I thought I would share the journey of taking my Abyss AB1266 Φ TC headphones to a new level of audio ecstasy (XTC).
It started after reading the popular article, “Taking the RAAL-requisite SR1a Headphones To Another Level'' here on Audiophile Style. After more digging, I also found and spoke with a few friends on Head-Fi that were very impressed with the superior sound produced by the SR1a filters mentioned in the Audiophile Style review. Needless to say, I was very intrigued and anxious to try the Raal SR1a convolution filters. So I reached out to Mitch Barnett of Accurate Sound to purchase his product. @mitchco responded with the purchase details very quickly.
But after some thought… As good as the Raal SR1a headphones are… and they are incredible. Like many, I find the SR1a to be more of a genre-specific headphone. I use them mainly for classical, instrumental, and soloist music. This is mainly due to the fact that they lack the visceral bass slam that many other genres require for musical reproduction and enjoyment. This is where the Abyss AB1266 Φ TC. (fun to say) takes over. The TCs are considered by many to be the unequivocal “King of Bass” when it comes to headphones. But they also offer an unbelievably detailed, full-range response that makes them my daily driver and all-around favorite headphone.
So I asked Mitch if he would be interested in using my 1266 TC headphones as a test model?
As you may know… Mitch is located in Vancouver, Canada, but my headphones reside in Pennsylvania, USA. Mitch was kind enough to inform me that “customs/insurance is a PITA, but can be done…” Fearful but undeterred, Mitch and I came to an agreement that if he promised to return my $6000 headphones as he received them... I'd be willing to ship them to him for testing. So I boxed them up, said a little prayer, and shipped them off to the other side of North America.
After a week of anxious waiting, UPS delivered my precious 1266 TCs to Mitch safely and undamaged.
Mitch spent the next week with my headphones performing his magical expertise. During that time, Mitch checked in with me daily to ask about my specific requirements, as well as fit, pad position, and to give me updates on his progress. Honestly, I think Mitch went above and beyond to make me feel comfortable with the entire process. I can’t put into words what a sense of reassurance this was to me. Before my initial email, I had never spoken to Mitch. So sending my very expensive headphones to a complete stranger, 2800 miles away was a little unsettling. Thanks again Mitch for treating my headphones as if they were your own.
It would be best if I let Mitch explain the details and process of measuring the 1266 TCs… as well as his creation of the convolution filters.
Mitch’s Measurements, Analysis, FIR filter design and verification:
As mentioned in, “Taking the RAAL-requisite SR1a Headphones To Another Level,” measuring headphones is hard and even harder to eq properly.
Following on with additional information. Regardless of measurement technique and apparatus used, headphone measurements above 3 to 4 kHz don’t reflect the actual frequency response of the headphone. There are several reasons for this, here are a couple of the main ones:
Our individual HRTF’s are unique. A great presentation is David Griesinger’s “Laboratory reproduction of binaural concert hall measurements through individual headphone equalization at the eardrum” (333 MB download). What is cool is that the slide presentation contains audio recordings of ten listeners perceived timbre at the ear canal and how each one is different. If you give it a listen, it will demonstrate how dramatically our individual HRTF’s can vary from each other. It is a real ear opener.
Another reason has to do with wavelengths and reflections. A 4 kHz wavelength is 3.4 inches and becomes smaller as frequency increases, to 0.68 inches at 20 kHz.
As shown in the photos, the size of the headphone driver, surface area/material, and ear pad present opportunities for high frequency reflections, which yield narrow band peaks and cancellations. This is an issue with most over the ear headphones, not just the TC’s.
Trying to eq narrow band peaks or dips above 3 to 4 kHz is an exercise in futility and will never sound right. I try to avoid any narrow band eq above 3 to 4 kHz as the measurement is no longer representative of the actual headphone response. Depending on how much high frequency energy is coming to ones ears, or prefers, may require an eq shelf or overall tilt of the frequency response above 3 to 4 kHz up or down a few dB. Basically one has to interpolate the high frequency response by following the overall spectral envelope and ignore the sharp peaks and dips.
These two points sets us up nicely to help interpret the measurements.
I took 30 stereo measurements of the 1266 TCs using the following equipment. For each measurement, I adjusted the headphone for most comfortable fit on my head as per manufacturer’s instructions and John’s tips. I also measured with the ear pads at their default 1pm/11am positions, 2pm/10am and finally at 3pm/9am positions. All pad positions measured virtually the same from a frequency response perspective. The different pad positions offer a different soundstage presentation, with virtually identical frequency response.
Remarkably identical, given I was adjusting the headphone for each measurement and using different pad positions. While I wear glasses, the low frequency seal (or unseal) is still quite consistent below 30 Hz. The response is consistent to up to 4 kHz before we start seeing the different reflection patterns as I move the headset around on my head.
I can average the result or hand pick a pair of measurements that best represent the headphones frequency response:
This is a full resolution measurement with no frequency “smoothing” applied. We do not hear those narrow dips as they are too narrow of a bandwidth for our ears/brain discriminate.
With respect to the Harman headphone target response, I have read the research with my AES membership on the subject. I have also written about it in a peer reviewed article by Sean Olive. For me, there is too much low bass and upper midrange energy, the latter being right in our ears most sensitive hearing range. This type of eq, whether baked into the headphone or one of the many PEQ’s available, does not sound flat or neutral to these ears. It sounds like loudness compensation to my ears, but applied at the wrong sound pressure level. Flat sounds better and well, neutral.
Digital and electronic gear are designed to have a flat frequency response. Our loudspeakers are designed to measure flat in an anechoic chamber. Speakers still measure flat in-room response, albeit a tilted downward response due the varying directivity of monopole loudspeakers. Bass being omnidirectional and speaker dispersion becoming more directional as frequency goes up. For headphones, there are complicating factors like individual HRTF’s and reflections in the ear cup. However, I don’t see why headphones should not be equalized to a flat response as well, sans individual HRTF and internal surface reflections.
When I listen to headphones that are eq’d flat, the tonal response sounds even, where all frequencies are at the same level. No one frequency or range sticks out over the other. To me, this a “neutral” frequency response. The idea is not to change the tonal response (i.e. audio signal) coming into the headphone. Nor should the stereo image or depth of field be changed in anyway. This means that both channels frequency and timing response should be as close to identical as possible. Accurate, neutral sound with no frequency response or time domain distortions.
I utilize a loudness control, both for headphone listening and loudspeakers. When I am not listening at critical listing or reference level, the loudness control kicks in so at low listening levels, the music still sounds full. JRiver has a good implementation of a loudness control.
Let’s start by applying a frequency dependent window (FDW) to the headphone measurement to better represent what our ears hear. For now, let’s forget beyond 3 kHz due to HRTF/internal reflections and zoom in on the response from 20 Hz to 3 kHz:
Note the vertical 1 dB increment vertical scale. As a brief refresher course, our ears/brain can notice about a 1 dB change in SPL relative to frequency. Here is one way to test your ears for this. We tend to notice peaks more so than dips and our ears follow the envelope of the frequency response.
Let’s say the black horizontal line is the flat target response we are after. This chart correlates well to what I am hearing when I listen to the 1266 TCs. The lower midrange is a little colored (sounds a bit hollow or thick depending on how you hear it) with that 4 dB peak to peak variation between 250 Hz and 450 Hz.
The 6 dB peak to peak variation that spans a broad range from 500 Hz to 2.3 kHz, with a peak at 1.2 kHz and dip at 2 kHz does color the midrange sound. The 1 kHz peak is too much for my ears. Not only the tonal response, but it brings the lead vocals and instruments in that frequency range too far forward in the mix. It sounds too dynamic in that range and a bit separated from the overall mix. The 2 kHz dip tends to make the upper midrange of the vocals sound recessed, or again, a bit hollow sounding.
From a stereo imaging perspective, the perceived stereo image in the 1 kHz range is so far forward in the mix, it actually adds height to the overall image, which should not be there. Rather than a 180 degree horizontal panoramic sound, there is height in the 11am to 1pm position and pulled too far forward in the mix.
Let’s eq the 1266 TCs to flat or neutral response. I don’t use PEQ’s as they are too crude of a tool to invert the minimum phase response with any level of accuracy or precision towards a flat target response. I use high resolution, minimum phase FIR filters with the equivalent of 32,768 graphic eq sliders to fine tune the response.
Aside from the +- 0.5 dB ripple at 300 to 400 Hz, the headphone easily measures +- 0.25 dB from 20 Hz to 3 kHz. Now were are talking about the same tolerance of frequency response as our DAC’s and headphones amps, going directly into your ears.
When I listen to the 1266 TC’s with FIR correction, the mix is perfectly balanced, not only in the tonal response, but also the stereo image is a continuous 180 horizontal plane with no discontinuities. Now the lead vocals and instruments “sit in the pocket” of the mix like they should and not so far forward to almost sound disembodied from the mix. Source: 10 year pro recording/mixing engineer.
Here is the “normal” full frequency scale comparison:
Ruler flat response to 3.5 kHz and then no more eq. Remember, because any headphone measurement technique/analysis is not going to be able to separate the direct sound from the internal reflections due to the wavelengths involved. There is also one’s individual HRTF to take into consideration. The purpose here is just to eq the headphone sans HRTF and internal surface reflections.
With this filter, the treble sounds the same without correction, but with a smoother bass response and nicely cleaned up midrange frequencies.
I created a 2nd filter that brings back the low bass extension, but still has the ruler flat response through the midrange:
The 1266 TCs uncorrected response is -3dB at 21 Hz. With the bass extension, it is -3dB at 18 Hz with a 1.5 dB lift in the low bass similarly following the uncorrected measurement.
A third filter with both bass and treble extension was also designed.
To complete the process, let’s look at the timing (step) response of the headphone.
Correcting the frequency response also corrected the step (and phase) response. No timing distortions of any kind. The 1266 TC is a near perfect minimum phase device.
I used the Lynx Hilo headphone output which did a great job of driving these relatively inefficient headphones. I also have a Class A headphone amp I built a long time ago from the “Audio Amateur” magazine that had no problem driving these headphones cleanly to levels way beyond what my ears could take.
I built a convolver that is designed to level match FIR filters and switch filters instantly while listening to music. This allows me to focus on the design and sound of the eq without worrying about the mechanics of how to compare filters or no filter.
Note that these FIR convolution filters only cut and never boost frequencies. That means there is some level of filter insertion loss that one can manually compensate for by adding positive digital gain. In the case of these convolution filters:
FS1 neutral = -3.9 dB of insertion loss
FS2 neutral plus bass extension = -5.2 dB
FS3 neutral plus bass and treble ext = -4.6 dB
High resolution DSP FIR filtering allows smoothing of the frequency response to a tolerance that we expect from our DAC’s and headphone amplifiers. This level of headphone eq accuracy and precision cannot be achieved any other way.
John’s Subjective Listening Impressions:
Like The Computer Audiophile and his SR1a, I was interested in taking my 1266 TCs to the next level. But honestly… I didn’t know what that would sound like. How do you improve on an already incredible sounding headphone?
For those with interest… My audio chain consists of Roon > Intel NUC > Lumin T2 > Simaudio Moon 600i > Abyss AB-1266 Φ TC.
Mitch provided me with three custom filters for the 1266 TC. One neutral, another with a bit of bass extension, and the third filter with both bass and treble extension. For ease of use, he also provided his Hang Loose Convolver application (sold separately) to allow for quick and seamless transitions between the three filters.
After many hours and genres of music… Here are some details of what I heard.
Like many, I have a list of tracks that I have listened to many hundreds of times. For me, these tend to be from artists that reach deep into my soul like Hans Zimmer, Vacant, Emancipator, and my recent discovery of Max Cooper but occasionally, I’m drawn back to the ambient and layered vocals of Enya…
“Pax Deorum” by Enya from The Memory of Trees... Love, Love, Love it! According to Roon, I have listened to this track 83 times. Since I’ve only been using Roon for two years, and the album came out in 1995, I expect the play count to be many times higher. That being said, using the 1266 TC Neutral Filter makes the track brand new again. Enya’s voice is more precise, beautiful, and haunting. As with all things in audio, a little can go a long way. There are no night and day differences in the track. But the space between the instruments and Enya’s voice is now clean and void of harshness. What I love most about the filter is, the distorted horn-like (think seashell at your ear) sound around her voice is now gone. Enya’s voice now sits back in the mix providing a somewhat larger sound stage. Now for the bass. Wow! The 1266 TCs are bass monsters. This track really showcases the ability of this planar. But like the vocals, the neutral filter cleans up the bass. This allows for more of the sub-bass character of this track to pass through the headphone. This is not a bass boost novelty! Mitch has managed to transform one of my favorite tracks for the better… much better.
“Now We Are Free” by Hans Zimmer, Lisa Gerrard From the Gladiator, film score. Play count: 137… This film score changed my musical preferences forever. Gerrard’s vocal is a dialect or “glossolalia” of her own imagination. It’s bloody fantastic. She is able to convey the heart-felt emotion, victory, and honor of the scene without contaminating it with the barriers of language. For this track, I enjoy both the neutral and bass and treble filter versions. Like before, the seashell-like sound of certain frequencies has been eliminated allowing Gerrard's voice to sound cleaner and in sections, less shouty. The bass and treble extensions do just that. They both add very subtle changes to the leading edge of the bass drum notes as well as improving the out-of-your-head sounds at 3:02. The instrument choices of Zimmer become more noticeable because of the superior imaging these filters provide. Can’t get enough.
Electronic ambient music is my favorite genre. Vacant and Emancipator get most of my attention these days. This music tends to be well recorded, loaded with macro detail, and has very very precise imaging. For my taste, this music needs little sonic improvement. So why use filters at all? Well, that’s what I thought BEFORE trying Mitch’s filters on my favorites. For this genre, I enjoy the neutral filter the most. It has the most improvement in sound. With the filter enabled, I am mesmerized by the clarity of bass and snare drum hits. The edge of the hit is quicker and more defined which leads to a stronger impact/slam... Even at lower volumes!
“All Through the Night” by Emancipator from Safe In The Steep Cliffs. There is so much detail in this track that it is difficult to focus on one thing. It’s a musical rollercoaster that I never want to end. With the neutral filter, the sound of bass guitar notes is smoother (without gaps) in their frequency response. The cello has lost the unnatural horn-like sound. The snare drum and hi-hat are more present and alive… Even the subtlety of the crickets has been enhanced… The play count of this track is climbing!
“Nocturnal” by Vacant from Nocturnal. Play count… 273! This track is less detail intense but offers a robust ambient sound. This is the sound I love. This track has no walls to contain it… Imaging and soundstage galore! Turn on the neutral filter, close your eyes and the music jumps out of your head. This one is difficult for me to explain. Clarity is dramatically improved… which for this track means exaggerated. It’s wild. Wait until you hear the shells hit the floor sound of the slide of the gun. Awesome!
I used all three filters to test my favorite genres of music. The results were the same for everything I listened to. The change is subtle... but it's a change I really enjoy! I now use the neutral filter most, along with the one with extended low and high frequencies. One of the two is active at all times. As good as the stock 1266 TCs are/were… There's no going back now.
With that being said… I understand how easy it can be to get caught up in the audiophile hype, lingo, and high expectations. Will these filters give you a “night and day” experience with your 1266 TCs? No. Are the improvements to the music subtle? Yes. Were the risks associated with the filter-making process worth it? Absolutely! If you are looking to take your 1266 TCs to the bleeding edge of performance, I highly recommend these filters.
Auditioning the 1266 TC filters with music demo
You will need Audacity: Download
... and a way to hook your Abyss TCs to your computer... preferably with a separate DAC and amp.
Download demo files here: Download
Open the files (one at a time) with Audacity...
Click "View" on the toolbar and choose "Mixer Board"
Click the "Solo" button on "Bypass" Filters, starting from the left: Bypass, Neutral, Bass Ext, and Bass & Treble Ext.
Click the Green Play button...
Then click the individual solo buttons in the "Mixer Board" to instantly and seamlessly switch between filters.
About the Author:
John Teixeira, a connoisseur of all things audio, especially ambient, binaural, OOYH (Out Of Your Head), and multichannel recordings. On most days, John is listening to audio/music for at least 10 hours of the day. It has become an obsession for him to dive deep into the detailed and minute world of sound. John has recently taken on a new passion project in the development of PRIRs for exclusive use in the Smyth-Research, Realiser A16. There is much to be explored in the new world of Dolby Atmos Music. More info about this project can be found here on Audiophile Style.