[Binaural38]

I recently had the opportunity to spend 42 minutes with a Stax SR-X9000 and SRM-T8000 brought to the Toronto Audiofest by Headfoneshop.




Indeed beautiful and huge in person. I believe the outer diameter is comparable to the Susvara and Shangri-La Jr., likewise the Audio-Technica ATH-ADX5000 which I shortly heard at the show and can only describe as having sounded properly neutral and hence agreeable as suggested by its measured frequency response.



Pleasingly large earpad apertures for someone coming from the HiFiMan Arya Stealth and Meze Elite Tungsten. For context, my documented audio journey starts in https://www.head-fi.org/threads/mez...eadphone-official-thread.959445/post-17509345 (post #4,541) and the chain of posts its links to, post #4,665 of that thread covering my approach to EQ, then post #5,152 showing my latest EQ results as aided by in-ear microphones (includes comparison to the Sennheiser HE-1), my finally combining my outdoor in-ear microphone HRTF measurements with the Reaper DAW and SPARTA's Binauraliser NF, AmbiRoomSim, and AmbiBIN plug-ins, the latter two finally presenting a nigh convincing head-tracked imaging of an orchestra positioned 14 meters away.

Given this, my main reason for interest in the X9000 other than its looks was the desire for an ultra-low distortion (confirmed by ASR's measurements of the 009S) electrostatic headphone with exquisitely sharp and clean transients (which I've found is one thing you can't really EQ), and a frequency response (which you can find in post #1,346 of this thread) that is easy to EQ to diffuse field. Per post #1,346, it indeed looks quite neutral, perhaps with a boosted midrange, then impressively no major treble nulls until around 1.6 kHz which could be beneficial for allowing binaural head-tracking to EQ in the natural treble nulls that move depending on the direction of the sound source. If someone could listen to a treble and top octave sine sweep through their X9000 and confirm whether they can hear any sharp/deep dips between 6 kHz and 15 kHz, that would be great. Otherwise, the shallow high-Q jaggedness of that treble gives me some concerns regarding the driver quality or its control of resonances. Otherwise, the hype did have me working on first saving up for a Mjolnir Audio Carbon CC as a high-power, low-distortion, solid-state driver, after my seeing measurements showing that the Topping EHA5 doesn't cut it.

Comfort:

Finally, unto my subjective impressions, the first thing is that despite my hopes of pleasant earpad spaciousness, the comfort just somehow didn't cut it for me. I was apparently the first person there to complain about a concentration of pressure on the upper and lower parts of the pads, there not being enough pressure around my temples and at the back of the pads. Positioning the pads higher didn't help much and approached the unacceptable case of my lower ear lobes being touched. The HiFiMan Serenity pads on the other hand fit my head perfectly, the Meze Elite despite its flat profile still being very comfortable and managing a full seal with great sub-bass extension, its main flaw being the pads' flexibility having the cups moving around a lot more when I rotate my head toward the extremes.

I was really hoping the X9000 pads would fit snugly and be more stable for use in endgame binaural head-tracking, but with this inability to get a good seal. I don't know if third-party pads are or will ever be offered (so long as the bass extension is good, I'm going to EQ it anyways). Maybe I could do a mod that inserts a spacer to shape the pads to better conform at the front and back. At least there was no driver crinkle. I had inside the Headfoneshop shortly tried the Audeze CRBN which was unfortunately driven by the Stax SRM-D10 which could barely get it loud enough (I likewise couldn't properly assess the sharpness of single-sample transients), this being surprisingly the first time I had heard a quite noticeably reduced soundstage or shrunken imaging; the pads though soft were too claustrophobic or unwieldy for me, my finding that pad size and feel can have subjective effects on image size; the driver crinkle was also horrid. I did shortly during the audio show just try on the 009S whose pads were quite more comfortable, having just a bit of driver crinkle, but its frequency response doesn't meet my standards. Otherwise, this and my relative satisfaction with the Meze Elite Tungsten may be the deal-breaker saving me $20k CAD.

Sound:

Expectedly from the frequency response, it simply sounded "agreeable" as I had found all other headphones at the audio show that were some form of neutral with sufficient ear gain. There were some points of noise or conversation nearby, but I'd say it wasn't too noisy or busy during my session. At best, the tonal presentation sounded comparable to my final Meze Elite EQ shown in post #4,665, dynamics and bigness (e.g. with Boulez' recording of Mahler 5) likewise being comparable. My Meze Elite EQ was already having excellent recordings like Gardiner's Beethoven 4 sound exquisitely rich and details, whereby at best, the stock X9000 sounded similar. Transients in Rodrigo y Gabriela's "Oblivion" sounded clean and sharp. Track after track, it simply sounded "good" or "comparable", my already having a quite excellent sound at home, though I know that in isolation, the main limitation of the Meze Elite is that single-sample transients like in http://pcfarina.eng.unipr.it/Acustica-samples/Dirac.wav don't sound as freakishly sharp and incisive as through the HiFiMan Arya Stealth with its sub-micron diaphragm (though you have to crank them up pretty loud for these differences to be noticeable, the T8000 having not been able to get this transient loud enough for me to tell whether the X9000's single-sample transients did sound sharper or comparably sharp; the same went for the Sennheiser HE-1). Yosi Horikawa simply sounded good, nothing special. Likewise, the panning of the distant sound sources in didn't blow my mind like the Meze Elite did when I first auditioned it (that effect wore off on all subsequent listens). Regardless, there was always the constant nagging of this discomfort with the pads.

Conclusion:

So it's a bummer that comfort has seemed like the big deal-breaker in my selecting the X9000 as an end-game EQing and binaural head-tracking platform, but at least that may save me a lot of money to spend on other hobbies. Meanwhile, the HE1000se which I did try at the audio show with third-party sheepskin memory foam pads was exquisitely comfortable, its simply sounding "good" by my standards, its otherwise having no magical properties for classical music at least when driven with a Chord Mojo 2 isolated with fancy AudioWise RF Stop gear. If the HE1000se or whatever HiFiMan comes up with next manages to have competitive distortion performance to the Meze Elite, then it may be worth it for the faster transients.

I recently had the opportunity to spend 42 minutes with a Stax SR-X9000 and SRM-T8000 brought to the Toronto Audiofest by Headfoneshop.




Indeed beautiful and huge in person. I believe the outer diameter is comparable to the Susvara and Shangri-La Jr., likewise the Audio-Technica ATH-ADX5000 which I shortly heard at the show and can only describe as having sounded properly neutral and hence agreeable as suggested by its measured frequency response.



Pleasingly large earpad apertures for someone coming from the HiFiMan Arya Stealth and Meze Elite Tungsten. For context, my documented audio journey starts in https://www.head-fi.org/threads/mez...eadphone-official-thread.959445/post-17509345 (post #4,541) and the chain of posts its links to, post #4,665 of that thread covering my approach to EQ, then post #5,152 showing my latest EQ results as aided by in-ear microphones (includes comparison to the Sennheiser HE-1), my finally combining my outdoor in-ear microphone HRTF measurements with the Reaper DAW and SPARTA's Binauraliser NF, AmbiRoomSim, and AmbiBIN plug-ins, the latter two finally presenting a nigh convincing head-tracked imaging of an orchestra positioned 14 meters away.

Given this, my main reason for interest in the X9000 other than its looks was the desire for an ultra-low distortion (confirmed by ASR's measurements of the 009S) electrostatic headphone with exquisitely sharp and clean transients (which I've found is one thing you can't really EQ), and a frequency response (which you can find in post #1,346 of this thread) that is easy to EQ to diffuse field. Per post #1,346, it indeed looks quite neutral, perhaps with a boosted midrange, then impressively no major treble nulls until around 1.6 kHz which could be beneficial for allowing binaural head-tracking to EQ in the natural treble nulls that move depending on the direction of the sound source. If someone could listen to a treble and top octave sine sweep through their X9000 and confirm whether they can hear any sharp/deep dips between 6 kHz and 15 kHz, that would be great. Otherwise, the shallow high-Q jaggedness of that treble gives me some concerns regarding the driver quality or its control of resonances. Otherwise, the hype did have me working on first saving up for a Mjolnir Audio Carbon CC as a high-power, low-distortion, solid-state driver, after my seeing measurements showing that the Topping EHA5 doesn't cut it.

Comfort:

Finally, unto my subjective impressions, the first thing is that despite my hopes of pleasant earpad spaciousness, the comfort just somehow didn't cut it for me. I was apparently the first person there to complain about a concentration of pressure on the upper and lower parts of the pads, there not being enough pressure around my temples and at the back of the pads. Positioning the pads higher didn't help much and approached the unacceptable case of my lower ear lobes being touched. The HiFiMan Serenity pads on the other hand fit my head perfectly, the Meze Elite despite its flat profile still being very comfortable and managing a full seal with great sub-bass extension, its main flaw being the pads' flexibility having the cups moving around a lot more when I rotate my head toward the extremes.

I was really hoping the X9000 pads would fit snugly and be more stable for use in endgame binaural head-tracking, but with this inability to get a good seal. I don't know if third-party pads are or will ever be offered (so long as the bass extension is good, I'm going to EQ it anyways). Maybe I could do a mod that inserts a spacer to shape the pads to better conform at the front and back. At least there was no driver crinkle. I had inside the Headfoneshop shortly tried the Audeze CRBN which was unfortunately driven by the Stax SRM-D10 which could barely get it loud enough (I likewise couldn't properly assess the sharpness of single-sample transients), this being surprisingly the first time I had heard a quite noticeably reduced soundstage or shrunken imaging; the pads though soft were too claustrophobic or unwieldy for me, my finding that pad size and feel can have subjective effects on image size; the driver crinkle was also horrid. I did shortly during the audio show just try on the 009S whose pads were quite more comfortable, having just a bit of driver crinkle, but its frequency response doesn't meet my standards. Otherwise, this and my relative satisfaction with the Meze Elite Tungsten may be the deal-breaker saving me $20k CAD.

Sound:

Expectedly from the frequency response, it simply sounded "agreeable" as I had found all other headphones at the audio show that were some form of neutral with sufficient ear gain. There were some points of noise or conversation nearby, but I'd say it wasn't too noisy or busy during my session. At best, the tonal presentation sounded comparable to my final Meze Elite EQ shown in post #4,665, dynamics and bigness (e.g. with Boulez' recording of Mahler 5) likewise being comparable. My Meze Elite EQ was already having excellent recordings like Gardiner's Beethoven 4 sound exquisitely rich and details, whereby at best, the stock X9000 sounded similar. Transients in Rodrigo y Gabriela's "Oblivion" sounded clean and sharp. Track after track, it simply sounded "good" or "comparable", my already having a quite excellent sound at home, though I know that in isolation, the main limitation of the Meze Elite is that single-sample transients like in http://pcfarina.eng.unipr.it/Acustica-samples/Dirac.wav don't sound as freakishly sharp and incisive as through the HiFiMan Arya Stealth with its sub-micron diaphragm (though you have to crank them up pretty loud for these differences to be noticeable, the T8000 having not been able to get this transient loud enough for me to tell whether the X9000's single-sample transients did sound sharper or comparably sharp; the same went for the Sennheiser HE-1). Yosi Horikawa simply sounded good, nothing special. Likewise, the panning of the distant sound sources in didn't blow my mind like the Meze Elite did when I first auditioned it (that effect wore off on all subsequent listens). Regardless, there was always the constant nagging of this discomfort with the pads.

Conclusion:

So it's a bummer that comfort has seemed like the big deal-breaker in my selecting the X9000 as an end-game EQing and binaural head-tracking platform, but at least that may save me a lot of money to spend on other hobbies. Meanwhile, the HE1000se which I did try at the audio show with third-party sheepskin memory foam pads was exquisitely comfortable, its simply sounding "good" by my standards, its otherwise having no magical properties for classical music at least when driven with a Chord Mojo 2 isolated with fancy AudioWise RF Stop gear. If the HE1000se or whatever HiFiMan comes up with next manages to have competitive distortion performance to the Meze Elite, then it may be worth it for the faster transients.

Great review. Bummer about comfort level. Hopefully I can give these a try someday. Never heard them but they’re always at the top of my endgame list.

 

[SolarCetacean]

Tried the SR-X9000 today at CAF for a considerable amount of time on 4 different amps: the LTA Z10e, the BHSE, the Eksonic Aeras, and briefly on the Grand Cayman. On all 4 amps, it was exemplary. It was my favorite headphone of the show by a significant margin.

Comfort is amazing. I think it's more comfortable than my Shangri-La Jr, my current benchmark for comfort. I even think it's a bit more comfortable than the Meze Elite/Empyrean. Barely any clamp force or head pressure. I use the headband sliders at one notch above the bottom. Even though it's technically heavier than the SGL Jr, I think it's more comfortable. The X9K would win on comfort alone.

Physical appearance is exemplary, and even better in person than in pictures. I've thought that the X9K was the best-looking headphone since its announcement and holding it in my hand has not changed that opinion. It also feels sturdy in the hand.

The X9K has decent bass kick, at least on the amps I tried it on. I actually thought it had more impact in the leading edges of notes than the new Meze Empyrean 2. When listening to Adele's "Rolling In the Deep" at the LTA booth, drums had a sharper attack on the X9K than on the Empy2, which had this duller impact even if it had more bass quantity. At no point did I feel the X9K to be lacking in bass, though I didn't test with any songs with significant bass <40 Hz where measurements show a drop-off in extension.

Midrange is where my only quibbles with the X9K's tuning would be. Vocals are pretty forward here, and that's something I'm a bit sensitive to and I generally prefer a more recessed vocal presentation (e.g. Hifiman). On certain amps (more on this later), I thought vocals could be a bit piercing or grating. However, Staxen do forward vocals better than most other vocal-forward headphones in my experience. Also, show conditions required me to turn up the volume more, and higher volume generally causes the frontal elements in the soundscape to sound closer. I may not have the same experience in proper listening environments at home. However, I've heard criticism of "3-blob" imaging on the X9K, and I think I can understand that, since it's like the vocals are very prominent in the center of the soundstage and other things seem to float around them, and I can see how people can interpret that as being blobby.

Treble is wonderful. It's open, airy, and lacks significant harsh peaks while also not being so smooth that detail is blurred. There's bite and incisiveness to treble-heavy notes like cymbals where my SGL Jr will soften them too much to the point where they get smeared. On the SGL Jr, it's like every cymbal is tapped, rather than hit, with this soft "tShhh" instead of sharp "Tss". The treble also conveys a lot of openness and layering in the tuning.

Sounds are perceived as floating around me and emanating into space rather than being beams of sound aimed at me. As an analogy, think of a squirt bottle vs an aerosol spray. The squirt bottle produces a tight stream of water, while the aerosol produces a mist that dissipates into the air. Notes appear and decay more like the aerosol, gradually fading away into space, than just being shot at me like the squirt bottle. That sense of "openness" or "airiness" is a trait that I really value in sound. And the X9K is probably the best I've heard, though I'd need a direct comparison vs the SGL Jr to decide that. The X9K also has good layering of sound, and I don't really perceive any true blobbiness in the sound presentation.

Regarding the amp comparison, I could only compare the Aeras with the BHSE as both of those were next to each other on the same table so I could switch the plug from one to the other. I think I like the BHSE more, though both are very good amps, and I can see people preferring the Aeras too. I thought the BHSE wasn't as sharp in the midrange compared to the Aeras, and the sense of openness was greater on the BHSE. The Aeras seemed slightly warmer, with a bit more heft to the lower midrange and bass, but not really any more impact. These changes were noticeable, but still subtle changes; I would be fine with the Aeras and its compact form factor is certainly a mark in its favor, though HeadAmp may be coming up with something in response...

I thought the X9K sounded very good on the Grand Cayman, but unfortunately I didn't get any dedicated time on it; I was listening to someone else's music at the time. I thought the bass sounded very solid, but I can't tell if that was the amp doing that, or just the mastering of the track (it was a track by Tool beginning with the letter 'I', but I don't recall the exact title). I don't live very far from HeadAmp HQ, so perhaps I could arrange a private demo with the GC. In any case, even without the GC, I think the X9K is an excellent headphone.

Vs other estats: there was a 007 on the BHSE too, and I preferred the X9K to the 007. Some people get triggered when this opinion is voiced here, but for the attributes that I value in sound, like openness and treble refinement, I think the X9K expresses them better than the 007. The X9K is also more comfortable to me. The X9K was also significantly preferable to the DCA Corina (both heard on the LTA Z10e). The Corina just lacks openness. It (and most DCA headphones) is the epitome of that "squirt bottle" sound that I mentioned earlier. The Corina sounds constricted and dull in comparison, and the layering isn't there compared to the X9K. Going from the Corina to the X9K is like loosening the belt with regards to soundstaging; everything gets more open, more separated, more layered, more spacious, and more relaxed. The CRBN comparison is a bit less precise, as I didn't spend as much time on the CRBN. I think the CRBN goes for the bassier, more mid-centric sound like the new Audezes, and it's not bad on its own, but I don't know if I would pick it over the X9K. The X9K still has that "Stax magic" in its overall presentation, and I didn't think it was a slouch in bass either.

This was long-winded, but I wanted to get my impressions written down while they are still fresh on my mind. The X9K is definitely a headphone I want to pick up in the future. I'd love to give it a head-to-head comparison vs my SGL Jr (I didn't bring it today due to lacking a good carrying bag/case for it), because it's a headphone that could actually be considered an upgrade!

 

[number1sixerfan]

Vs other estats: there was a 007 on the BHSE too, and I preferred the X9K to the 007. Some people get triggered when this opinion is voiced here, but for the attributes that I value in sound, like openness and treble refinement, I think the X9K expresses them better than the 007.

An amazing read overall. Really great and detailed impressions.

And on this point, it shouldn't be triggering at all. The x9000 is a league or two above the 007, and even the 009--and I am a 009 zealot lol. For me, the imaging issues you spoke to do stand out and I think the 007 and 009 have a bit more cohesive presentation, but I'd take the x9000 over both of them any day of the week. Now between the Omega, it's more of a toss up and I'd take the Omega due to a bit more natural tone despite the rest of their similarities.

 

[gammi]

Tried the SR-X9000 today at CAF

First off, thanks for your impressions.
Sad I missed CAF, it's a quick train ride away but by the time I realized when it was happening I already had conflicting plans for this weekend.

Same thing happened to me where I heard the X9000 at a show and I just couldn't shake it, it was the only pair at the show that stood out to me as something special. You'll wind up with one within a few months now that you have the bug

Comfort, I agree entirely. They're extremely comfortable and very low clamp force, they just kind of barely rest on your head. That and the fact they let through 99% of outside sounds contributes a lot to them feeling like speakers/the sound is in the room and not on my head. They are the most "I'm not wearing any headphones" of any headphones I have. They just disappear.

Glad we agree on bass. The "estats have no bass" stereotype is definitely incorrect. That <40Hz drop-off? You can actually get it back counterintuitively by breaking the seal. If you lift the bottom of the X9000 off your head ever so slightly you'll hear a bass lift.

I also prefer X9000 over 007 and I bet people who don't would change their minds if they EQed it to match the 007. It's simply better but some people prefer the sound signature of the 007.

I also demoed the Corina along side the X9000 before I owned one and same thing, it felt too closed. The AMTS tech they have in blocks some outside sound just like in the Expanse. I thought the Corina was good but if I was looking for an estat type sound, I wouldn't choose it, I think it sounds more like a planar actually because the AMTS muffles it and you lose some of that estat air magic.

 

[WernerW]

Hi gammi,
this is indeed interesting to read:
"That <40Hz drop-off? You can actually get it back counterintuitively by breaking the seal. If you lift the bottom of the X9000 off your head ever so slightly you'll hear a bass lift."

Exactly the same was mentioned by Michael in the comments section of "Director's Garage".



In the comments section to that video it was said (citations by Michael):
"Sorry you asked for the sound— If you’ve ever owned or heard the Abyss 1266 TC, it’s a bit like that bass thump. Maybe not quite as extreme, but the soundstage gets even wider and the bass kicks much more than with the headphone sealed. I think a pad that opens up the seal (like Abyss did for the Diana) would be a great mod for the X9000. Ya feel a little goofy sitting there holding the headphone off your head a bit, but it produces a great effect."

and:

"If you get the chance, tell me what you hear when you pull them off your head a bit. I’d like to know I’m not crazy! Cheers! Thanks for watching."

I have not yet tested this by myself although I have an X9000.
I still have to do and try this!

And it does not sound illogical as the 'Abyss 1266 Phi TC' has this very loose seal, either.

In a video (@HiFiInsider)

'HiFi Insider' says (citation) "[...] you don't want a tight fit, you don't want it to clamp [...] because you would lose bass response [...]".

So, maybe this might be a very exciting experiment with the X9000.
Cheers,
Werner

 

[gammi]

Yup, director's garage is actually where I got the idea to try it. He's right.
Simply pull outward on the cable connectors to lift the bottom slightly off your head and you'll hear it too.

I found it too bloomy and as such didn't pursue a way to make it stay like that without me holding it. A simple piece of foam double sided taped to the earpad should do it though.

 

[yaps66]

Tried the SR-X9000 today at CAF for a considerable amount of time on 4 different amps: the LTA Z10e, the BHSE, the Eksonic Aeras, and briefly on the Grand Cayman. On all 4 amps, it was exemplary. It was my favorite headphone of the show by a significant margin.

Comfort is amazing. I think it's more comfortable than my Shangri-La Jr, my current benchmark for comfort. I even think it's a bit more comfortable than the Meze Elite/Empyrean. Barely any clamp force or head pressure. I use the headband sliders at one notch above the bottom. Even though it's technically heavier than the SGL Jr, I think it's more comfortable. The X9K would win on comfort alone.

Physical appearance is exemplary, and even better in person than in pictures. I've thought that the X9K was the best-looking headphone since its announcement and holding it in my hand has not changed that opinion. It also feels sturdy in the hand.

The X9K has decent bass kick, at least on the amps I tried it on. I actually thought it had more impact in the leading edges of notes than the new Meze Empyrean 2. When listening to Adele's "Rolling In the Deep" at the LTA booth, drums had a sharper attack on the X9K than on the Empy2, which had this duller impact even if it had more bass quantity. At no point did I feel the X9K to be lacking in bass, though I didn't test with any songs with significant bass <40 Hz where measurements show a drop-off in extension.

Midrange is where my only quibbles with the X9K's tuning would be. Vocals are pretty forward here, and that's something I'm a bit sensitive to and I generally prefer a more recessed vocal presentation (e.g. Hifiman). On certain amps (more on this later), I thought vocals could be a bit piercing or grating. However, Staxen do forward vocals better than most other vocal-forward headphones in my experience. Also, show conditions required me to turn up the volume more, and higher volume generally causes the frontal elements in the soundscape to sound closer. I may not have the same experience in proper listening environments at home. However, I've heard criticism of "3-blob" imaging on the X9K, and I think I can understand that, since it's like the vocals are very prominent in the center of the soundstage and other things seem to float around them, and I can see how people can interpret that as being blobby.

Treble is wonderful. It's open, airy, and lacks significant harsh peaks while also not being so smooth that detail is blurred. There's bite and incisiveness to treble-heavy notes like cymbals where my SGL Jr will soften them too much to the point where they get smeared. On the SGL Jr, it's like every cymbal is tapped, rather than hit, with this soft "tShhh" instead of sharp "Tss". The treble also conveys a lot of openness and layering in the tuning.

Sounds are perceived as floating around me and emanating into space rather than being beams of sound aimed at me. As an analogy, think of a squirt bottle vs an aerosol spray. The squirt bottle produces a tight stream of water, while the aerosol produces a mist that dissipates into the air. Notes appear and decay more like the aerosol, gradually fading away into space, than just being shot at me like the squirt bottle. That sense of "openness" or "airiness" is a trait that I really value in sound. And the X9K is probably the best I've heard, though I'd need a direct comparison vs the SGL Jr to decide that. The X9K also has good layering of sound, and I don't really perceive any true blobbiness in the sound presentation.

Regarding the amp comparison, I could only compare the Aeras with the BHSE as both of those were next to each other on the same table so I could switch the plug from one to the other. I think I like the BHSE more, though both are very good amps, and I can see people preferring the Aeras too. I thought the BHSE wasn't as sharp in the midrange compared to the Aeras, and the sense of openness was greater on the BHSE. The Aeras seemed slightly warmer, with a bit more heft to the lower midrange and bass, but not really any more impact. These changes were noticeable, but still subtle changes; I would be fine with the Aeras and its compact form factor is certainly a mark in its favor, though HeadAmp may be coming up with something in response...

I thought the X9K sounded very good on the Grand Cayman, but unfortunately I didn't get any dedicated time on it; I was listening to someone else's music at the time. I thought the bass sounded very solid, but I can't tell if that was the amp doing that, or just the mastering of the track (it was a track by Tool beginning with the letter 'I', but I don't recall the exact title). I don't live very far from HeadAmp HQ, so perhaps I could arrange a private demo with the GC. In any case, even without the GC, I think the X9K is an excellent headphone.

Vs other estats: there was a 007 on the BHSE too, and I preferred the X9K to the 007. Some people get triggered when this opinion is voiced here, but for the attributes that I value in sound, like openness and treble refinement, I think the X9K expresses them better than the 007. The X9K is also more comfortable to me. The X9K was also significantly preferable to the DCA Corina (both heard on the LTA Z10e). The Corina just lacks openness. It (and most DCA headphones) is the epitome of that "squirt bottle" sound that I mentioned earlier. The Corina sounds constricted and dull in comparison, and the layering isn't there compared to the X9K. Going from the Corina to the X9K is like loosening the belt with regards to soundstaging; everything gets more open, more separated, more layered, more spacious, and more relaxed. The CRBN comparison is a bit less precise, as I didn't spend as much time on the CRBN. I think the CRBN goes for the bassier, more mid-centric sound like the new Audezes, and it's not bad on its own, but I don't know if I would pick it over the X9K. The X9K still has that "Stax magic" in its overall presentation, and I didn't think it was a slouch in bass either.

This was long-winded, but I wanted to get my impressions written down while they are still fresh on my mind. The X9K is definitely a headphone I want to pick up in the future. I'd love to give it a head-to-head comparison vs my SGL Jr (I didn't bring it today due to lacking a good carrying bag/case for it), because it's a headphone that could actually be considered an upgrade!

Thanks for sharing! Enjoyed the read!

I am in agreement with your comparisons with Corina and CRBN. I had exactly the same impressions as you did with the Corina and I too found the Corina lacking in the openness and did not enjoy the presentation from the Corina. I am also in agreement with your CRBN comparisons. I do have the CRBN as well and I think that the CRBN is more complimentary and I would pick the CRBN over the X9k for jazz genres. I would pick the X9k anyday for classical music.

The X9k is special and I have no regrets in piecing together the stack to drive them. I do hope you get to own the X9k soon. It looks like you will enjoy it very much, as much as I do!

Kudos to Michael from Director's Garagej (@reeltime). I am a long time subscriber to his channel very much enjoy his videos!

 

[MrHaelscheir]

So after watching the following which had been shared by someone a number of pages ago



I've again fallen in love with the sheer craftsmanship of this thing. Again, as in post #2,659, until a suitable third-party pad comes out that offers a better seal and comfort for my head, this headphone just won't work for me, but I do long to someday add this work of art to my collection alongside the Meze Elite Tungsten.

Anyways, two weeks ago, I had a chance to take in-ear measurements of this headphone at a shop, but under strict instruction from the distributor not to post said measurements per Stax being against sharing such (I can't say how much control they have after you've already purchased a unit). Anyways, as a baseline, consider the below in-ear measurement from my 20-minute session with the Sennheiser HE-1 (I had asked permission beforehand) when it made a stop by Bay Bloor Radio:

Figure 1: Sennheiser HE-1 frequency response measured from the partially blocked left ear canal entrance. Please ignore the green phase response trace as this measurement was taken with the DAC and ADC on separate devices such that samples are misaligned.

Anyways, what impressed me was the sheer smoothness (smoother than expected from existing published measurements) and neutrality of the frequency response other than the known upper midrange dip, and for my ears, a 6 kHz peak that would be easy to EQ down. My philosophy is that "resolution" can be derived from a lack of major peaks or dips (except where they naturally occur in the ear) that would cause certain details to be diminished, or certain details to be overamplified and mask the details within the surrounding frequencies.

Here are my Meze Elite's and HiFiMan Arya Stealth's stock magnitude and phase response versus frequency charts with more accurate and consistent phase responses thanks to both the test signal DAC and measurement ADC being on the same MOTU M2 audio interface this time:

Figure 2: Meze Elite magnitude (upper trace) and phase (lower trace) response versus frequency. Frequency response ripples from driver modes.

Figure 3: HiFiMan Arya Stealth magnitude (upper trace) and phase (lower trace) response versus frequency. Frequency response ripples from driver modes, resonance problems at 4 kHz, and sharp peaks and deeper dips in the upper treble.

Here is my previous "V3 PEQ" EQing result for the Meze Elite for which I had had excellent results:

Figure 4: Meze Elite magnitude (upper trace) and phase (lower trace) response versus frequency with preferred Harman-based EQ with adjustments for perceived peaks; basically, my original EQing by ear to adjust a rough Harman EQ profile resulted a similar but much more wavy result. Here, you can see that the "phase distortions" introduced by minimum-phase EQ actually corrected (smoothed) the phase distortions introduced by the headphone itself.

Finally, here is my left in-ear quasi-anechoic frequency response for a perfectly neutral speaker positioned 30 degrees to the left:

Figure 5: Meze Elite magnitude (upper trace) and phase (lower trace) response versus frequency with quasi-anechoic HRTF EQ applied for a perfectly neutral speaker positioned 30 degrees to the left. The my EQing 3 kHz and the 6 kHz peak down is corroborated by this measurement. "Neutral" never has to mean" lifeless. In this case, it was indeed brighter than average, but if a recording contains exquisite warmth and lushness, it will be passed through unmodified.

(2024-03-19: See "Calibration using threshold of hearing curves" in https://www.head-fi.org/threads/rec...-virtualization.890719/page-121#post-18027627 (post #1,812). Everything I have said about neutral speakers actually having a lot more ear gain than neutral headphones was wrong.)

What these graphs show is that these relatively lower-end large planars have been limited in how well they could control resonances and present an ideal frequency response to the ear for transparency. The Sennheiser HE-1 then presented what was clearly a substantial achievement for that time without DSP (like Warwick Acoustics uses, and as was needed for my Meze Elite). Given that, I will simply say that the Stax SR-X9000 presented similar excellence in frequency response smoothness (smoother than the measurements already shown on post #1,346 and post #2,370 would suggest) and control of resonances, for a fraction of the price. The other estat headphones in post #1,346 possessed major treble peaks which while now probably drilled into the "estat identity" simply aren't accurate to true stereo reproduction; the Stax SR-X9000 does away with those exaggerated peaks toward the path of true fidelity.

The midrange is indeed elevated like has been seen in those existing measurements and as heard by peoples' ears, but nothing is stopping you from EQing that down (other than system-related inconveniences), and as was shown in Figure 4, minimum-phase EQ can actually correct phase distortions (if there were ever more audible than the tonal distortions in the first place).

As for sub-bass extension, for my head, the limited seal at the front and back of the pads did incur a few dB of sub-bass roll-off, but while pressing the cup toward my head for a better seal, it became much more competitive with the Meze Elite hybrid pads. This compression also measurably slightly reduced the upper midrange elevation and partly smoothed the treble peaks and dips; shallow treble nulls are desirable to me since they are easier to EQ up to flat so I can more accurately overlay the nulls present in one's HRTF as applied by a binaural head-tracking renderer. The bass to lower midrange is dead-flat neutral, but with estat distortion performance, nothing is stopping you from EQing in a decent bass shelf.

The measured distortion performance was as excellent as in https://www.audiosciencereview.com/...sr-009s-electrostatic-headphone-review.47390/, which is to say competitive with the Meze Elite, DCA Expanse, and Audeze CRBN, though both these estats are not reaching the unnecessary depths of distortion achieved by the Sennheiser HE-1 as verified by headphones.com in their video. Now, the noise floor even with 4M-length measurements and 8 repetitions for measurement averaging within Room EQ Wizard wasn't ideal, and maybe some of the distortion components originated from the Stax SRM-T8000, so it might measure more impressively in another measurement environment.

It would already be apparent from looking at these Stax headphones' drivers and HiFiMan's that they do not have much damping beyond driver tensioning and air resistance. My theory has been that for absolute "speed" or "cleanness" of sound, you would want a cumulative spectral decay (CSD) that is as clean as possible (though said decay may only be audible when playing transients really loud; I don't think the Stax SRM-T8000 could play said transients to the point of hearing audible decay components). The Meze Elite with its outer-side (back volume) damping had a cleaner CSD than the HiFiMan Arya Stealth which itself had a cleaner CSD than both the Stax SR-X9000 and Sennheiser HE-1 while my closed-back Audio-Technica ATH-M50xBT and Jabra Elite 85h had the cleanest CSDs (virtually nothing in the midrange above the noise floor), but one assessment consistent with that measure as well as what I have heard out of some headphones that I haven't measured is that CSD cleanness tends to be correlated with increased driver damping and hence duller transients. E.g. The DCA Expanse when I auditioned it had very clean transient decay (when listening to http://pcfarina.eng.unipr.it/Acustica-samples/Dirac.wav, a single 0 dBFS sample at 48 kHz sampling rate, very loud), but felt the dullest out of the headphones I auditioned per all the damping and tuning materials used. So "speed" per the sense of incisiveness or how rapidly the driver accelerates does not always correlate with how rapidly the driver comes to a stop. The Final Audio D8000 Pro when I listened to said transients seemed like a combination of reasonable transient sharpness and very good transient cleanness. Another theory I have is that "impact" or transient sharpness may be related to the smoothness of the bass group delay measurement, of which among those headphones I have measured, the HiFiMan Arya Stealth had a particularly smooth bass group delay, its indeed being the most incisive headphone I've heard to date (until I can drive an estat loud enough for a better comparison).

tl;dr: The Stax SR-X9000 is an exceptional headphone with tuning performance, driver control, and EQability rivaling the Sennheiser HE-1, whereby you should all be very happy owners or aspirants.

 

[Thomsen132]

@MrHaelscheir
Very interesting posting and i would have loved to see your measurements of the SR-X9000, but of course i understand, why you won't post it.

This is the frequency response measurement (compensated), which i made with my SR-X9000 (mini ears dsp).
Maybe it is a little bit similar to your measurement and interesting for some people

 

[MrHaelscheir]

@MrHaelscheir
Very interesting posting and i would have loved to see your measurements of the SR-X9000, but of course i understand, why you won't post it.

This is the measurement (compensated), which i made with my SR-X9000 (mini ears dsp).
Maybe it is a little bit similar to your measurement and interesting for some people

Let's say the measurements are most similar to the general features captured in post #1,346 below 8 kHz, but flatter in the bass (that concave arc is probably an artefact of the B&K 4128 rig and a bad seal; otherwise, there is around 1 dB of elevation around 60 Hz on my head when the seal isn't perfect), having a smoother rather than stepped roll-off below 60 Hz (my Meze Elite can sometimes measure as having that sub-bass step like on published measurements and for the 009S, but with smoother corners), sharing some features of the upper midrange shape (e.g. an little dip at 1.39 kHz), smooth and a bit concave through 2 kHz to 4 kHz (the 3 kHz peak seen in post #1,346 is just the natural ear canal resonance which appears particularly sharp in that measurement rig, whereby my partially blocked canal entrance measurements exclude that peak), a round dip and then peak situated around 5 kHz (this can differ depending on your ear geometry), a very shallow 9.5 kHz dip (relative to the preceding drop in magnitude) as opposed to the sharper dip or null induced by some of the other headphones I've measured on my head, and finally a top octave (10 kHz and up) profile somewhat similar to the HE-1 with some more nulls here and there, but not like the Arya.

As for distortion, like with the Meze Elite even when EQed up to my free-field HRTF, even at 100 dB SPL, the distortion components would mainly be second-order and at or below your listening room's noise floor so as to be inaudible.

This shall have been my 100th post on Head-Fi.

 

[SolarCetacean]

Another theory I have is that "impact" or transient sharpness may be related to the smoothness of the bass group delay measurement

Have you ever experimented with dynamic EQ or any sort of square wave measurements? My experience with using dynamic EQ as an expander leads me to theorize that transient impact is due to overshoot of the transient target level, or crest factor. If I add a +3dB dynamic shelf to bass (not just a static +3dB from parametric EQ), the bass transients become more impactful, not just that the bass becomes more present. It's the same feeling of punch that I've experienced with Focal headphones and other subjectively punchy and impactful gear. I have not increased bass level by 3 dB, but added 3 dB to the crest factor of transients in the bass.

This tracks with what Dan Clark has mentioned about his tuning goal. His headphones are tuned to never overshoot the target levels of transients. He mentioned as much in an interview with Passion For Sound, I believe. That's why his headphones have that duller transient. If I add expanding dynamic EQ to a DCA headphone, the transients it renders become sharper, more like other headphones.

Dan Clark referred to this metric as "impulse response overshoot". I'd call it "finite impulse response overshoot" or just "transient level overshoot", as technically the Dirac impulse has infinite height, and thus cannot be overshot. But any real impulse would have a finite height, and the transducer (or signal chain) can overshoot it. Another metric that theoretically could show this is the square wave response, focusing on how much the leading edges of the transition overshoot the steady-state level. Since you've done so many measurements and you mentioned perceived transient sharpness, I was wondering on your take on crest factor and dynamic overshoot.

 

[MrHaelscheir]

Have you ever experimented with dynamic EQ or any sort of square wave measurements? My experience with using dynamic EQ as an expander leads me to theorize that transient impact is due to overshoot of the transient target level, or crest factor. If I add a +3dB dynamic shelf to bass (not just a static +3dB from parametric EQ), the bass transients become more impactful, not just that the bass becomes more present. It's the same feeling of punch that I've experienced with Focal headphones and other subjectively punchy and impactful gear. I have not increased bass level by 3 dB, but added 3 dB to the crest factor of transients in the bass.

This tracks with what Dan Clark has mentioned about his tuning goal. His headphones are tuned to never overshoot the target levels of transients. He mentioned as much in an interview with Passion For Sound, I believe. That's why his headphones have that duller transient. If I add expanding dynamic EQ to a DCA headphone, the transients it renders become sharper, more like other headphones.

Dan Clark referred to this metric as "impulse response overshoot". I'd call it "finite impulse response overshoot" or just "transient level overshoot", as technically the Dirac impulse has infinite height, and thus cannot be overshot. But any real impulse would have a finite height, and the transducer (or signal chain) can overshoot it. Another metric that theoretically could show this is the square wave response, focusing on how much the leading edges of the transition overshoot the steady-state level. Since you've done so many measurements and you mentioned perceived transient sharpness, I was wondering on your take on crest factor and dynamic overshoot.

Interesting. I hadn't heard of dynamic EQ. For square wave measurements, I would expect most overshoot artifacts to just be a result of the bandwidth limiting of the signal by the sample rate, reconstruction filter, and transducer. Otherwise, I remember finding https://www.superbestaudiofriends.org/index.php?threads/burst-response-hd800-sr-207-hd650.3688/ interesting, though one must note that those signals comprised a rectangular pulse being multiplied with a sinusoid in the time domain which leads to convolution of their frequency domain counterparts. I suppose for any given bandwidth or reconstruction filter, one could predict the theoretically "perfect" square wave and its inherent overshoot. As for DCA targeting to "never" have any overshoot, while in control systems, this can be useful mind essential, for audio, I think overshoot is really a natural part of the reproduction of transients as a result of the bandwidth limiting, whereby intentional overdamping to prevent said overshoot may technically be "incorrect". The other option to "never" overshoot the signal is to supply transients of infinite bandwidth, but then your own ears' physical filtering will probably reintroduce said overshoots, but at least that frequency or phase content will be available as opposed to the case where it has been damped away as it may be with DCA.

Otherwise, I would expect any such time-domain overshoot to be predictable from frequency-domain data. As such, these differences in subjective transient sharpness would relate to how the frequency-domain influences bandwidth-limited transient overshoots. As for why this is "harder" to parse from traditional frequency response displays, it's because your average human can't convolve and conduct the inverse Fourier transform on a signal with their eyes. If, transient overshoot can indeed be entirely predicted by frequency-domain measurements (I would expect the effects of damping to show up in such), then the theoretically most "correct" amount of transient overshoot would depend on an individual's HRTF with respect to a perfectly neutral sound source.

Anyways, I have yet to figure out how to take time domain measurements in REW.

 

[SolarCetacean]

Interesting. I hadn't heard of dynamic EQ. For square wave measurements, I would expect most overshoot artifacts to just be a result of the bandwidth limiting of the signal by the sample rate, reconstruction filter, and transducer. Otherwise, I remember finding https://www.superbestaudiofriends.org/index.php?threads/burst-response-hd800-sr-207-hd650.3688/ interesting, though one must note that those signals comprised a rectangular pulse being multiplied with a sinusoid in the time domain which leads to convolution of their frequency domain counterparts. I suppose for any given bandwidth or reconstruction filter, one could predict the theoretically "perfect" square wave and its inherent overshoot. As for DCA targeting to "never" have any overshoot, while in control systems, this can be useful mind essential, for audio, I think overshoot is really a natural part of the reproduction of transients as a result of the bandwidth limiting, whereby intentional overdamping to prevent said overshoot may technically be "incorrect". The other option to "never" overshoot the signal is to supply transients of infinite bandwidth, but then your own ears' physical filtering will probably reintroduce said overshoots, but at least that frequency or phase content will be available as opposed to the case where it has been damped away as it may be with DCA.

Otherwise, I would expect any such time-domain overshoot to be predictable from frequency-domain data. As such, these differences in subjective transient sharpness would relate to how the frequency-domain influences bandwidth-limited transient overshoots. As for why this is "harder" to parse from traditional frequency response displays, it's because your average human can't convolve and conduct the inverse Fourier transform on a signal with their eyes. If, transient overshoot can indeed be entirely predicted by frequency-domain measurements (I would expect the effects of damping to show up in such), then the theoretically most "correct" amount of transient overshoot would depend on an individual's HRTF with respect to a perfectly neutral sound source.

Anyways, I have yet to figure out how to take time domain measurements in REW.

I find it odd that so many people, even technical people in the headphone field, have never heard of dynamic EQ. It seems to be a pretty normal tool within music production circles. I use FabFilter Pro-Q 3 as my dynamic EQ plugin, in VST form.

If overshoot was part of the FR, then it would be possible to adjust overshoot-related phenomena via parametric EQ. Thus far, I have not been able to do so. Dynamic EQ adds that sense of punch while not changing the tonality in the same way that static parametric EQ does. While I think it's still possible that just changing the static measured FR can produce that same transient effect, it would involve looking at and modifying parts of the FR that we just aren't looking at right now, because it's very easy to make bass punchier using dynamic EQ just by expanding the size of the transient. I suspect that a set of static parametric EQ filters that would only boost the leading edge of a transient, then have little effect afterwards (just like a dynamic filter) would be pretty convoluted, if it is even possible.

My suspicion is that overshoot would factor into some sort of distortion measurement. If the signal contains a 10 dB crest, and that transient is rendered as 13 dB, that leading edge rises more in the same amount of time, so it must contain higher-frequency content, which should show up as distortion vs the original signal. Perhaps the extremely low distortion levels observed in DCA headphones are related to their lack of overshoot. Perhaps less distortion isn't better after a certain point...?

 

[SolarCetacean]

Anyways, back to the SR-X9000. I went to The Source AV outside LA to try the X9000, but alas, it wasn't there! The store had been burglarized a few weeks ago and the X9K was stolen in the robbery! It took an hour and a half to get there from Pasadena, so it was quite disappointing (for both me and TSAV). I had hoped for a second, more critical listen to follow up on my CAF experience and see if I really wanted to buy it. But I had a good time anyways listening to other headphones. If you see any suspiciously cheap X9Ks on the used market, it might be worth double-checking its origin.

I also learned a few interesting things about the business backend of Stax. According to Jason at TSAV, only 12 SR-X9K units are shipped to the US every 6 months. The vast majority of produced X9Ks are kept in Japan for the domestic market. Only 6 units have been shipped to Canadian dealers since the headphone was released two years ago. Honestly, this makes me tempted to import, because it seems they are relatively plentiful in Japan and the yen is so weak compared to USD right now that I'm basically getting 30% off the list price.

Another dealer comment on a different forum was that in late July of this year, Stax shipped their 1000th X9000. It took just under two years to build 1000 units.

 

[MrHaelscheir]

I find it odd that so many people, even technical people in the headphone field, have never heard of dynamic EQ. It seems to be a pretty normal tool within music production circles. I use FabFilter Pro-Q 3 as my dynamic EQ plugin, in VST form.

If overshoot was part of the FR, then it would be possible to adjust overshoot-related phenomena via parametric EQ. Thus far, I have not been able to do so. Dynamic EQ adds that sense of punch while not changing the tonality in the same way that static parametric EQ does. While I think it's still possible that just changing the static measured FR can produce that same transient effect, it would involve looking at and modifying parts of the FR that we just aren't looking at right now, because it's very easy to make bass punchier using dynamic EQ just by expanding the size of the transient. I suspect that a set of static parametric EQ filters that would only boost the leading edge of a transient, then have little effect afterwards (just like a dynamic filter) would be pretty convoluted, if it is even possible.

My suspicion is that overshoot would factor into some sort of distortion measurement. If the signal contains a 10 dB crest, and that transient is rendered as 13 dB, that leading edge rises more in the same amount of time, so it must contain higher-frequency content, which should show up as distortion vs the original signal. Perhaps the extremely low distortion levels observed in DCA headphones are related to their lack of overshoot. Perhaps less distortion isn't better after a certain point...?

My "technical" origins are mainly from circuits and signals courses in my computer engineering degree, hence the concert of frequency response being familiar when I first sought out my ATH-M50xBT back in 2019. Otherwise, I do not "produce" music, unless I am right in front of a physical piano.

Regarding overshoot, the reason parametric EQ may not do the trick on its own is that it may also depend on a very specific corresponding phase response that we may additionally not know how to best manipulate. As for dynamic EQ, I'm curious as to how it looks for harmonic distortion under a sine sweep. Otherwise, I would argue that those square wave overshoots would be a case of linear distortion which is captured in the plain magnitude and phase versus frequency response. I have yet to match two headphones to the same magnitude response with minimum phase EQ and see whether the phase response ends up being the same.

As for DCA's distortion measurements, at least based off of Jude's measurements, it is "merely" competitive with the Meze Elite and Audeze CRBN, both surely having sharper subjective transient effects. As for ASR's measurements, I suppose the 009S does have worse lower midrange distortion than the DCA Expanse. I can say that the Sennheiser HE-1 had decent impact and transient sharpness.

solderdude's measurements in https://diyaudioheaven.wordpress.com/headphones/measurements/focal/utopia/ and whatnot show step responses. Per https://diyaudioheaven.wordpress.com/headphones/measurements/brands-a-i/dt-1990-pro/, Beyerdynamics while not having the most competitive distortion at least in the bass show a case of an exquisitely clean CSD coinciding with a noticeably rounded step response. The Susvara in https://diyaudioheaven.wordpress.com/headphones/measurements/hifiman/susvara/ is a case of a poor CSD coinciding with an initial "bump" in the step response that might correlate with "impact". Audezes in solderdude's measurements can have similarly shaped step responses, so I guess it's plausible. I'd say the D8000 in https://diyaudioheaven.wordpress.com/headphones/measurements/final/d8000/ is indeed like I heard (for the D8000 Pro) a case of a good balance of a reasonably clean CSD coinciding with a step response maybe as sharp as the Focal Utopia (both of which don't overshoot).

Also, that figure on how many X9000s have been shipped to Canada is surprising. There are two Stax dealers in Toronto, whereby last I remember, one had it in stock and for the Toronto Audiofest managed to have the distributor open one as a demo unit. Quite scarce.