Zaylli Lyrö. These on-ear headphones have promising measurements.

[Ventfull]

I have have come across a new kickstarter headphone, which usually I would ignore but the kickstarter is well funded, and they did send review samples to ORATORY, Unheard Labs, who provided freqeuncy response and distortion measurements; along with GREQ, who gave a glowing review on Youtube. These seem very promising to those who dont like using IEMS such as myself. I believe retail is 250 USD while earlybird pricing is 139 USD. The head cups are made of CNC-Machined Aluminium and the headphone is very light at 112 grams. They openly market third party independent measurements such as Oratory on their website page which I found promising. I'll show FR, THD measurements. There is a removable cable. Kickstarter link is here.

The image of the product is later down the post. Oratory confirmed distortion to be below 0.05% at 94db (at 500hz). Link to all of Oratory's measurements of the headphone. Full credit to him of course. Manufacturer claims third-order harmonics below 0.02% across the entire audible range at 90db, and less than 0.5db channel imbalance.

The headphones measure quite close to Harman. The headphone has a rotating knob on the outside of the earcup. With this knob you can adjust the headphone to have less bass than Harman or more bass than Harman. The middle position is closest, leaving enough room to adjust to your own preference. There is also a setting for a Diffuse Mode tuning. Apologies for any formatting issues, Im not used to posting on here. This is not an advertisements, I'm genuinely hopeful of this product especially to those who can't wear IEMs such as myself.

Oratory Measurments​

 

[solderdude]

If only they would have gotten rid of that 3.5kHz peak then it would not have desperately needed EQ.

Oh wait.... there would be an easy fix for that.

 

[Ventfull]

The only objective con with this headphone is that 3.5hz peak. Although it seems easy to EQ. GREQ reviews claimed it wasnt that audible. Maybe beacause its an on-ear headphone, but Im not sure. Maybe because not dependent on seal?

 

[solderdude]

Sai did not measure the peak though. At least not nearly as high.

Another measurebator saw it more like an HD600 with bass (5128)

The peak only seems to be as prominent on Oratory's fixture.
This means the peak is coupling dependent and/or positioning dependent.
It is right in the ear-gain range for on-ears and perfectly illustrates that headphone measurements aren't exact science but indicative at best.
Who's measurements to trust ?

Seems like a promissing headphone but still in kickstarter phase.

 

[Ventfull]

Sai did not measure the peak though. At least not nearly as high.

View attachment 503321

Another measurebator saw it more like an HD600 with bass (5128)
View attachment 503323

The peak only seems to be as prominent on Oratory's fixture.
This means the peak is coupling dependent and/or positioning dependent.
It is right in the ear-gain range for on-ears and perfectly illustrates that headphone measurements aren't exact science but indicative at best.
Who's measurements to trust ?

Seems like a promissing headphone but still in kickstarter phase.

I would assume the 5128 measurments would be more accurate especially in the treble then all the other non 5128 measurments no? It is considered the most accurate rig especially in that region and I believe its the most expensive.

I was surprised to see the kickstarter promotion showed the company had a 5128 measurement rig which costs around USD 50k.

 

[solderdude]

The 'issue' with the 5128 comparison is a bit lame though and suggests there is no problematic peak at 3.5kHz.
The 'trick' that is use is where to put the reference.
In this case the reference is the dotted line.

Notice how they cross at 250Hz which is not particularly standard.Should the traces overlay at 1kHz (which is kind of a standard) then the Lyro line would be 2dB higher and peak about +5dB above the upper tolerance band and about +7dB above the target (dotted line) so while the 5128 plot suggests there is no 'troubling' treble peak there secretly still is.... also on the 5128.

One can also rent a 5128 and maybe the picture is taken at some lab where they tested it and took piuctures.

 

[Somafunk]

Could be an interesting “blind buy” as it’s an acceptable price of £100 for a small portable headphone, it reminds me of a grownup koss porta pro type on ear which as the Koss was my first proper headphone back in 1988 along with my DC2 Walkman leaves me with that irrational need/desire to buy one.

I’ll have a closer look at the “promo” vids and check out the fanboys/girls on headfi but I can imagine one winging my way come April

 

[harvey.zhx]

I have have come across a new kickstarter headphone, which usually I would ignore but the kickstarter is well funded, and they did send review samples to ORATORY, Unheard Labs, who provided freqeuncy response and distortion measurements; along with GREQ, who gave a glowing review on Youtube. These seem very promising to those who dont like using IEMS such as myself. I believe retail is 250 USD while earlybird pricing is 139 USD. The head cups are made of CNC-Machined Aluminium and the headphone is very light at 112 grams. They openly market third party independent measurements such as Oratory on their website page which I found promising. I'll show FR, THD measurements. There is a removable cable. Kickstarter link is here.

The image of the product is later down the post. Oratory confirmed distortion to be below 0.05% at 94db (at 500hz). Link to all of Oratory's measurements of the headphone. Full credit to him of course. Manufacturer claims third-order harmonics below 0.02% across the entire audible range at 90db, and less than 0.5db channel imbalance.

The headphones measure quite close to Harman. The headphone has a rotating knob on the outside of the earcup. With this knob you can adjust the headphone to have less bass than Harman or more bass than Harman. The middle position is closest, leaving enough room to adjust to your own preference. There is also a setting for a Diffuse Mode tuning. Apologies for any formatting issues, Im not used to posting on here. This is not an advertisements, I'm genuinely hopeful of this product especially to those who can't wear IEMs such as myself.

Oratory Measurments​

View attachment 503292View attachment 503293View attachment 503296View attachment 503295View attachment 503297View attachment 503298

I would like to sincerely thank you for posting in support of our project and for providing such detailed and insightful information. Your contribution means a lot to us.


We have already reached out to Amir (Amir's review platform) and will be sending out the units as soon as possible. We are hopeful that Amir will provide us with a thorough and professional evaluation.


Thank you again for your continued support and for being an integral part of this journey.

 

[harvey.zhx]

The 'issue' with the 5128 comparison is a bit lame though and suggests there is no problematic peak at 3.5kHz.
The 'trick' that is use is where to put the reference.
In this case the reference is the dotted line.
View attachment 503369Notice how they cross at 250Hz which is not particularly standard.Should the traces overlay at 1kHz (which is kind of a standard) then the Lyro line would be 2dB higher and peak about +5dB above the upper tolerance band and about +7dB above the target (dotted line) so while the 5128 plot suggests there is no 'troubling' treble peak there secretly still is.... also on the 5128.

One can also rent a 5128 and maybe the picture is taken at some lab where they tested it and took piuctures.

First, thank you for bringing up this question. We are very happy to have the opportunity to explain our tuning method here. We truly value feedback from our users and are excited to share more details about the tuning of Lyro. Here, we will explain the reasoning behind the 3.5 kHz treble boost and clarify why we did not strictly follow the GRAS 711 (including RA0402 and KB5000) Harman target curve for tuning Lyro.

GRAS 711 and Harman Target Curve​

GRAS 711 series ear simulators (including RA0402 and KB5000) are commonly used in headphone measurements and are often paired with the Harman target curve. The Harman target curve is based on years of auditory research and is designed for different types of headphones.

• Over-ear headphones: The Harman curve has a specific target curve for over-ear headphones, based on various reference headphones. This curve takes into account the unique acoustic characteristics of over-ear headphones, including the position of the ear cups relative to the ear.
• In-ear headphones: The Harman target curve also defines a specific target for in-ear headphones, considering the differences in ear canal and ear response.

However, the Harman curve does not define a specific target for on-ear headphones. On-ear headphones (like the Lyro) have a unique wearing style that differs significantly from over-ear and in-ear headphones. Therefore, using the Harman curve as a standard does not fully apply to on-ear headphones.

Zaylli's Acoustic Approach: GRAS 711 Series (RA0402 and KB5000)​

For Lyro's development, we chose to measure using the GRAS 711 series (including RA0402 and KB5000) ear simulators. The GRAS 711 series provides extremely low distortion and high precision, suitable for designing ultra-low distortion headphones like Lyro.

• RA0402 and KB5000: These two ear simulators have similar ear models and are suitable for on-ear headphones. The KB5000 is especially well-suited for on-ear and over-ear headphone development due to its softer ear model, which more closely simulates the actual fit of on-ear headphones. Therefore, we chose this series to ensure accuracy and reproducibility in our measurements.

Because the GRAS 711 series differs from real human ears, we decided to develop a custom tuning curve for Lyro—the "Zaylli Curve." This curve was finalized after multiple tests and feedback from a variety of listening experts, ensuring the best possible listening experience for users. Compared to the two Harman 711 curves (one for in-ear headphones and the other for over-ear headphones), the Zaylli curve shows higher consistency with the sound signature of flat loudspeakers in a laboratory environment. The Zaylli curve also exhibits better linearity in the high frequencies. We found that the Zaylli curve more closely matches the flat loudspeaker response, achieving a more accurate reproduction of natural, balanced sound without excessive boosts or cuts in any frequency range.

3.5 kHz Boost in Lyro Tuning​

During the tuning process, we chose to slightly boost the 3.5 kHz range, a critical frequency for vocal clarity and spatial representation. This adjustment was made to enhance clarity in relatively quiet or slightly noisy environments (such as home or sleep settings). The boost is very subtle, and while most people would not readily notice it, it helps to improve clarity, especially in speech and high-frequency details.


Compared to the Harman target curves for in-ear or over-ear headphones, this boost was intended to improve overall sound quality without introducing harshness or excess. This adjustment took into account the typical usage scenarios of Lyro, where the headphone is used in relatively noisy or very quiet environments. Therefore, we decided to increase some low frequencies and lightly boost the 3 kHz region to achieve an equal loudness balance.

Performance Comparison: Lyro vs HD600 in BK5128 Testing​

In multi-platform validation, we tested Lyro using the B&K 5128 (BK5128) Head and Torso Simulator (HATS), which provides highly accurate high-frequency response testing and is especially suitable for high-end headphones. The results showed that Lyro's performance in the 3 kHz to 10 kHz range almost perfectly matched that of the 2025 version of Sennheiser HD600, verifying Lyro's high-frequency accuracy and excellent sound performance.


This result might be due to BK5128's internal coupling cavity being better suited for different wearing styles compared to GRAS 711, which offers more precise adaptation for on-ear headphones.


Compared to the 2022 version of HD600, Lyro is slightly higher in the 3 kHz range by 2-3 dB, mainly due to the BK5128's harder ear simulator, which often causes slight high-frequency boosts as a result of the simulator's characteristics.

AB Testing Results: Lyro vs HD650 and HD600​

In AB testing with Sennheiser HD650 and HD600, we found that Lyro's high-frequency performance lies between the two, exhibiting excellent linearity and natural sound. This result demonstrates that Lyro's tuning is highly linear, especially in the 3 kHz to 10 kHz range. Through these tests, we confirmed that Lyro's response in this range approaches the ideal standard for flat sound reproduction, showing a more balanced and natural high-frequency response.


It is not a headphone with excessively boosted bass and treble. So, if you are judging it as a V-shaped headphone simply based on the curve showing boosted bass and treble, we want to clarify: under the Zaylli curve, it is a headphone with a more passionate and warm sound, while in the Diffuse-Field mode, its performance is fast and accurate, without perceiving the enhanced bass and treble (this is important as we are concerned that the curve might mislead potential buyers).

Future Development​

We are continuing to develop softer ear simulators to further improve the accuracy of tuning on on-ear headphones, making future headphone development even closer to real human ear response. This will help us achieve more precise tuning in future products, enhancing performance and user experience.

 

[Jim Taylor]

@harvey.zhx

@amirm has rules regarding posting in the forum as a business. Please contact him.

 

[Somafunk]

@harvey.zhx

Welcome to asr Harvey, good to see your presence on the forum and the willingness to expose yourself to amir’s testing regime

A quick question regarding the tuning dial, is the design fully implemented?, i ask due to my medical condition of secondary progressive multiple sclerosis which leaves my fingers numb - this makes it rather awkward to manipulate small items/dials etc. Is there an opportunity to make the dial slightly larger or have a raised section on it to provide a firm grip.

I’ve backed them so looking forward to trying my pair out in may

 

[solderdude]

Future Development​

We are continuing to develop softer ear simulators to further improve the accuracy of tuning on on-ear headphones, making future headphone development even closer to real human ear response. This will help us achieve more precise tuning in future products, enhancing performance and user experience.

Yep, I had problems with measuring on-ears in the 3-4kHz area that correlated with how I heard them too and had to make an 'adapter' for on-ears.
That adapter reacted differently from a fake pinna.

One other issue with on-ears is that the shape of pinnae differs substantially resulting in seal issues affecting the lows as well as positioning issues with relatively large driver membranes close to the ears.

So indeed it is possible the 3.5kHz peak on measurement fixtures is a measurement artifact.
However, when looking at Konstantin's measurements of (the few) on-ear headphones they did not peak at 3.5kHz on his fixture.

I have noticed that some drivers happen to peak around that (troublesome) frequency band for headphones and it is known that there is a substantial difference in ear-gain in that part of the audible range. This means that some people may be bothered more or less than others. Also that peak is rather narrow which may also help 'disguise it' and is around 2 ERB wide so noticeable but in music may not be too detrimental to the sound.

'Tuning' that frequency requires the usage of acoustic resistors around the driver (which is too large in this case) or requires tinkering with the membrane.

 

[harvey.zhx]

Yep, I had problems with measuring on-ears in the 3-4kHz area that correlated with how I heard them too and had to make an 'adapter' for on-ears.
That adapter reacted differently from a fake pinna.

One other issue with on-ears is that the shape of pinnae differs substantially resulting in seal issues affecting the lows as well as positioning issues with relatively large driver membranes close to the ears.

So indeed it is possible the 3.5kHz peak on measurement fixtures is a measurement artifact.
However, when looking at Konstantin's measurements of (the few) on-ear headphones they did not peak at 3.5kHz on his fixture.

I have noticed that some drivers happen to peak around that (troublesome) frequency band for headphones and it is known that there is a substantial difference in ear-gain in that part of the audible range. This means that some people may be bothered more or less than others. Also that peak is rather narrow which may also help 'disguise it' and is around 2 ERB wide so noticeable but in music may not be too detrimental to the sound.

'Tuning' that frequency requires the usage of acoustic resistors around the driver (which is too large in this case) or requires tinkering with the membrane.

The “adapter / on-ear specific interface” approach you mentioned sounds like a very workable solution to me — and it matches my own experience that measurement fixtures can be extremely useful, but still have real limitations once you get into on-ears.
Speaking personally (and not as a formal company statement), I think the biggest trap is assuming a fixed “3–4 kHz” story applies equally to everyone. Real ears vary a lot: in real-ear measurements, the peak resonance frequency has been reported to vary roughly from about 2.1 kHz to 4.8 kHz across subjects, and more broadly the whole 2–6 kHz region is quite sensitive to ear-canal geometry and coupling. So a narrow “3.5 kHz peak” seen on a fixture can easily shift in effective frequency and audibility from person to person — and on-ears add even more variability via seal and driver positioning.

Reliability and intersubject variability of the real ear unaided response - PubMed

Intratester test-retest reliability of the real ear unaided response (REUR) was determined on 49 ears using the Frye 6500 real ear analyzer. Results revealed mean differences of less than 1 dB for repeat measurements at seven test frequencies between 250 and 4000 Hz. The average peak resonant...

pubmed.ncbi.nlm.nih.gov

 

[harvey.zhx]

@harvey.zhx

Welcome to asr Harvey, good to see your presence on the forum and the willingness to expose yourself to amir’s testing regime

A quick question regarding the tuning dial, is the design fully implemented?, i ask due to my medical condition of secondary progressive multiple sclerosis which leaves my fingers numb - this makes it rather awkward to manipulate small items/dials etc. Is there an opportunity to make the dial slightly larger or have a raised section on it to provide a firm grip.

I’ve backed them so looking forward to trying my pair out in may

thanks a lot for the support — and for sharing your situation.


Regarding the tuning dial: yes, the design is fully implemented. I designed it with a number of fairly deep grooves around the edge, so it can be rotated by “hooking” it with a fingernail. It might not be the prettiest solution (haha), but it should be more practical in real use, especially when fine finger control is limited.


The dial itself isn’t small, but in general (for anyone) I don’t recommend adjusting it while wearing the headphone, since you won’t be able to clearly see the scale. The way it’s shown in the video is mainly for demonstration. In practice, most users will set it once to their preferred position and rarely touch it again.


I personally do adjust it occasionally when switching between indoor and outdoor use. Once you receive your pair in April, feel free to let me know how it works for you.

 

[berjine]

Looks awesome, can't wait to recieve mine !
I'm very curious about the gaming cable, it could be a banger gaming headphone

And I like the on-ear form factor as a glasses wearer. I have a modular headphone (AIAIAI TMA-2) and I prefer it with on-ear ear pads rather than over-ear. Now obviously this doesn't have the low-latency wireless feature but it's so lightweight it must be a delight to wear !

 

[solderdude]

So a narrow “3.5 kHz peak” seen on a fixture can easily shift in effective frequency and audibility from person to person — and on-ears add even more variability via seal and driver positioning.

A question that is still not answered is why the 3.5kHz peak is seen in various fixtures even when having different pinnae and ear canals.
Given the fact that other on-ear headphones measured on different fixtures do not exhibit a peak at that frequency band and they occur at the same frequency.
What is the rationale for that observation ?

I like the idea behind the headphone and the low weight.

 

[harvey.zhx]

A question that is still not answered is why the 3.5kHz peak is seen in various fixtures even when having different pinnae and ear canals.
Given the fact that other on-ear headphones measured on different fixtures do not exhibit a peak at that frequency band and they occur at the same frequency.
What is the rationale for that observation ?

I like the idea behind the headphone and the low weight.

On why a ~3.5 kHz peak can still be seen across different fixtures: in my view, on‑ears are largely measuring an acoustic coupling system (a relatively large driver placed very close to the ear + a small front cavity + the ear‑entrance/pinna boundary). When that geometry interacts with a hard‑coupler style artificial ear, energy can easily “pile up” around 3–4 kHz (especially ~3.5 kHz). Even with different pinnae/canals, the frequency can stay similar while the level changes with pinna compliance, seal, positioning, and front‑cavity volume. Although BK5128 has a coupling volume in the same general order as 711, its overall canal/cavity structure is more complex, so you can often see this ~3.5 kHz buildup reduced compared to a more traditional 711‑style coupler—yet it remains an approximation, since real ears also involve skin/soft tissue and underlying cartilage/bone mechanics. As I mentioned earlier, we also intentionally added a small amount of energy around 3.5 kHz via a small rear‑coupling feature; combined with fixture coupling, that can make the measured peak look higher, but it still remains below a full diffuse‑field target. Subjectively, the goal is to preserve clarity and information in portable use: light ambient noise (especially in the 2–4 kHz region) tends to mask timbre and intelligibility, so vocals stay intelligible; in quiet listening, multi‑listener feedback aligns more with “slightly more forward vocals and better separation,” rather than obvious sharpness or a classic V‑shaped bass/treble emphasis.

 

[harvey.zhx]

Looks awesome, can't wait to recieve mine !
I'm very curious about the gaming cable, it could be a banger gaming headphone

And I like the on-ear form factor as a glasses wearer. I have a modular headphone (AIAIAI TMA-2) and I prefer it with on-ear ear pads rather than over-ear. Now obviously this doesn't have the low-latency wireless feature but it's so lightweight it must be a delight to wear !

Thank you so much! I’m a gamer too, haha.
If you’re also into wireless features, what we’re working on next might bring a small surprise in a few months (you should start seeing some progress by then).
That’s also why Lyro was designed to be so modular from the beginning.

 

[Somafunk]

Thank you so much! I’m a gamer too, haha.
If you’re also into wireless features, what we’re working on next might bring a small surprise in a few months (you should start seeing some progress by then).
That’s also why Lyro was designed to be so modular from the beginning.

Cool, so it’s going to be a Bluetooth headband with battery that plugs into the drive units?, sounds ideal.

 

[solderdude]

Subjectively, the goal is to preserve clarity and information in portable use: light ambient noise (especially in the 2–4 kHz region) tends to mask timbre and intelligibility, so vocals stay intelligible; in quiet listening, multi‑listener feedback aligns more with “slightly more forward vocals and better separation,” rather than obvious sharpness or a classic V‑shaped bass/treble emphasis.

Agreed there... for lower listening levels a bit more 'warmth' and 'presence' (is what I call the 2-6kHz region) does not heard.

I prefer a little bit more 'presence' myself (as if listening to near-field monitors) but don't like it 'peaking' there and prefer a 'wider' and more gentle 'bump' if it has to be there.
Especially when listening at studio ref levels (80-85dB average so to speak) in that case a light depression there (lightly reduced 'brightness') is more desirable at those levels.
It also turns out that most people seem to prefer some mild recession in that area rather than peaking as it is known that the test fixtures 'over-report' that area compared to human perception and peaks are more detrimental to SQ than mild recessions (dips). BBC dip effect in speakers is an example.

Are there time domain plots available (like CSD) ? and is that peak visible as ringing ?
Some ringing in that part of the frequency range is not problematic (as the ear 'rings' there too and the brain ignores it) but should not be too long.

Anyways... I like the modular design, simple acoustic tuning for the bass, low weight, easy (consumer) serviceable at least when parts are available for > 10 years.
The acoustic bass tuning is often only seen in closed headphones and can actually work there to compensate for seal issues which is a major issue in closed headphones.

I'd like to see you succeed in your efforts.

A low weight also could mean low clamping force which would be good news for people wearing glasses.

Another desirable, but not easy to achieve, issue is cable microphony. Is there attention for this issue ?
Would be solved wireless (in an active headband).
Nothing as annoying (to me) when a cable is microphonic.

Another (overlooked and often not even recognized) issue is the cable being 4 wire all the way up to the TRS jack. This is not feasible with mic/remote anyway.
How is the cable that comes with the Lyro ? 3-wire from split down or 4-wire all the way to the plug ?

Another question .... how is the impedance curve? Are there plots available for that ?