Heir Audio 4.Ai - Appreciation/Impression/Review Thread - Page 92

There's no such thing as burn-in in BAs and the effect will have to be massive in that specific region, that's the psychological effect you guys are experiencing. Heck to me it sounded better with time as well, but subjective listening is hardly concrete to evaluate such effects.

Edited by Inks - 11/6/12 at 12:39pm

 

So would you say you're more concerned with the data and graphs than you are with what the heir 4ai's sound like?  Will you be selling yours?
 

We have different hearing response and we all have a different idea of what sound should be... Ink, you don't like them, cool... I love it 2x more than others... and others who also have a lot of experience and skills do so too...punkt!

My point of view comparing to W4.

Best treble sparkle
Best medium and timbre
Best bass even if 4 ai doesn't have the quantity of W4 bass rumble
Best imaging, deep, high...coherence
More air and details
More dynamic
Best instrument separation

Is there a dip somewhere or a crossover issue? Frankly, I keep on saying that I don't care...

I love them and know why...

 

 "First the Magnus cable is brought about to bring about the benifits of higher conductivity......."

 

 

 Ahhhhh, so you were there when I designed the Magnus 1?  You saw on the white board the #1 priority of the Magnus 1 cable was to decrease conductivity, and we failed horribly....

 

         

 Inks, really.... do you think that the higher the silver content the lower the conductivity is somehow "lost upon me?"  

 

 

         Sure, I could of decreased the impedance even more, but with a trade off that I wasn't willing to sacrifice....    Most, if not all high content silver cables have the flexability of a coat hanger. I didn't want the Magnus 1 to be a multi-purpose product that would allow one to "Jimmy" a car lock when in a pinch or that could be used to clean a fish tank after attaching a sponge to it, or better yet, used to clean a clogged sink.

 

Truth be known, the #1 priority of the Magnus 1 was to keep the "spirit" of the standard cable, but make it better. The essense of a tinsel cable is its flexability, and its durability.http://en.wikipedia.org/wiki/Tinsel_wire

 

I wanted to maintain that spirit but compound the attributes of the standard cable by:

 

1) adding kevlar rather than Nylon increasing the durability

2) giving it the ability to be converted to a balanced cable in under 20 minutes

3) quick repairs if required

4) and for an added pinch of quality, a higher silver content

 

I believe I reached my objective for the Magnus 1

 

The Magnus 1 is what "I wanted to design" and it may not be what you want (or would design) but that does not make it "Snake Oil," and frankly I am surprised one would use such slanderous terms about Heir Audio.

 

 

"Enough said...." 

 

Actually, I'm still waiting for an apology

Snap.

 

Honestly I've been of the school of thought that buying new IEM cables is for aesthetic and comfort purposes.

 

Also that multi-use "Jimmy" cable bit had me splitting my sides.

 

 

 

 

the irony....it hurts

now repeat after me:

 

"I shall not fall victim to placebo effects"

I shall not fall victim to expensive cables and burn-in.

 

 

 

 

Aaaaaaaand goodbye non-existent reputation.

The tone of some of the criticism of the 4ai is quite bitter and sounds like somebody with a chip on their shoulder.
 

Well I'm glad we can agree on something. As for me,

 

I shall fall victim to the preferences of my ears.

 

The criticism I feel is warranted by the user, as well has his objective data (multiple sources) that verify what he hears as well.  It is written in a "stick-it-to-the-man" sort of fashion, which is the reason many are being offended though.  Could it be a little over stated?  Sure, but I still feel that both sides are overstating things regardless...  I'm talking about the dip in the 4k region BTW, I will not comment on the cable. 

Edited by tinyman392 - 11/6/12 at 4:59pm

There's nothing wrong with criticism - if there's a dip that shouldn't be there, then by all means shine a light on it. I'm glad to hear something about this, but there's no need to besmirch Heir in the process.

I'd still like to hear what Heir has to say about the 4k dip or cross over error. For an IEM of this price what is $450? plus dinner for 4 in shipping one would assume they're purchasing absolute perfection. I don't know if it's something that's simply excusable, and members are turning up saying they heard something wrong, maybe they didn't want to believe it and kept quite all this time. I was never really interested in 4Ai for the price but to find there's a problem rights it off for me completely.

There's a lot of 3Ai turning up on out FS boards to recently I've noticed.

 

Attacking the cable is something I don't agree with, so I'm with you on that. 

So, because the topic is big and convoluted and takes a huge amount of time to cover, I'm just going to address it once:

 

Let's start off with the big issue. Does the 4.Ai/4.A have a notch cut at ~3.7 kHz? Yes. It will be seen in measurements conducted both with Zwislocki-style and 2cc couplers. A secondary issue is bass boost. Is the 4.A/Ai boosted in the bass region? Yes, according to measurements conducted by both Zwislocki-style and 2cc couplers. A tertiary issue is high frequency extension. According to the measurements conducted in the blog post in question, the 4.Ai lacks high frequency presence in the highest registers. I have no other evidence that corroborates this finding.

 

Let's then go to whether these findings are relevant to a person's listening experience. Is the notch audible? To my ears, yes. It is audibly recessed in the 3-4 kHz region, if I listen to tone sweeps and pure tones. However, tone sweeps and pure tones don't really constitute the definition of music for most people. Is this region recessed when I play actual music? Er... I really can't tell you, because I don't know. Subjectively, the only thing I can say when I play music is that vocals can come off a bit flat, and that does correlate with the the notch cut at ~3.7 kHz, as some vocal harmonics reside in that range. I've mentioned that I've tried EQing up the region, and a 3-5 dB bump does seem to help give better definition to vocals, correcting its dimensions, especially with regard to height.

 

I didn't plan on releasing my own graphs publically, because they were an intellectual exercise for my own learning purposes (I had to go to great lengths and call in personal favors in order to obtain these graphs and thus it was a one-time thing), but here they are:

 

 

I fear these graphs may perpetuate further confusion amongst those that do not have ample experience with reading quantitative measurements, but to those with a good idea, they will show that the data both supports and questions the results of the measurements in question. The axes are not of the same scale, so people will have to readjust, but while there is that specific ~3.7 kHz notch dip, it is not as severe by 2-3 dB, even when readjusted for IEC-60711 (or IEC-60318) or even ANSI 3.25 specifications. Keep in mind that this (raw) set of measurements was performed with a 2cc coupler, with clay wrapped around my own (that I bought with my own money) 4.A. I do not make any claims as to the absolute accuracy of these graphs, and I've released this information just to allow the public to see more information in light of all this discussion.

 

---

 

Perhaps it'd be prudent to explain a tiny bit about how measurements are conducted for canal measurements. Putting down the words "Zwislocki" and "coupler" does no one any favors.

 

For measurement, and standardization of measurements, it'd be prudent to have an IEC-60711 (or IEC-60318) compliant coupler, as it is designed as an occluded ear simulator. The basic design was laid out by Dr. Jozef J. Zwislocki way back in the 60s (IIRC, don't quote me on that) as a cavity with two different dual Helmholtz resonance chambers (four in total, surrounding a central chamber) designed to simulate the natural impedance of the ear canal. The design has been revised in later years to yield greater accuracy and/or more simplicity, but the basic design has remained the same.

 

In hearing aid design and QC, many people use the 2cc coupler, which is a simple cavity of two cubic centimeters large. Because it doesn't simulate the acoustic properties of the ear canal, it is mostly used only for performance comparison between drivers and designs, as well as QC.

 

Any coupler, whether it be the 2cc or the IEC711-compliant coupler, when compared to the ear drum response measured with microphones, has a real ear to coupler difference (RECD), the difference being that the RECD of the 2cc coupler is fairly significant (due to the fact that it responds nothing like a real ear) and that the RECD of an IEC711 coupler is much smaller. There's also a correlation between the response of 2cc and IEC711 couplers that is measurable, given that varying acoustic diaphragm impedance in earphones is negligible.

 

Here is a diagram illustrating the differences between a 2cc coupler and Zwislocki coupler:

 

 

Couplers, as can be inferred, merely simulate the response of the human ear canal. They are by no means perfect, and are often unreliable and with irrepeatable results above 8 kHz.

 

The acoustic environmental standard that Choi adheres to when measuring occluded ears is diffuse-field equalization. There are other equalization standards, including free-field and independent of direction (championed by HEAD Acoustics, and employed by Tyll Hertsens). A very short and simplified explanation of diffuse-field equalization versus free-field is found here. Both free-field and independent of direction are claimed to be flawed by Choi. Diffuse-field equalization is also employed by Etymotic Research in its design of the ER4B.

 

---

Back to the issue at hand.

 

Since the 4.A/Ai has a notch at ~3.7 kHz, that must mean it's unlistenable right? Well, no. There are plenty of owners with the 4.Ai or the 4.A that enjoy the sound with nary a bit of complaint. Is it that we're all too used to the sound? Perhaps, perhaps not. Was there a bit of irrational exuberance about the performance of the 4.Ai? Perhaps. Group psychology is often highly influential on Head-Fi. Does that mean that everyone who got a 4.Ai because of this basically bought a lemon. Definitely not.

 

Let's talk about human physiology and neuroscience, a topic of personal expertise. Measurements may tell us the behavior of sound from the source down to the typanic membrane, but there's a whole lot of stuff in between the ears that's equally as important. The human auditory perceptive system is highly adaptable. Even with noise-induced hearing loss (NIHL), we can often function without signficant deficits. Discounting conductive hearing loss at any path between the ear canal and the basilar membrane, we know that cochlear hair cells have overlapping frequencies of redundancy, beginning with the 12,000 or so outer hair cells and the 3500 or so inner hair cells. In an incredible show of neuroplasticity, auditory ganglia can also be 'recalibrated' to sensitize to other frequencies. There's also short-term sensitization of sound due to processing from within the auditory cortices.

 

Right now, you might be thinking, "So, the defense is that I should keep listening to my flawed 4.Ai because my brain can adjust? What a poor argument!" No, but it helps explain why our hearing system is much more sensitive to peaks than valleys. I will freely admit I have a tiny bit of NIHL (at the common 4 kHz area) in my left ear, whereas my right ear is fine. The notch cut at ~3.7 kHz on the 4.A/Ai affects my left ear more than it does my right ear, but the difference is not as severe as one may think. That's one reason why "brain burn-in" affects listening experiences so wildly, but also why people shouldn't just think measurements are the gold standard for how an earphone sounds. A headphone's sound has as much to do with a person's mood and physiological state as it has to do with frequency/impulse response, resonance dampening, etc.

 

Hence, there's a bit of a chicken-or-egg thing going on with measurements, listening impressions, and their assessments. Do we like an earphone more/less while everyone's hyping it, before everyone's hyping it, before it gets measured "amazingly", after it gets measured "amazingly", before it gets measured "horribly", or after it measures "horribly"? Let's review what Inks wrote about the 4.Ai initially, when he first received the earphones:

 

 

So, the only thing that addressed the ~3.7 kHz dip was "midhighs a bit too conservative"; certainly not "severe dip" or "ridiculous dip" that he now claims. Certainly, however, the words 'severe', 'ridiculous' and 'horrible' serve to deliver greater shock and awe than 'conservative'.

 

Anyways, after those two posts, nothing else, until the graphs came out. So, did the his hearing back up the measurements, or did the measurements inform him what he was hearing? I highly doubt Inks would've flat out said that he heard a significant dip in the upper mids if he hadn't seen the graphs, and he didn't, unless he made some private grumbles that none of us were aware of.

 

The real point is, though, is that whether it's from listener impressions, or measurements, your frame of thought can be significantly biased, depending on the information presented to you. Objective information easily transforms into subjective judgment.

 

---

 

I'm now going to take some time to address IEM design and how the notch isn't as horrendous an error as it might seem.

 

I beta-test for an unnamed CIEM company (please don't ask me which company, as I'm not allowed to disclose that fact; but it's not Heir, though, obviously). To improve my knowledge base, I often ask the electroacoustic engineer that I provide feedback to about the difficulties in tuning balanced armatures.

 

Why does a notch even develop? I've mentioned this fact briefly a few times before; the TWFK is a joining of two BA designs, the FK and the WBFK. The WBFK is designed to have the best high-end bandwidth of any Knowles BA driver; as such, it does not have much presence below 4 kHz. When combined with the FK, the net effect is this:

 

 

 

See the notch between 3 and 4 kHz? Well, the notch gets bigger once we start fiddling with capacitors:

 

With a 1.5 uF capacitor attached to the TWFK, the response drops some 10 dB between 2.5 kHz and 3.7 kHz. Because of its popularity in IEM designs, Knowles has made many variants of the TWFK driver, mostly with the goal of bridging the gap between the FK and WBFK, to limited results. It has definitely gotten better, but it is by no means perfect.

 

Case in point: let's look at an actual earphone, the (very expensive) AKG K3003 (Rin's own data is here), which also uses a TWFK for its mids and highs. Here's SONOVE's data:

 

 

Why am I not surprised that there's a dip is at 3.7 khz? Granted, AKG uses an advanced (and much more expensive) version of the TWFK, rather than the generic version, so the gap is smaller.

 

Let's look at another example, an earphone very similar (and very well-received) to the 4.Ai --- the Westone 4. Unfortunately, there are no raw graphs, only compensated graphs from GoldenEars. If we look at the compensated graphs in the light of their specific small room x-curve equalization (-3 dB per octave above 2 kHz, as well as bass boost, but the bass isn't the issue at question here) that they apply on top of diffuse-field equalization, we see that the Westone 4 is also "severely" recessed in the mid-highs. But many don't hear it that way, including me. While this doesn't excuse the 4.A/Ai from the notch dip at ~3.7 kHz, we can start to understand the limitations of design and implementation in multi-BA earphones.

 

As I mentioned very early on in some fairly old posts, it is quite remarkable how nice the 4.Ai sounds, considering that it uses completely generic (read: non-proprietary) components and simple resistors/capacitors. I'm sure Heir could opt to order proprietary drivers and laser-trim its resistors and capacitors to perfect crossover alignment, but then the 4.Ai sure as hell wouldn't be available at $399.

 

Now here's a little bit of an editorial and a challenge:

 

Rin has been measuring (almost exclusively) IEMs for many months. He has a sizable amount of knowledge in electroacoustics. So, it wouldn't at all be a stretch of the imagination to let us know about why exactly that the design of the 4.Ai is a "crossover network error"? The balanced armature units employed in the 4.A/Ai are the TWFK-30017 and DTEC-30008, both designed and manufactured by Knowles Electronics. Being the generic versions, these specific driver models are readily available for purchase from Mouser or Digikey. Knowles posts full specifications of these two drivers on their company website. The acoustic dampers utilized are the white (680 ohm) for the TWFK and the red (2200 ohm) for the DTEC. I'd love to see what Choi would do to 'correct the error'. Wouldn't that be constructive feedback, instead of merely stating that there is an error and saying how absolutely horrendous it is?

 

The fact of the matter is, CIEM design is a complicated process that requires a lot of trial-and-error, and most of all, a lot of compromise, due to the fact that these drivers aren't really designed for audio in the first place, and their highly variant electrical behavior.

 

Can the 4.A/Ai improve in the future? Absolutely. I highly encourage the Wizard to come up with a follow-up product that will address the design shortcomings of the 4.A/Ai.

 

Is it the "awful" earphone that Inks would like for you to believe it is? Absolutely not.

---

 

Inks and I have had our differences in the past. While he's not incorrect about certain things here and I've included much information that backs up some of his/Rin's claims, I absolutely disagree with the manner in which he characterizes things and his naturally irreverent nature. He often attempts to excuse himself by saying he's merely 'busy' and likes to be direct, but he often spends more time dogmatically defending his sensationalistic statements that could easily have been much more diplomatic; there are better ways to inform the public of his crusade that are unfortunately lost on him.