.
Would love to hear your thoughts — has anyone tried anything similar? Interested in collaborating on testing or refining the idea?
Thanks!
care to share this design or at least let us see your creation?
really?? if this is true why do they garner so much love and attention??
Great questions!
The tapped horn I built is based on the original Alpine 12 tapped horn design posted here: https://www.stereonet.com/forums/topic/282701-alpine-12-tapped-horn-home-theatre-subwoofer-build/. I used that as a foundation and tuned mine to 20 Hz to emphasize the sub-bass energy from 20 to ~50 Hz, which is critical for the kind of drum and bass/live bass music I run at events.
And you're right. Tapped horns offer incredible efficiency, low distortion, and deep extension, especially for DIYers who want max output per watt. But that doesn’t mean they’re perfect. One tradeoff is time-domain performance: they can exhibit longer group delay and modal ringing in the sub region, which can make transients feel a little “blurred” or smeared, especially compared to sealed boxes or front-loaded horns.
That said, the benefits usually outweigh the drawbacks for most people. My goal with this new system is to retain the output and extension of the tapped horn. But clean up defects in the signal with small transients bursts or cancellation.
I will put a secondary driver with a sealed back, firing through a very short port or waveguide halfway up the horn.
The entire system becomes a tunable, intelligent acoustic device. A convolutional neural network (CNN) will be used during development to learn an optimized, low-latency function that transforms the incoming audio into a correction signal for the secondary driver.
This correction function will aim to:
A. Detect and reinforce fast transients, generating sharper pressure peaks in the horn output.
B. Inject phase-inverted signals to cancel modal ringing or low-frequency artifacts that would otherwise smear the response.
But this approach isn’t limited to those two tasks. By comparing the predicted output (from the dry input signal) to the actual pressure waveform measured at the horn mouth, the CNN can learn to minimize the error by discovering complex interactions between the horn, main driver, correction driver, and acoustic environment. In theory, this could lead to a system that completely corrects time-domain distortion, unlocking levels of transient control previously impossible in horn-loaded designs.
Thanks again for the interest. Below is a picture of the horn. (The writing on it was from the crowd at one of my parties)
The tapped horn I built is based on the original Alpine 12 tapped horn design posted here: https://www.stereonet.com/forums/topic/282701-alpine-12-tapped-horn-home-theatre-subwoofer-build/. I used that as a foundation and tuned mine to 20 Hz to emphasize the sub-bass energy from 20 to ~50 Hz, which is critical for the kind of drum and bass/live bass music I run at events.
And you're right. Tapped horns offer incredible efficiency, low distortion, and deep extension, especially for DIYers who want max output per watt. But that doesn’t mean they’re perfect. One tradeoff is time-domain performance: they can exhibit longer group delay and modal ringing in the sub region, which can make transients feel a little “blurred” or smeared, especially compared to sealed boxes or front-loaded horns.
That said, the benefits usually outweigh the drawbacks for most people. My goal with this new system is to retain the output and extension of the tapped horn. But clean up defects in the signal with small transients bursts or cancellation.
I will put a secondary driver with a sealed back, firing through a very short port or waveguide halfway up the horn.
The entire system becomes a tunable, intelligent acoustic device. A convolutional neural network (CNN) will be used during development to learn an optimized, low-latency function that transforms the incoming audio into a correction signal for the secondary driver.
This correction function will aim to:
A. Detect and reinforce fast transients, generating sharper pressure peaks in the horn output.
B. Inject phase-inverted signals to cancel modal ringing or low-frequency artifacts that would otherwise smear the response.
But this approach isn’t limited to those two tasks. By comparing the predicted output (from the dry input signal) to the actual pressure waveform measured at the horn mouth, the CNN can learn to minimize the error by discovering complex interactions between the horn, main driver, correction driver, and acoustic environment. In theory, this could lead to a system that completely corrects time-domain distortion, unlocking levels of transient control previously impossible in horn-loaded designs.
Thanks again for the interest. Below is a picture of the horn. (The writing on it was from the crowd at one of my parties)
Attachments
it's an interesting premise i just don't know if your proposed solution would succeed.
i may be wrong but it seems that your complaint could be resolved with proper system alignment.
i may be wrong but it seems that your complaint could be resolved with proper system alignment.
Good choice.
Yes, combining the output of the front and back of the drivers separated in time by the path length may have sonic implications.
The "transient response" has two arrival times.
The "low end ringing" in your sub is primarily above your 20-50Hz pass band.
"Tigtening up" the timing of the front and rear output is the equivalent of eliminating the advantage the TH provides.
Give it a try- you will find what occurs is not what you propose.
Yes.
Not more than I already have 😉
Art
If the time-domain distortion in a subwoofer were to be eliminated, we then are left with the "elephant in the room", the room's low frequency time domain problems dwarf those of any sub.
When elephants use sounds 20Hz and below for communication, they don't have to worry about a room destroying their phase relationships.
Art
Thanks for the thoughtful response, Art. You raise some important points. Let me clarify what I’m proposing and how it differs from what it sounds like you're interpreting.
First, I agree that the tapped horn's quarter-wave principle gives it massive output by summing front and rear wave energy, that’s the point of the design. But the cost of that efficiency is time-domain compromise, especially in the first few cycles of a transient where phase alignment isn’t fully developed.
This is where the correction driver comes in not to cancel sustained energy (which would indeed undo the horn’s gain), but to reinforce or realign early-cycle transients where the horn hasn’t yet fully “charged” its pressure path. Think of it like restoring the first punch in a drum hit, before the wavefront has had time to build resonance in the horn path.
A quarter-wave resonator is inherently slow to build energy. It emphasizes sustained tones but undershoots fast transients. That’s why some tapped horns feel “lazy” or bloomy, even if they measure flat in the frequency domain. I'm addressing that gap: not full-band cancellation, but targeted injection during the first ~10–50 ms of a hit.
To be clear:
Instead, I’m shaping the pressure at specific timing intervals to reduce:
As for room correction. Yes, the room dominates below 100 Hz. But what I’m proposing is not a substitute for room correction. It's a way to enable a cleaner source signal, which makes room correction more effective by starting from a faster, more controlled transient.
In short: I'm not trying to erase the horn's nature. I'm trying to give it the punch it often lacks at the moment it matters most.
Happy to dive into data or tuning results as I build and test this out!
First, I agree that the tapped horn's quarter-wave principle gives it massive output by summing front and rear wave energy, that’s the point of the design. But the cost of that efficiency is time-domain compromise, especially in the first few cycles of a transient where phase alignment isn’t fully developed.
This is where the correction driver comes in not to cancel sustained energy (which would indeed undo the horn’s gain), but to reinforce or realign early-cycle transients where the horn hasn’t yet fully “charged” its pressure path. Think of it like restoring the first punch in a drum hit, before the wavefront has had time to build resonance in the horn path.
A quarter-wave resonator is inherently slow to build energy. It emphasizes sustained tones but undershoots fast transients. That’s why some tapped horns feel “lazy” or bloomy, even if they measure flat in the frequency domain. I'm addressing that gap: not full-band cancellation, but targeted injection during the first ~10–50 ms of a hit.
To be clear:
- I’m not cancelling the continuous wave, which builds constructively over the horn length
- I’m not flattening the horn’s response, or destroying gain
- I’m not trying to "tighten" front-to-rear path time alignment, I’m working with it, not against it
Instead, I’m shaping the pressure at specific timing intervals to reduce:
- The group delay tail
- Overshoot
- And first-cycle underrepresentation, especially below 60 Hz
As for room correction. Yes, the room dominates below 100 Hz. But what I’m proposing is not a substitute for room correction. It's a way to enable a cleaner source signal, which makes room correction more effective by starting from a faster, more controlled transient.
In short: I'm not trying to erase the horn's nature. I'm trying to give it the punch it often lacks at the moment it matters most.
Happy to dive into data or tuning results as I build and test this out!
@BlakeJones
Reading about transient response is always interesting.
Did you consider before adding a "corrector" driver (any images of the intended construction?) to correct with dsp techniques only the main driver first? Like FIR.
I know there are PA loudspeaker cones which are more modern than paper cones but saw these rarely used.
As impulse response is also a question of cone stiffness - did you check if the kind of driver you use excels here?
Reading about transient response is always interesting.
Did you consider before adding a "corrector" driver (any images of the intended construction?) to correct with dsp techniques only the main driver first? Like FIR.
I know there are PA loudspeaker cones which are more modern than paper cones but saw these rarely used.
As impulse response is also a question of cone stiffness - did you check if the kind of driver you use excels here?
now i'm really confused...here i thought that a transient was by definition a one time short duration event that should not have any cycles...
What I meant by “the first few cycles” is actually the system's response to the transient, not the transient itself. In tapped horns, especially, there’s a bit of a ramp-up period where the horn loads and builds pressure, particularly around its resonant frequencies. That response takes time usually measured in cycles of the system’s tuning.
So, while the transient itself is one-shot, the horns output from that transient includes some slow buildup that’s what I’m aiming to reinforce or correct. My phrasing could have been clearer. T
hanks for catching that.
So, while the transient itself is one-shot, the horns output from that transient includes some slow buildup that’s what I’m aiming to reinforce or correct. My phrasing could have been clearer. T
hanks for catching that.
I have played around with the default iir filtering in the minidsp2x4. Along side the Dirac live filters which acts as a fir filter to achieve room correction and a flat response. I believe it aligns the group delay across the entire spectra. I believe the phase and impulse response need to be aligned properly for each crossover element at the crossover frequency first to ensure alignment as Dirac processes the output signal as one signal. Overall I have Achieved a great sounding system. With improvements to be made soon in my midbass region when I construct a front loaded horn kick bin array.
And referring to driver selection. I chose a driver that models well against my enclosure in hornresp for frequency response, spl and extension to 20hz. I should have looked more in to the T/S params and the transient ability but the drivers recieved many accolades online so I assumed they were appropriate. And which drivers are you referring to? I am interested. I believe velodyne acoustics makes a product with a realtime feedback loop that attenuates driver excursion.
@BlakeJones
In my eyes adding an extra correcting driver is the more difficult way than applying some elaborate dsp correcting the time domain and or f response on one driver.
I saw PA drivers with carbon fiber cones which are much stiffer than any paper.
If you want to make diy an existing driver more stiff here some info
its then for free
diy stiffening like tannoy or Mr Reck see post in the thread diy stiffening techniques
In my eyes adding an extra correcting driver is the more difficult way than applying some elaborate dsp correcting the time domain and or f response on one driver.
I saw PA drivers with carbon fiber cones which are much stiffer than any paper.
If you want to make diy an existing driver more stiff here some info
How to DIY making a paper cone (or others) more stiff.
Here first example:
fast drying hairspray liquid
Never apply too much at once so the paper fibers do not become wet and weak and throw wrinkles!
Better less but several times applying
then if possible fast dry in the sun
Here first example:
fast drying hairspray liquid
Never apply too much at once so the paper fibers do not become wet and weak and throw wrinkles!
Better less but several times applying
then if possible fast dry in the sun
its then for free
diy stiffening like tannoy or Mr Reck see post in the thread diy stiffening techniques
Attachments
I believe that the blurring of transients to which you refer is caused not by any 'slow buildup...' of the output, but by the inevitable long decay of the energy subsequently built up in poorly damped resonant systems present in loadings such as tapped horn and ported. Even if we take a couple of cycles to achieve full output, this is as nothing compared with step response tests showing enclosures still resonating 200ms or more after the signal has ceased.
Perhaps we should be focusing on reducing transient decay, a very rarely considerd and significant failing of most currently implemented low frequency speaker loading techniques, but examined so closely at higher frequencies where pretty waterfall plots showing good damping and quick decay seem to be king...
So impulse compensated arrangement of two woofers vice versa would help, too.
If the design allows for it
A small driver won't have the displacement required to enhance "first-cycle underrepresentation, especially below 60 Hz".
https://data-bass.com/systems
A well designed 20Hz tapped horn's group delay tail doesn't exceed one cycle duration (the blue line) below ~60Hz:
Most studies show even 1.5 x cycle group delay, (the red line) is imperceptible under controlled A/B tests.
Your concept is based on the premise that "punch" below 60Hz is lacking and reduction of group delay will matter in your typical high-impact house parties and drum & bass rave events.
You could check your premise comparing SPL matched sealed woofers covering the same 20 to 50 Hz pass band as your tapped horn, using FIR filters to match phase and amplitude response.
The latency of those filters would be too long to be useful for live sound reinforcement, but not a problem for playback events.
Looking forward to seeing your test results!
Art
Great to see your enthousiasm.
As you might know, the impulse response and frequency response are related. You cannot change one without affecting the other, as they contain the same data. They are linked by the Fourier transform. Sharpening the impulse response tends to extend the bandwidth beyond 60 Hz. Or: if the system has resonances and you fix the frequency response (which includes phase) with a DSP, the ringing is also reduced.
As you might know, the impulse response and frequency response are related. You cannot change one without affecting the other, as they contain the same data. They are linked by the Fourier transform. Sharpening the impulse response tends to extend the bandwidth beyond 60 Hz. Or: if the system has resonances and you fix the frequency response (which includes phase) with a DSP, the ringing is also reduced.