LH Labs, a subsidiary of ultra high-end audio company Light Harmonic (maker of the $20,000 Da Vinci DAC), has wafted massive ripples across the personal audio industry over the last couple of years. With their seminal Kickstarter campaign for the pocket-sized DAC/amp the Geek Out, they popularized crowdfunding as a viable option for audiophile products, spurring a bunch of follow-up campaigns in the Geek Pulse and Geek Wave, as well as campaigns by other companies.
As a corollary to their Geek Wave campaign on Indiegogo, LH Labs introduced the Verb, an IEM designed to accompany the Geek Wave and other portable audio players. The Verb was introduced as a $39 perk to the Geek Wave during Cyber Monday late last year, to be sold later for a much higher retail price upwards of $160. As so, the massive “discount” equitably excited backers of the Geek Wave, and many opted to order the Verb on account of it being a part of the LH Labs ecosystem.
The marketing language proved to be particularly attractive as well:
- Metal Alloy – Better resonance control for a natural sound
- Deep Bass Extension – Ported design produces VERY deep bass
- Ergonomic Design – Doesn’t fall out when you’re on the move
- Comfort Fit – Silicone ear tips in three different sizes for a universal fit
- Convenience – Inline microphone for chatting with your peeps
Through the marketing speak, most people assumed the Verb would be an audiophile-oriented earphone, and looked at it as such. There was even talk of a Verb X model, terminated to the TRRS outputs of the balanced Geek Wave models. LH Labs subsequently began a small beta pilot program, and the development of the Verb itself seemed like pretty smooth sailing compared to the setbacks and controversy within the Geek Pulse and Geek Wave campaigns that they helmed. It was a product that was not merely going to be released on-time, but early. Backers were veritably happy to have something delivered early from the crowdfunding world.
When shipments of the Verb began going out to the general populace in March last month, however, the tide began to shift. While there were positive reviews of the product from individuals largely unknown to the veteran enthusiast community for IEMs, quite a few experienced listeners began coming out in droves to complain about poor sound quality in the Verb. Our very own shotgunshane was one of these backers.
He received his LH Labs Verb and immediately remarked (internally to us at CYMBACAVUM) that it “sucks” and was “muddy, bassy, veiled, and with crappy resonances all over”. Nothing prepared us to hear these kinds of remarks. We were genuinely surprised. Those of you who have been readers of CYMBACAVUM know that most of us lean toward a more neutral sound signature, but we can appreciate a high-performance bassy signature when it is presented to us, and shotgunshane assured the rest of us that this was not merely a bass issue. Apparently, on top of sloppy bass, it had no midrange and bad ringing in the treble region.
Meanwhile, more troubling revelations came to light about a possible spurious origin to the Verb. Whilst LH Labs has maintained that the Verb was designed and developed exclusively for their own company, the Xuma PM73, a low-cost, mass-produced IEM sold at New York based Photo/Video superstore B&H for about $25, is a near identical product to the Verb, and had come out years before the Verb was even released.
So was the Verb a true premium IEM whose sound signature one of our writers happened to be allergic to, or was it a throwaway item that happened to have slick marketing around it? We decided to investigate. The IEMs were sent to our new electroacoustic measurement contributor, speakerphone, in Korea, and given a full battery of tests — measurements that are considered internationalized gold standards for earphone design and quality control. For the full details of our measurement methodology, please visit this page.
Editor’s Note: Since the publication of this article a number of individuals have called into question our testing methodology and whether we were biased in the way we tried to “spin” our results. For full disclosure, the earphones were sent from shotgunshane in the US directly to speakerphone South Korea, and the interpretation of the results were made by Mr. T, who has no experience with the Verb. We’re not trying actively to “besmirch the good name” of LH Labs or anything like that. But the Verb has problems. A lot of problems — problems we feel the public needs to be aware of.
…when we got the results back, they were even more horrific than we ever anticipated
Based on SGS’ unflattering characterization of the sound, we suspected these measurements wouldn’t look great, but when we got the results back, they were even more horrific than we ever anticipated:
Let us examine the uncompensated measurements first. Normally, when measurements are made via an international industry-standard IEC60318-4 compliant ear simulator, they don’t look anything like this:
The international IEC standard stipulates normalizing plots relative to 500 Hz at 94 dB, but for practical purposes we’ll stray from that kind of firm interpretation here, or else the results will be even more skewed.
For raw, uncompensated ear simulator measurements, we expect to see a broad peak ideally about 9-12 dB above 1 kHz levels (though lesser is acceptable as well, but even a 3 dB broad peak can be found on the most colored earphones, almost never recessed), centered between 2.5 to 3 kHz, but instead, the Verb are recessed by 6 dB relative to 1 kHz. This means that there’s an entire chunk of midrange that’s basically missing from the music being reproduced. Psychoacoustically, this means the Verb will sound distant warm, hollow, and indistinct.
Then, there’s a broad resonance peak in the lower treble at 6 kHz, where sibilance is accentuated. Relative to the point of ear canal gain (between 600-800 Hz), the bass itself is accentuated around 12 dB. Basically, there’s nothing good to report, other than the fact that the bandwidth is, er, somewhat impressive.
Addendum: Regarding the accuracy (or lack thereof) of artificial ear simulators — IEC-60711 (now IEC-60318-4) couplers are not known to be accurate beyond 8 kHz or so, and clearly introduce an artificial resonance around 12 kHz. However, none of the problem areas highlighted in this article, other than resonance at 9 kHz, are within this fuzzy range. Keep in mind that while the coupler isn’t necessarily “accurate”, it does give people a “ballpark estimate” of general levels of treble.
We’re at a loss for words.
The uncompensated plot is just completely different from anything any of us at CYMBACAVUM have ever seen.
The results are even more glaring when the FR is compensated for diffuse field, GoldenEars’ modified target, or Harman International’s Olive-Welti equalization:
There’s really not much to remark, other than the fact that these are the absolute worst set of frequency response measurements we’ve laid our eyes on. We’ll let your eyes be the judge.
Does it get any better if we vary the insertion depth or block off some vents? Nope.
With varying insertion depth, we see some expected variation, such as the shifting of quarter-wave resonances in the treble downward in frequency with respect to lessened insertion depth. The big issue at 6 kHz would remain at all insertion depths.
Addendum: Once again, these results only affect the regions where the ear simulator generally is not considered accurate (above 8 kHz), so take those results with a grain of salt.
With its one visible air port blocked, the Verb did not seem to change much in frequency response, suggestive that the vent is not there for bass porting but rather relief of over-pressurization to the dynamic driver.
…these are the absolute worst set of frequency response measurements we’ve laid our eyes on.
Frequency response doesn’t tell us the whole story, however. Provided that the transducer is extremely well-damped for resonances (which LH Labs claims in its marketing materials for the Verb), an earphone can still sound somewhat decent, despite being completely off-kilter with respect to FR.
Cumulative Spectral Decay (CSD)
The results for CSD do not acquit the Verb, however. In fact, they add fuel to the fire, as they corroborate shotgunshane‘s “crappy resonances” statement. The Verb rings heavily at its resonance peaks of 6 kHz (where sibilance becomes prominent) and 9 kHz (where harshness can be emphasized). Bass decay and group delay are off the charts bad. We’ve never seen bass that creeps so egregiously deep into the midrange.
Addendum: What’s most important here is not so much the absolute scale of the time course for ringing, but rather the relative time course. When bass decays so broadly, and treble ringing occurs at many times the scale of the non-ringing midrange region, the result, undoubtedly, is poor sound quality. CSD protocols vary from software to software, and from testing rig to testing rig. Results for this test (performed on ARTA), are not directly comparable to results acquired by, for example, InnerFidelity (an excellent website). For reference, we’ve included the specific parameters for testing in ARTA here: FFT Block Shift: 2, Max FFT Block: 60
At this point, readers might be wondering whether or not this Verb unit was a bad unit.
But distortion results aren’t so terrible. Our testing methodology involves testing at 94 dB at 500 Hz, and distortion values are mostly under 0.1% with the exception of the bass:
Since the bass is so grossly boosted in the Verb, the distortion in that region is probably audible. However, what these distortion results reveal best is the fact that our unit of the LH Labs Verb is almost surely in good functional order. When there is a performance problem with an earphone, it’s usually distortion that goes first (lest it just completely breaks and doesn’t produce sound), be it a driver loose in the housings, misaligned driver, etc. These problems would also manifest themselves in the FR as well, but the FR and distortion for both the left and right sides were reasonably well-matched.
Unfortunately, that means the FR and CSD results really are that terrible — we don’t need to re-state what that revelation means. We’d been partially hoping that the distortion had been really bad, so that we could then attribute the poor measurements to the possibility of the drivers in the Verb being broken, but we can’t.
The Verb is just terrible, plain and simple.
Editor’s Note: A number of individuals have called into question the “unrealistically good” performance of the distortion. Keep in mind that these results are reflective of only Total Harmonic Distortion (THD), not THD+N. Indeed, the Verb does perform pretty well with respect to distortion, even when compared directly to other results acquired by the same measurement setup. For example, the distortion of the Etymotic ER4S, acquired by the same setup, is on the whole much “worse” than that of the Verb. However, all of this is a moot point. The point is that no amount of amazingly clean distortion will make up for a frequency response that defies reality. The distortion performance here only highly suggests that the Verb unit we have on our hands is not faulty.
The impedance results are as stated, hovering around 16 Ω.
The LH Labs Verb is unbelievably bad. There really is no way around that statement. If, today, the Verb were marketed as a casual, lifestyle/fashion type of earphone product, there’d be some room for discussion, but even most Beats products these days measure and sound better.
The most serious infraction is that LH Labs marketed this product to audiophiles and gadget enthusiasts — the very market segment that cares deeply about acoustic performance. The electroacoustic measurements we got are just incomprehensibly terrible. There are $15 earphones on the market that perform better.
We’ve got more bad news for you — unless the Verb is completely reworked, it’s not getting any better with the balanced “Verb-X” version, either. Balanced output won’t perform any miracles, so don’t hold your breaths if you’re one of the backers for the balanced version of the Verb.
What makes these revelations most troubling is the fact that LH Labs has long hung its hat on great performance numbers to vindicate its digital and analog design methodologies. Larry Ho, the principal founder of Light Harmonic and LH Labs, is an experienced electrical engineer with admirable audio design chops. At the same time, knowing how to design DACs and amps doesn’t make a company qualified to understand the priorities in the design of headphones and earphones.
The Verb is at best a serious case of negligence on the part of LH Labs, and at worst, outright fraud. They need to do a serious investigation into the sound quality of the Verb. These earphones defy acoustic expectations of anyone remotely interested in quality audio. So even if the Verb as intended was much better than the current product being delivered, the entire project has to be the epitome of the Swiss cheese model — clearly, someone’s head needs to roll over at Light Harmonic for this project.
The Verb is available on Amazon for $89, but we don’t recommend anyone buying it. If you’re even remotely serious about sound quality, do NOT purchase the LH Labs Verb. It is not worth paying $38 for, let alone $89.
Light Harmonic itself was founded with the spirit of Renaissance genius Leonardo da Vinci in mind, but the Verb is undoubtedly a shameful representation of invention.
Editor’s Note: We performed these measurements in hope that the Verb our writer received was actually faulty, so they were done with care and adhered to international standards from the IEC. If LH Labs would like to respond to this article, we welcome company representatives to contact us and present your side of the story. We will then append that response to this article.