Myth-Busting Hi-Fi: Tweeter Tricks and Off-Axis Myths

[Bob Rapoport]

The Myth:
Some manufacturers recommend turning speakers inward or outward to “smooth out” the treble response. The story goes that off-axis listening gives you a more natural sound.


The Truth:
Physics says otherwise. As frequency rises, wavelengths shorten. Bass spreads like a balloon, but treble beams forward like a flashlight. The only way to hear the full frequency response — especially in the high frequencies — is to listen on-axis.


When a manufacturer tells you to toe-in or toe-out unnaturally, it’s often to mask flaws such as tweeter ringing or resonances at the top end. In other words, it’s a design compromise, not a performance feature.


Why It Matters:
Stereo recordings are mixed to create a 3D soundstage, with instruments and voices placed precisely across the stage. If you’re listening off-axis, you’re throwing away localization cues and detail. On-axis, with careful setup, you hear what the artist and engineer intended.


The Takeaway:
Don’t let myths or marketing gimmicks steer you off course. Start with on-axis listening — the physics are on your side.

 

[w_sizer]

It depends…If you’re listening to speakers with an elevated in-room treble response, you can reduce listening fatigue by angling them slightly off-axis from the MLP. Moreover, some speakers (eg, ones with concentric drivers) are actually designed to be angled out by 10-15 degrees). See attached YT review of the KEF R3 Meta by Erin’s Audio Corner.

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

 

[Bob Rapoport]

It depends…If you’re listening to speakers with an elevated in-room treble response, you can reduce listening fatigue by angling them slightly off-axis from the MLP. Moreover, some speakers (eg, ones with concentric drivers) are actually designed to be angled out by 10-15 degrees). See attached YT review of the KEF R3 Meta by Erin’s Audio Corner.

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

"Some kind of diffraction element if you don't toe the speaker out 10-15 degrees" sounds to me like a design flaw being covered up. This means the highest frequencies are aimed away from the listener and end up as reflections that can interfere with the coherent soundwaves, smearing the stereo image. The ASR measured frequency response graph shows the tweeter output in red with a significant peak at 10 KHz with a rapid roll-off beyond. This may account for what the reviewer heard.

 

[JStewart]

This means the highest frequencies are aimed away from the listener and end up as reflections that can interfere with the coherent soundwaves, smearing the stereo image.

Likely not in this case due to the directivity, which is narrow and very even compared to most speakers and I think that was the point.

The ASR measured frequency response graph shows the tweeter output in red with a significant peak at 10 KHz with a rapid roll-off beyond. This may account for what the reviewer heard.

Amir on ASR measured the previous version, not the meta version.

I’d like think the correct approach for the average home listener is to first set speakers up on axis and then experiment with toe-in or out to suit. Folks have different tastes and different rooms, not to mention wildly different speakers.

 

[Bob Rapoport]

Likely not in this case due to the directivity, which is narrow and very even compared to most speakers and I think that was the point.



Amir on ASR measured the previous version, not the meta version.

I’d like think the correct approach for the average home listener is to first set speakers up on axis and then experiment with toe-in or out to suit. Folks have different tastes and different rooms, not to mention wildly different speakers.

The response chart above says its testing the Meta version. The top octave of audible sound in humans is 10KHz to 20KHz;. Most of it is like a flashlight beam 6: wide. The very highest frequencies are like a laser beam, less than an inch wide. The higher the frequency, the narrower the wavelength. Being off-axis just a little bit means that beam of high frequencies can miss our ears and hit the wall behind us. Its the difference between hearing the localization cues in the music or not. We humans are quite good at that. Most mammals have even better hearing than we do.

The human auditory system evolved over millennia for survival, hearing the sound of a twig snapping at 50 yards away and knowing within inches the location of the twig was the difference between life and death. This ability is put to use by audiophiles who value pin-point precision imaging. I think the Uni-Q technology puts the tweeter where it belongs, as a point source, it has a better chance of delivering pin-point imaging than most speakers. The industrial design and craftsmanship are state-of-the-art.

 

[JStewart]

The response chart above says its testing the Meta version.

Yes, it does. Not my point though.

Most of it is like a flashlight beam 6: wide. The very highest frequencies are like a laser beam, less than an inch wide. The higher the frequency, the narrower the wavelength. Being off-axis just a little bit means that beam of high frequencies can miss our ears and hit the wall behind us.

I believe this is conflating wave length and beam width. It’s either that or I am completely missing the point on charts such as this:

I read this as for a good portion of the upper HF range above 10k the frequency is not falling off between +20° and -20°. If correct, the width of the beam is greater than 7 ft at a distance of 10ft. Is this not correct?

 

[ddude003]

@JStewart had said: "I’d like think the correct approach for the average home listener is to first set speakers up on axis and then experiment with toe-in or out to suit. Folks have different tastes and different rooms, not to mention wildly different speakers."

Not to mention ears and brain... Although I may not be the "average home listener", I like my Martin Logan ESLs pointed to cross about 12 inches behind my head (toe-out from on-axis)... And that is with precise room location positioning, time aligning and DRC FIR EQing over a B&H type house curve...

Another thing to consider might be the Fletcher Munson effect and how do you effectively A/B variations of toe...

 

[Bob Rapoport]

Yes, it does. Not my point though.

I believe this is conflating wave length and beam width. It’s either that or I am completely missing the point on charts such as this:
View attachment 85893

I read this as for a good portion of the upper HF range above 10k the frequency is not falling off between +20° and -20°. If correct, the width of the beam is greater than 7 ft at a distance of 10ft. Is this not correct?

You make a good point about starting on-axis. Just to clarify the physics: wavelength is the speed of sound divided by frequency. At 10 kHz it’s about 1.35 inches, at 15 kHz around 0.9 inches, and at 20 kHz only 0.67 inches. That’s why tweeters beam like a laser beam at the top end — the wavelengths are tiny, not several feet wide. It’s exactly this narrowing that makes on-axis setup so important for hearing a precise stereo image. Thanks.

 

[Bob Rapoport]

@JStewart had said: "I’d like think the correct approach for the average home listener is to first set speakers up on axis and then experiment with toe-in or out to suit. Folks have different tastes and different rooms, not to mention wildly different speakers."

Not to mention ears and brain... Although I may not be the "average home listener", I like my Martin Logan ESLs pointed to cross about 12 inches behind my head (toe-out from on-axis)... And that is with precise room location positioning, time aligning and DRC FIR EQing over a B&H type house curve...

Another thing to consider might be the Fletcher Munson effect and how do you effectively A/B variations of toe...

AS you mentioned earlier, taste is a factor that cant be measured, its personal. Electrostats are famously directional and deliver pin-point imaging. They're also dipoles, creating a first early reflection from the wall behind them. 1/2 the output goes backwards. I had Final Electrostats for many years myself and loved that spacious sound field in my home theater system. It was fine for immersive soundtracks although the pin-point imaging was missing.

 

[JStewart]

You make a good point about starting on-axis. Just to clarify the physics: wavelength is the speed of sound divided by frequency. At 10 kHz it’s about 1.35 inches, at 15 kHz around 0.9 inches, and at 20 kHz only 0.67 inches. That’s why tweeters beam like a laser beam at the top end — the wavelengths are tiny, not several feet wide. It’s exactly this narrowing that makes on-axis setup so important for hearing a precise stereo image. Thanks.

Perhaps I’m confused. If the point is that SPL drops for a 15kHz if more than 0.9” off the tweeter axis then I can’t agree. If the meaning is other, then my apologies for being dense.

Also, and I know not part of the original article, my experience is equal time arrival and equal SPL by frequency of the two speakers are more important factors for imaging than being directly on axis.

 

[Bob Rapoport]

Perhaps I’m confused. If the point is that SPL drops for a 15kHz if more than 0.9” off the tweeter axis then I can’t agree. If the meaning is other, then my apologies for being dense.

Also, and I know not part of the original article, my experience is equal time arrival and equal SPL by frequency of the two speakers are more important factors for imaging than being directly on axis.

Thanks for your thoughtful follow-up — no worries at all, this is exactly the kind of discussion that helps everyone sharpen their understanding.


On the physics: the wavelength at 15 kHz is about 0.9 inches (speed of sound ÷ frequency). That means if you move just an inch or so off the tweeter’s central axis, the output in that top octave begins to roll off. Those very short wavelengths are what carry much of the fine spatial detail, so losing even a little of that direct energy does affect localization cues.


And on your second point — you’re right that equal arrival time and balanced SPL are critical for imaging. The key is that being off-axis changes both: one speaker arrives slightly sooner and louder than the other, and the precedence effect (our brain locking on to the first arriving sound) can skew the image. That’s why starting from an on-axis, equidistant sweet spot gives the most accurate reference before making any personal tweaks.

 

[Kal Rubinson]

On the physics: the wavelength at 15 kHz is about 0.9 inches (speed of sound ÷ frequency). That means if you move just an inch or so off the tweeter’s central axis, the output in that top octave begins to roll off.

Again, as someone else pointed out, you are conflating beam width with wavelength.

 

[w_sizer]

Again, as someone else pointed out, you are conflating beam width with wavelength.

Yes, I’m no physicist, but I believe wavelength is just that: length. Beam width is measured in radians or degrees. As I understand it, beam width expands as sound travels through the air. I agree with Bob on one point - this discussion is very educational.

 

[Kal Rubinson]

Beam width is measured in radians or degrees.

Beam width can be measured in radians/degrees with reference axis or as a length/width orthogonal to the axis.

 

[ddude003]

AS you mentioned earlier, taste is a factor that cant be measured, its personal. Electrostats are famously directional and deliver pin-point imaging. They're also dipoles, creating a first early reflection from the wall behind them. 1/2 the output goes backwards. I had Final Electrostats for many years myself and loved that spacious sound field in my home theater system. It was fine for immersive soundtracks although the pin-point imaging was missing.

My ML ESLs are a Curvilinear Line Source Dipole design which has a wider dispersion then flat panel designs like your Final Electrostats... I also have bass and wide band absorbers in the corners behind them minimizing the back side wave bounce off the wall... Both speakers are time alined to that MLP point 12 inches behind my head making each ear on axis to its respective L and R speaker... 15kHz and above is the in the air band area of the frequency spectrum and yes it may contain spacial information given the quality mastering of the content...

Honestly, are you here to teach us something new or are you here to hock your hardware?

 

[Bob Rapoport]

Again, as someone else pointed out, you are conflating beam width with wavelength.

I hear you, and I appreciate the chance to clarify. I’m not actually conflating the two — wavelength and dispersion (or beamwidth) are directly connected. The physics are straightforward: as frequency goes up, the wavelength gets shorter (10 kHz ≈ 1.35″, 15 kHz ≈ 0.9″, 20 kHz ≈ 0.67″). When the wavelength approaches the size of the tweeter diaphragm, the sound naturally becomes more directional — the beam narrows.


So the “beam-like” behavior isn’t a separate phenomenon from wavelength, it’s a direct consequence of it. That’s why listening on-axis is so important: those short treble wavelengths carry the fine localization cues, and sitting off-axis means you’re hearing less of them, which smears the stereo image.

 

[Bob Rapoport]

My ML ESLs are a Curvilinear Line Source Dipole design which has a wider dispersion then flat panel designs like your Final Electrostats... I also have bass and wide band absorbers in the corners behind them minimizing the back side wave bounce off the wall... Both speakers are time alined to that MLP point 12 inches behind my head making each ear on axis to its respective L and R speaker... 15kHz and above is the in the air band area of the frequency spectrum and yes it may contain spacial information given the quality mastering of the content...

Honestly, are you here to teach us something new or are you here to hock your hardware?

My ML ESLs are a Curvilinear Line Source Dipole design which has a wider dispersion then flat panel designs like your Final Electrostats... I also have bass and wide band absorbers in the corners behind them minimizing the back side wave bounce off the wall... Both speakers are time alined to that MLP point 12 inches behind my head making each ear on axis to its respective L and R speaker... 15kHz and above is the in the air band area of the frequency spectrum and yes it may contain spacial information given the quality mastering of the content...

Honestly, are you here to teach us something new or are you here to hock your hardware?

This post was meant to be informative, it does not mention the hardware I manufacture and sell. About 75% of the consumers I talk too are not as well informed as you are and have never heard precision stereo imaging and don't know why proper speaker setup is important.

 

[Kal Rubinson]

I hear you, and I appreciate the chance to clarify. I’m not actually conflating the two — wavelength and dispersion (or beamwidth) are directly connected. The physics are straightforward: as frequency goes up, the wavelength gets shorter (10 kHz ≈ 1.35″, 15 kHz ≈ 0.9″, 20 kHz ≈ 0.67″). When the wavelength approaches the size of the tweeter diaphragm, the sound naturally becomes more directional — the beam narrows.

So, yes, tweeter diameter/size is a factor in dispersion but its dispersion needs to match with that of the associated drivers or timbre and that is influenced by its mounting and immediate environment. We see a number of speakers that remove the tweeter from the main box and place it in its own small housing in order to increase dispersion such that it matches that of the midrange driver. Even more significant is the employment of "acoustic lenses" which, like horns, reduce reflection/interference from the surrounding panel on the direct radiation from the diaphragm to widen dispersion, smooth the blend with the output of the other drivers and reduce distortion.

Well engineered modern speakers are characterized by uniform dispersion over most of the spectrum, especially above the "critical" frequency, and it is wide enough to permit reasonable toe-in, if necessary, to accomodate to room conditions. Here's an example:

 

[Bob Rapoport]

So, yes, tweeter diameter/size is a factor in dispersion but its dispersion needs to match with that of the associated drivers or timbre and that is influenced by its mounting and immediate environment. We see a number of speakers that remove the tweeter from the main box and place it in its own small housing in order to increase dispersion such that it matches that of the midrange driver. Even more significant is the employment of "acoustic lenses" which, like horns, reduce reflection/interference from the surrounding panel on the direct radiation from the diaphragm to widen dispersion, smooth the blend with the output of the other drivers and reduce distortion.

Well engineered modern speakers are characterized by uniform dispersion over most of the spectrum, especially above the "critical" frequency, and it is wide enough to permit reasonable toe-in, if necessary, to accomodate to room conditions. Here's an example:
View attachment 85953

That’s a great example of how designers use waveguides and lenses to manage dispersion and integrate tweeters with midrange drivers — no doubt it improves smoothness and room-friendliness. I’d just add that while uniform dispersion is valuable, especially for tonal balance off-axis, there’s another side of the coin: localization cues.


Our auditory system locks onto the very first arrival of high-frequency content to place sounds in space. If a lot of that treble energy is spread broadly into the room, reflections arrive nearly at the same time as the direct sound. That can smear the image and weaken the precise stereo illusion. Starting on-axis minimizes that, because you’re getting the full, unblurred top-end directly.


So I think we’re both pointing to the same truth: dispersion needs to be well-managed for good blending, but for the sharpest imaging, on-axis listening remains the most reliable reference.
I think some of this comes down to listening goals. In a surround setup, a more diffuse, room-filling presentation is the aim, and the extra channels (and resulting reflections) help create that sense of envelopment.


But in 2-channel stereo, the target is different — it’s all about precision. The magic of a holographic phantom image between the left and right speakers relies on those tiny high-frequency localization cues arriving cleanly. That’s why on-axis listening is so important: it minimizes the smearing effect of reflections and preserves the sharpest possible stereo image.

 

[ddude003]

This post was meant to be informative, it does not mention the hardware I manufacture and sell. About 75% of the consumers I talk too are not as well informed as you are and have never heard precision stereo imaging and don't know why proper speaker setup is important.

So talk about proper speaker placement, physical room correction via traps, absorbers and diffusers, digital room correction... This focus on on-axis positioning is a bit baffling to me... The room is the biggie... Speaker positioning in the room next... Time alignment between speakers and the MLP... Then toe? Seems like you would want a kit that can deliver 20Hz to 20kHz at a minimum...

And yes my focus is on 2-channel stereo... Actually 2.1 to be more precise...

 

[Bob Rapoport]

So talk about proper speaker placement, physical room correction via traps, absorbers and diffusers, digital room correction... This focus on on-axis positioning is a bit baffling to me... The room is the biggie... Speaker positioning in the room next... Time alignment between speakers and the MLP... Then toe? Seems like you would want a kit that can deliver 20Hz to 20kHz at a minimum...

And yes my focus is on 2-channel stereo... Actually 2.1 to be more precise...

Thanks for your thoughtful input — you’re absolutely right that the room, treatments, and digital correction are all major factors in sound quality, and I definitely plan to cover those in a future follow-up.

For this first piece I deliberately kept the focus narrow: on-axis listening. That’s always where I begin, because it sets a reference. My starting point is with a good pair of sealed headphones — that takes the room out of the equation completely and gives me a benchmark for what’s actually on the recording.

From there, I try to match that same clarity and focus with the loudspeakers in the room, beginning in a near-field position (around 4 feet apart and 4 feet back). At that distance the direct sounds hit your ears before the reflections, so you really hear the stereo image lock in. As you move farther away — 8 feet, 10 feet — the reflections start to arrive in greater proportion, and beyond that the room takes over. The image goes from being pin-sharp to more diffuse, with echoes and reverberation smearing those localization cues we talked about earlier.

So yes, the room is a big factor, and I’ll definitely dig into traps, absorbers, diffusion, and digital room correction in another article. But it all starts with getting the on-axis geometry right, because that’s the foundation of everything else.

 

[Bob Rapoport]

When most people say “digital room correction,” using DSP, they’re usually talking about EQ filters that notch down peaks in the frequency response. The challenge is: EQ can’t remove the physics of a standing wave. It can only reduce energy at certain frequencies, and in the process it often adds its own coloration, processing artifacts, or even distortion.

Instead, I rely on an old audiophile solution that's rooted in solid acoustics, adding a second sub.

• Standing waves in the bass region create both peaks (reinforcement) and nulls (cancellations).
• An equalizer can reduce peaks but it cannot fill in the nulls, because a cancellation is literally a cancellation — the energy is gone.
• By adding a second subwoofer, placed strategically, you excite the room differently. The overlapping wave patterns help “fill in” those nulls while smoothing out the peaks, leading to a flatter, more even low-frequency response without the downsides of heavy EQ.
• The result is tighter, more natural bass that integrates seamlessly with the mains, preserving clarity instead of masking it with extra processing.

That’s another strong myth-busting point: sometimes the simpler physical solution (like adding a sub) is more effective — and more faithful — than trying to “fix it in the DSP.”

 

[ddude003]

When most people say “digital room correction,” using DSP, they’re usually talking about EQ filters that notch down peaks in the frequency response. The challenge is: EQ can’t remove the physics of a standing wave. It can only reduce energy at certain frequencies, and in the process it often adds its own coloration, processing artifacts, or even distortion.

You are talking about old skool analogue equalizers and even many digital EQ VST type plugins... Check into what is being called "the state of the art" in digital room correction... Convolution using Finite Impulse Response (FIR) filters...

Here is a sip from the fire hydrant...

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies