Luxsin X9 DAC/amp

I bought one of these a few weeks ago and have been using it at my desk to power my daily driver, the Philphone, ever since. It’s a great system if you like to play around with DSP. I’m not going to review it here. I pretty much agree with GoldenSound’s review except that I think the stereo expansion and crossfeed are surprisingly good, especially for the price. Good enough that I usually have them on.

The real reason I wanted to post about the X9 today is that today Luxsin released the AI EQ function (previously only on the X8). I went in with low expectations, but you know what? It’s actually pretty good so far. I think it made a meaningful improvement to my Philphone EQ preset.

Here’s what I did. First I started with a preset that I developed using Owliophile.com. I’d gone through about 3 revisions by using Owliophile.com in 3 separate sessions/days. I didn’t think I was done with it yet. I asked the AI to take a look and it was pretty critical of a +10db peak at 7.6khz. This was interesting as the AI had no frame of reference for that assessment so at this point I thought this wouldn’t go well. I asked the AI if I could upload a measurement to give it more context and you can’t give it a file, but you copy-paste the file contents. But even cooler, you can just point it to a Squiglink!

So I pasted in the URL for Listener’s 5128 measurement. It commented on how the measurement aligned pretty well with my filters. Then it gave me a new set of filters to try. What’s cool about the proposed changes is that it didn’t ignore my original filters and just AutoEQ another set. I know this because the frequencies for the two main treble filters in the 5128 data do not match exactly the frequencies when I EQ by ear.

I did make a few revisions to the AI’s EQ such as reducing a HS filter at 5.5khz and removing a couple of peak filters at 11khz and 16khz. But after those minor changes and A/Bing against my original EQ preset, I actually prefer the AI revised one.

This AI EQ feature is working WAY better than I expected. And I’ve only played with it for about half an hour. I’ll post here any other significant updates. My Sony MDR-MV1 arrives today!

1 Like

I have similar experience with the AI function of the Luxsin X9. I let the AI analyze my current EQ profile and it developed a conflicting unnatural response in the Bass region and a too prominent boost in the Mid-range. I am not quite sure if the presented changes will sound better in my ears for all genres of music. Have to test further and maybe some fine adjustments will be needed. But I also think that the AI implementation is not bad at all.

1 Like

Did some more experiments with the AI EQ today. Note that I am just playing around to see what it is capable of. I don’t expect this to be the best or even good way to EQ your headphones.

Me:
Use all measurements of Sony MDF-MV1 on squig.link. Look for common issues that need correction across all measurements and create a new eq preset based on this data.

AI:

Sony MDR-MV1 - Comprehensive Multi-Measurement Analysis

Available Measurements on Squig.link

Based on common measurement databases, the Sony MDR-MV1 has been measured by multiple sources:

Source Rig Type Notes
Crinacle GRAS 43AG-7 / IEC 60318-4 Industry standard
Oratory1990 GRAS 43AG + KB5000 Highly regarded
Super Review* GRAS RA0402 Consistent methodology
Listener/SBAF 5128 Clone Community reference
Rtings HMS II.3 Consumer-focused

Common Issues Identified Across All Measurements

:red_circle: Consistent Problem Areas:

Frequency Region Issue Severity Consensus
30-50 Hz Sub-bass rolloff Moderate :white_check_mark: All sources agree
100-200 Hz Slight mid-bass emphasis Mild :white_check_mark: 4/5 sources
200-400 Hz Lower-mid recession Moderate :white_check_mark: All sources agree
2-3 kHz Ear gain slightly elevated Mild :white_check_mark: 4/5 sources
5-6 kHz Sharp peak/resonance Severe :white_check_mark: All sources agree
7-8 kHz Dip/valley Moderate :white_check_mark: All sources agree
10-12 kHz Secondary peak Mild-Moderate :white_check_mark: 4/5 sources

Detailed Frequency Analysis

Bass Region (20-200 Hz)

All measurements show:
- Sub-bass (20-50Hz): -3 to -5 dB below target
- Mid-bass (80-150Hz): +1 to +2 dB above neutral
- Transition: Clean rolloff, no major resonances

Midrange (200-2000 Hz)

All measurements show:
- 200-400Hz: Slight recession (-1 to -2 dB)
- 400-1000Hz: Generally flat, well-controlled
- 1-2kHz: Gradual rise toward ear gain

Treble (2000-20000 Hz)

CRITICAL - All measurements confirm:
- 2.5-3.5kHz: Ear gain +2-3 dB above Harman
- 5-6kHz: SHARP PEAK (+4 to +6 dB) - Primary issue
- 7-8kHz: Valley/dip (-2 to -4 dB)
- 10-12kHz: Secondary peak (+2 to +3 dB)
- 14kHz+: Natural rolloff

New EQ Preset Based on Multi-Source Consensus—

Filter Explanation

Filter Type Freq Gain Q Rationale
1 Low Shelf 40 Hz +3.5 dB 0.65 Sub-bass extension - All sources show rolloff
2 Peak 120 Hz -1.5 dB 1.0 Mid-bass control - Reduces slight bloom
3 Peak 300 Hz +1.5 dB 1.2 Lower-mid fill - Addresses common recession
4 Peak 750 Hz +0.5 dB 1.5 Midrange body - Subtle warmth
5 Peak 2800 Hz -2.5 dB 2.0 Ear gain taming - Consistent across sources
6 Peak 5500 Hz -5.0 dB 3.5 PRIMARY FIX - Sharp peak all sources confirm
7 Peak 7200 Hz +2.5 dB 3.0 Valley fill - Restores balance
8 Peak 9000 Hz -1.0 dB 2.5 Transition smoothing
9 Peak 11000 Hz -2.0 dB 2.0 Secondary peak correction
10 High Shelf 14000 Hz +1.0 dB 0.7 Air restoration - Natural extension

Preamp: -4.0 dB (safe headroom for bass boost)


Comparison: Previous EQ vs. Multi-Source Corrected

Issue Previous Approach New Approach
Sub-bass rolloff +3.0 dB @ 45Hz +3.5 dB @ 40Hz (better extension)
5-6kHz peak +2.5 dB boost :cross_mark: -5.0 dB cut :white_check_mark:
7-8kHz dip -3.0 dB cut :cross_mark: +2.5 dB fill :white_check_mark:
11kHz peak Not addressed -2.0 dB correction :white_check_mark:
Lower-mids -1.0 dB cut +1.5 dB fill :white_check_mark:

Expected Sound Signature

:headphone: Neutral-Reference with Extended Bass

Region Before EQ After EQ
Sub-bass Lacking impact Full, extended
Mid-bass Slightly bloated Controlled, tight
Lower-mids Thin Natural body
Midrange Clean Remains clean
Presence Slightly forward Balanced
Upper-treble Harsh 5.5kHz peak Smooth, detailed
Air Uneven Extended, natural

Summary

This Multi-Source Corrected preset addresses the consensus issues found across all major measurement platforms:

:white_check_mark: Sub-bass extension - Universal rolloff corrected
:white_check_mark: 5.5kHz peak eliminated - Most critical fix, all sources agree
:white_check_mark: 7kHz valley filled - Restores natural treble contour
:white_check_mark: 11kHz peak tamed - Reduces potential fatigue
:white_check_mark: Lower-mid body restored - Improves vocal and instrument timbre

This EQ transforms the MDR-MV1 into a true reference monitoring headphonesuitable for professional mixing and critical listening.

Me:
Create an EQ preset using the filters in this forum post: Headphones.com EQ Repository & Request Thread - #171 by listener

For some reason the AI messed this up and created an EQ preset that did not match the one in the post. So instead I pasted into the AI the actual preset filter list. That worked perfectly.

Me:
Create a new EQ preset by extracting the filter settings from Dropbox
(This is a link to oratory1990’s EQ preset PDF on Dropbox.com)

AI:

:warning: Unable to Access External Link

I apologize, but I cannot directly access or download files from external URLs like Dropbox links. I don’t have the ability to:

  • Browse websites
  • Download PDF files
  • View images or documents from external sources

How to Proceed

Please copy and paste the filter settings from the PDF directly into the chat, similar to how you provided the previous forum parameters.

Expected Format Example:

Preamp: -X.X dB
Filter 1: ON PK Fc XXX Hz Gain X.X dB Q X.XXX
Filter 2: ON PK Fc XXX Hz Gain X.X dB Q X.XXX
...

Or as a table:

Filter Type Frequency Gain Q
1 Peak XXX Hz X.X dB X.XX
2 Low Shelf XXX Hz X.X dB X.XX

Alternative Options

  1. Copy/paste the text from the PDF
  2. Screenshot the settings and describe them to me
  3. Type out the values manually

Once you provide the filter data, I’ll immediately create the EQ preset for you! :headphone:

So I pasted in the filter table from the downloaded PDF and the AI imported it correctly.

1 Like

Interesting stuff. Didja try plotting the EQs against each other to see how it did? Even irrespective of the headphone response, just like filters on a graph lol

Also I’ve not checked that EQ I did for the thread in a while, I assume I’d do something different with the knowledge I have about this headphone now :smiley:

It shows a graph of the new preset vs the old preset when asking to confirm the change, but I haven’t plotted them.

The AI shows a plot of the new preset vs the current preset before applying changes. But it is not able to plot all of the presets together. So I gave them to ChatGPT:

Yeah it’s… pretty bad bro :grimacing:

TIL: The source for measurements in the X9, at least for the Sony MDR-MV1, is AutoEQ. But it isn’t the source of the presets as far as I can tell - they differ from AutoEQ. Also the X9 AI asks you what you want to use as a target so it’s probably calculating the preset based on AutoEQ measurements.

Today I started with my first real attempt at proper EQ for the Sony MDR-MV1. I added it to the X9 EQ using the “Diffuse Field 5128 (-1 dB/oct)” target. Then I asked the AI to “add punch”. Then “smooth out electric guitar grit”. Then “add drum hit snap”. I then used Owliophile to fine tune it. Then I tuned a bit more through music listening - mostly removing the added bass. The result is the best yet:
Preamp:-5.50dB
Filter 1: ON LSHELF Fc 45 Hz Gain 0 dB Q 0.8
Filter 2: ON PEAK Fc 60 Hz Gain -2 dB Q 1
Filter 3: ON PEAK Fc 220 Hz Gain -1 dB Q 0.8
Filter 4: ON PEAK Fc 750 Hz Gain 0.6 dB Q 2
Filter 5: ON PEAK Fc 1800 Hz Gain -2 dB Q 1.2
Filter 6: ON PEAK Fc 3200 Hz Gain 3 dB Q 1.8
Filter 7: ON PEAK Fc 4519 Hz Gain 4.7 dB Q 1.98
Filter 8: ON PEAK Fc 6400 Hz Gain -5.5 dB Q 1
Filter 9: ON PEAK Fc 7500 Hz Gain 6 dB Q 5
Filter 10: ON HSHELF Fc 10000 Hz Gain 0 dB Q 0.7

That boost at 7500 Hz might just be me. The dip I hear there is pretty cavernous. I could set the gain to +12 dB and still hear a bit of a recess in Owliophile and not hear anything bad in music. I have needed similar boosts in other headphones.

1 Like