Mad Lust Envy's Headphone Gaming Guide: (8/18/2022: iFi GO Blu Review Added)
Dec 16, 2021 at 9:41 AM Post #47,536 of 48,562
Installed the SXFI App again, did the head mapping, and it appeared gen3, I guess it has only to do with the scanning of your head this new gen3, probably a new algorithm to your personal profile.
Let me know if you like it more than Gen2.

I don't have my measurements from Gen2 anymore but I can share the frequency response from the Front Left speaker to the left ear for my Gen3 profile.

I measured GC7 with SXFI in the following scenarios:
1. With HD650 profile.
2. With K701 profile (K702 is K701 in black with removable cable).
3. With Emu Teak profile.
4. With Unknown Headphone profile.

Then I added the measurements of each respective headphone (from Oratory1990's website or Crinacle's for Emu Teak). For Unknown Headphone profile, I added the AKG K371's measurements as it's a good representation of the Harman Target.

I also included Harman Target in the graph for reference.
After all, HT is supposed to represent the sound as recorded by a dummy head, from a loudspeaker in a treated troom at 30 deg. angle IIRC.

Provided two graphs with varying degrees of smoothing.

So the end result (for my profile) for the headphones which have their SXFI profiles kinda looks like the Harman curve, but with reduced mid-bass to low-mid region and the lower treble being good few dBs higher than the Harman Target. Then there is a treble drop around 8-10k and an increase at the end of the hearing range.
From the headphones in the graph, K701 seems to have the smoothest upper treble response and I personally think it sounds best (even before I measured this).

The K371, which doesn't have their own profile, end up having similar resulting frequency response with SXFI to the other headphones. With the exception of bass, which is increased a lot in the sub region.

A conclusion I can draw from this is that if your headphones don't have a profile in SXFI App, equalising them to the Harman Target with a bit less bass could make them sound as intended with SXFI. Personally, I also like to reduce the 2 and 4 kHz sliders in Creative app's EQ to reduce the shoutiness when listening to music. You can also add few dBs in the 250 Hz slider for extra warmth.
 

Attachments

  • SXFI-48thoct-smoothing.png
    SXFI-48thoct-smoothing.png
    321.7 KB · Views: 0
  • SXFI_psy-smoothing.png
    SXFI_psy-smoothing.png
    204.8 KB · Views: 0
Dec 16, 2021 at 12:08 PM Post #47,537 of 48,562
My understanding of the sxfi, is that creative servers based on your head mapping, creates a personal algorithm which has to be uploaded to the sbxfi chip, very clever because you can change completely the surround effects, if the sbxfi has power enough and it has enought memory to be flashed, so is not a final product, which can be improved over and over.

The sbxfi app now works better, doesn't crash or forces close when doing the scanning of the new head mapping, or probably just got lucky.
 
Dec 16, 2021 at 12:20 PM Post #47,538 of 48,562
My understanding of the sxfi, is that creative servers based on your head mapping, creates a personal algorithm which has to be uploaded to the sbxfi chip, very clever because you can change completely the surround effects, if the sbxfi has power enough and it has enought memory to be flashed, so is not a final product, which can be improved over and over.

The sbxfi app now works better, doesn't crash or forces close when doing the scanning of the new head mapping, or probably just got lucky.
You're right, that's what happens in an overview. Likely the SXFI dsp does convolution with impulse responses that you download from your online profile to the device. Just guessing:)

Because of that, they can keep improving it like you wrote.

I wish they would offer some kind of calibration options to fine tune the effect and correct the machine learning guesswork. Who knows, maybe we'll see Gen4:)

There is a market for that, even Sean Olive mentioned Harman may be working on Speaker virtualization tech for headphones in a recent interview with headphones.com on YT. The future is bright:) That is if I remember correctly.
 
Last edited:
Dec 16, 2021 at 12:37 PM Post #47,539 of 48,562
You're right, that's what happens in an overview. Likely the SXFI dsp does convolution with impulse responses that you download from your online profile to the device. Just guessing:)

Because of that, they can keep improving it like you wrote.

I wish they would offer some kind of calibration options to fine tune the effect and correct the machine learning guesswork. Who knows, maybe we'll see Gen4:)

There is a market for that, even Sean Olive mentioned Harman may be working on Speaker virtualization tech for headphones in a recent interview with headphones.com on YT. The future is bright:) That is if I remember correctly.

Probably the sxfi chip is just an arm cpu, owing to today's standards, with relative processing power and multi core capabilities, if it doesn't run a complete operating system in top like Android or similar, it has lots of power to process the dsp effects.
 
Dec 16, 2021 at 1:01 PM Post #47,540 of 48,562
Probably the sxfi chip is just an arm cpu, owing to today's standards, with relative processing power and multi core capabilities, if it doesn't run a complete operating system in top like Android or similar, it has lots of power to process the dsp effects.
There is no separate ARM (or other CPU) in any SXFI device as far as I know. All it's doing is playing preset convolution and equalization that you downloaded to it from a PC or phone. You could record the same convolution manually. (Which is why it is frustrating that they try to lock in SXFI playback to their devices).
 
Dec 16, 2021 at 2:51 PM Post #47,542 of 48,562
There is no separate ARM (or other CPU) in any SXFI device as far as I know. All it's doing is playing preset convolution and equalization that you downloaded to it from a PC or phone. You could record the same convolution manually. (Which is why it is frustrating that they try to lock in SXFI playback to their devices).
I think you're right.
I'm actually thinking of uploading the default and my preset to HeSuVi once Creative fixes the 7.1 issues.
 
Dec 16, 2021 at 5:11 PM Post #47,543 of 48,562
Let me know if you like it more than Gen2.

HI there. Already tried gen3, is just an insane surround, yes I like better than gen2, looks like I'm playing a game in a theater, I always play games with dolby digital 5.1 output from the console, gc7 just decodes it and sxfi gives this crazy surround, just a huge upgrade from the g6 for me.

By far this the best dac for gaming.
 
Dec 16, 2021 at 7:35 PM Post #47,544 of 48,562
I think you're right.
I'm actually thinking of uploading the default and my preset to HeSuVi once Creative fixes the 7.1 issues.
Let me know how that goes, I'd love to hear how you do it if you don't mind. I tried it with Audacity following the Sourceforge tutorial, but couldn't get it to record all the channels correctly for some reason. I'm sure its possible but I am unfamiliar with the tools so I just gave up. :frowning2:
 
Dec 17, 2021 at 2:03 AM Post #47,545 of 48,562
HI there. Already tried gen3, is just an insane surround, yes I like better than gen2, looks like I'm playing a game in a theater, I always play games with dolby digital 5.1 output from the console, gc7 just decodes it and sxfi gives this crazy surround, just a huge upgrade from the g6 for me.

By far this the best dac for gaming.
Awesome, really glad you like it:)
I saw no announcement of Gen3 really, which surprised me as it is an improvement to me too.

Did some more experimentation with SXFI today.
Adding 2-3 dB to 8-16kHz and reducing 2-4 kHz by 1-2 dB makes the sound more pleasant to my ears, on K371 (Unknown headphone profile) and K702 (K701 profile).
Adding 1-2 on the Bass knob makes K702 really bassy, but with very clean bass at the same time. Fantastic!

And the KSC75 sounds fine with SXFI with just 3-4 bass setting. No need to correct the upper mids and treble.
They are very bright to being with, with a lot of energy in the topmost octave so I guess it balances out the dip that SXFI puts there.
 
Dec 17, 2021 at 9:46 AM Post #47,546 of 48,562
Awesome, really glad you like it:)
I saw no announcement of Gen3 really, which surprised me as it is an improvement to me too.

Did some more experimentation with SXFI today.
Adding 2-3 dB to 8-16kHz and reducing 2-4 kHz by 1-2 dB makes the sound more pleasant to my ears, on K371 (Unknown headphone profile) and K702 (K701 profile).
Adding 1-2 on the Bass knob makes K702 really bassy, but with very clean bass at the same time. Fantastic!

And the KSC75 sounds fine with SXFI with just 3-4 bass setting. No need to correct the upper mids and treble.
They are very bright to being with, with a lot of energy in the topmost octave so I guess it balances out the dip that SXFI puts there.

I have never seen just a profound surround effect as this new gen3, really the cherry of the cake, for a huge sound upgrade for consoles and general gaming.
 
Dec 17, 2021 at 2:31 PM Post #47,547 of 48,562
As promised I provide some more Fr captures from SXFI.

Both images use 1/24th octave smoothing for readability.

First image is Gen3 vs Gen2.
Gen2 clearly has more energy above 10k, but a bigger dip at 9k, and that wasn't very nice to listen to.

Second image is the Unknown Headphone profile vs Unknown In-Ear.
The Unknown In-Ear may be a better fit for headphones with a lot of sub-bass, as it doesn't amplify it that much.
There is also less of a peak at 700 Hz and less overall treble, but that could be easily compensated for.

Looks like the IEM profile might be good for K371, must do some play tests now:)
 

Attachments

  • SXFI Gen3 vs Gen2.png
    SXFI Gen3 vs Gen2.png
    195.1 KB · Views: 0
  • SXFI UH vs UIEM.png
    SXFI UH vs UIEM.png
    203.3 KB · Views: 0
Dec 17, 2021 at 6:13 PM Post #47,548 of 48,562
I was wrong, I was not able to use K371 with Unknown In-Ear profile and compensate for the reduced high treble using the GC7 on-board equalizer. Instead, I preferred the Unknown Headphone profile with -3 dB at 31 Hz. Still a lot of sub-bass though.

If anyone else is interested in their FR graph for the GC7 with SXFI, you can basically connect Headphone out to Line In and use Room EQ Wizard. I used low gain and full volume and it didn't clip using the default settings in REW (measurement sweep at -12 dBFS). Keep in mind to disable line in monitoring, because it would then create a nasty feedback loop.

The line in itself is decent, has 110 dB SNR and flat frequency response (it's what it showed when I used RMAA with Shanling M0 which had even better SNR and flat FR) and up to about 2.5 V RMS for the line in.
 
Last edited:

Users who are viewing this thread

Back
Top