Loudspeaker crossover parts and sound quality. (Regarding Danny Richie / GR-Research's claims, of course.)

At the heart of most subjective vs. objective heated audiophile debates is the concern about the ease by which a person can dramatically exaggerate claims about sound quality without controlled listening tests, measurements, or any other higher level of evidence beyond personal testimony.

Without any meaningful repercussions, we can say almost anything, claim we experience almost anything, regardless of whether it's true, misperceived, or intentionally exaggerated for financial gain, status among peers, or other benefits. It is necessary therefore that audiophiles who read, watch, or listen to such comments develop the skills for critical thinking to assess such "testimonies"; acceptance without critical thought is simply blind faith, potentially leading us in directions we regret or might even be unhealthy. (See also the article on the types of "subjectivisms" in audiophilia.)

For this post, let's talk about a topic which has been around awhile but perhaps more notable over the last few years about passive speaker crossovers. In particular, the idea that at times quite expensive parts upgrades make desirable, even "huge" differences toward better sound quality.

Grab a seat, open up a beverage, sit by the fireplace if it's cold outside, and let's think about this stuff for a moment, dear audiophiles...

I. Background

To start, I think many of us are aware that this discussion is happening more in the YouTube space, much of it centered on the work of Danny Richie and his GR-Research channel where he has discussed a number of speakers from various brands, typically sent in by owners for assessment and potential modifications.

In the videos, the usual sequence of events is that he will critique the physical hardware, then show and critique some frequency responses, impedance curves and waterfall plots. Then he often will talk about modifications and particularly will design and sell crossovers that those interested might want to buy as DIY upgrade kits - they're not inexpensive, typically costing US$250+ for the parts.

Certainly, there is potential to do good work here, and good education to be had in those videos. Audiophiles should check out his channel to look inside loudspeakers of various prices, the drivers, his observations about workmanship, cabinet issues (eg. edge diffraction), and what he's able to achieve with crossover mods that will usually flatten out the frequency response. Also, tips like using his "No Rez" damping material for resonant enclosures might help.

Among other things we'll get to, there are some obvious issues that can "rub audiophiles the wrong way" in these videos for various reasons. For example, I'm not so enthusiastic about his huffing-and-puffing, head-shaking, eye-rolling emotional demeanor with breathy sighs when he talks and typically criticizes the products he receives. I guess some people just talk like that as part of their character, maybe it works for others, but his non-verbal manner I feel projects a level of arrogance which I believe is over the top and simply unnecessary. This is typically mixed with subjective claims around how dramatic things could sound if these speaker designs were better, often based not on the measurements he shows but just general beliefs about what parts "sound" like, or how the music is "sucked out" due to components, wires, or connectors that otherwise electrically are working fine.

About 4 months ago, Andrew Robinson in a video probably made a veiled stab at GR-Research, basically stating that he didn't care about crossover parts - "Who cares?" if the XO is made from a "Buffalo nickel", paperclip, etc... If the speaker sounds good, then let's move on and not worry about those parts.

It seems like this got GR-Research's attention with Danny posting this response:

More or less, I'm with Andrew on this one (obviously he's being facetious with the nickel, paperclip, etc.). As consumers, do we really need to open up every box we buy to examine the quality of the components inside? Does it matter whether every resistor in our hi-res DAC is of tight tolerance so long as it sounds good and objectively performs well? If high tolerances and more expensive parts are needed in a speaker, shouldn't we leave it to the company's professional designers to know when and where it's beneficial so as not to waste money on insignificant gain?

If a hi-fi product satisfies a customer's needs including sounding good for the price (based on comparison with other products) and works reliably for the life of the product, that should be good enough. For more objective technically oriented consumers, we can look for objective reviews, or even measure devices ourselves, adding an extra level of quantitative evaluation.

If consumers don't have good experiences and measurements look poor, this is feedback which reduces interest in the product, and the company doesn't make as much money. They can then do better next time with their newer generation as part of normal capitalistic competition. To dig into the crossovers, judge components, wiring, connectors, as standard operating procedure with such drama, is outside normal mainstream hobbyist behavior. This is clearly intended to either "feed" obsessional enthusiasts who hold certain extreme, speculative, even misleading beliefs which we'll get to, or to create attention through taunting companies for being "too cheap".

Danny Richie does this not in the spirit of sharing how to improve the speakers as a fellow DIY enthusiast because he does not publish his designs freely as other DIY'ers typically do on various forums; rather, he has something more expensive to sell you. There's nothing immoral, illegal, or wrong with doing this, just that it's not the most noble course of action if the intent is primarily said to educate and improve the community in these videos.

If he had published the schematics as well as sold his kits, this would be a demonstration of transparency - just as he forces transparency and critique of other speaker companies and their products that he disassembles! The DIY community can then critique, even further improve, his crossover designs. Manufacturers might learn a thing or two to update future products. Win-win for everyone including ultimately the consumer. He doesn't do this, he's not that kind of hero. 🤔 [For comparison, Nelson Pass is very open with his DIY designs.]

Regarding higher quality components, these days, some brands will advertise better parts used in their marketing (like in the computer motherboard space, or in audio with amplifier companies talking about the use of Nichicon caps, Sumida inductors, and such). From what I can tell, the reason for doing this is NOT to insist that these parts directly correlate to making a computer faster, or sound quality better (although might be implied), but mainly to increase confidence that the product is built to last. Notice that many of these parts are from Japan or Europe to distance the device from just using very inexpensive generic Chinese components at rock bottom prices.

When it comes to speakers, obviously depending on the drivers and cabinetry, the passive crossovers need to be designed optimally. Typically, we expect a good speaker to have reasonable frequency response (relatively flat on- and off-axis as per research on preferences, decent treble extension, acceptable bass), and not extremely low impedance (nice to keep 4Ω and above, eh?). The optimized crossover design will target specific frequency ranges and filter steepness correlating to the limits of the drivers. DIY guys have been tinkering with crossovers for decades now with the assistance of design and simulation software like VituixCAD, LspCAD, Xsim, SoundEasy, WinSpeakerz, among others. Even if we insist that there is a significant amount of art in perfecting these designs, it's mostly not rocket science. Arguably, the best, most accurate, crossovers in the 21st Century would be programmable DSP-based anyways, not these old-skool passive designs.

The idea that components themselves - like resistors, or capacitors, or inductors - rather than the design, is somehow capable of making a big difference to the sound is pushing us into the territory of other kinds of tweaks like magical-sounding wires/cables or opamp "rolling" where objective results and controlled blind listening tests either have not been done (probably because there's likely little to show for the effort) or has not shown meaningful change regardless of whether some people insist they hear remarkable differences. Let's be honest, this is the territory of the audio snake oil salesmen who feel comfortable with exaggerating claims and marking up prices for questionable value.

While it's fine that Danny is creating these upgrade kits based on his experience to give audiophiles options (with some caveats discussed below), I think it's reasonable to say that if the result sounds better, it's most likely because of the improved frequency response as in the videos with graphs like this - for example the Wilson Watt/Puppy Series 8 recently:

Before and after. Notice his use of the Audiomatica CLIO measurement system, basic gated measurements down to 200Hz at 1 meter, and low-resolution 1/3-octave smoothed plots.

The fact that he uses ostensibly higher quality, more expensive parts for his premium crossover kits is fine. But unless there is evidence of big sonic changes, do we need to poo-poo stock crossovers just because the look of the parts are "cheesy" if the measured speaker performance is still objectively good, and consumers are not complaining?

Still frame from his Wharfedale Linton video (Lintons cost ~US$1500):
 

II. Evidence for parts upgrades?

Now when it comes to evidence, as far as I am aware, there's no proof that otherwise normally performing capacitors change "sound" significantly, binding posts make a big difference nor that non-inductive resistors will have a huge effect. 

The only blind listening test summary I've come across is this one on the Audioholics forum (thanks for the link AD):

Capacitor Sound Blind Test (Solen metalized polypropylene vs. mylar vs. electrolytic)

Notice that the summary is from 2004. The idea of "expensive parts sound better" is not a new one among audiophiles even if Danny has perhaps popularized the idea in these recent days of YouTube. As you might have guessed, the blind test did not find any evidence that the more expensive Solen capacitor sounded better than the cheap electrolytic.

[In the world of audio there are expensive capacitors like the Sonicap Platinums, another one is the Jupiter brand with fans of the "sound" like this guy. Audio Note has been calling theirs the "world's finest" for a long time.]

For further objective test results, I suggest having a look at this excellent post on AudioScienceReview where a number of these ideas have been examined. For convenience, here are the links (amazing work ctrl!):

Capacitor upgrade in crossover - Is it audible?

Capacitor upgrade - part two (film vs. electrolytic)

Capacitor upgrade - part three (with analysis of GR/New Record Day video)

Replace resistor by low-inductance resistor - Is it audible?

Different Binding Posts - is it audible?

Audible difference in high-end capacitors? - ABX samples

Have a look through those threads. Have a listen to the ABX samples in the last post to hear for yourself.

Here's another interesting comment that models the effect from component mismatch in crossovers:

Effect of crossover components +/-10% Tolerance (great work & discussion by MAB)

[BTW the whole thread is interesting about speaker-pair testing and matching.]

The bottom line is that even with the statistically very rare unfortunate situation where two speakers with crossover components like the capacitor or inductor mismatches at extremes of 10%, generally we'd be looking at most 1dB variation. Crossover design is important, but the effects of parts even when very unluckily mismatched but within tolerance, isn't massive. The crossover parts themselves are not the limiting factor in loudspeaker performance (focus on the drivers, enclosure, and ultimately your room).

Guys like Danny like to say that there are qualitative sound differences that cannot be measured to discount objective testing when it suits him. In that case, it would be nice to see his level-controlled blind listening tests of a stock crossover vs. an electrically equivalent unit with much more expensive parts showing that listeners consistently preferred the expensive one. If this doesn't exist, why not after all these years?

[One needs to be careful since parts upgrades with the same value capacitor or inductor can change the part's resistance. So it might not be as simple as a direct drop-in replacement when aiming to exactly replicate the crossover electrically.]

By the way, a possibly very good test subject for listening could be the Andrew Jones-designed MoFi SourcePoint 8 speaker which Danny made a video on last year. The reason being that frequency response from his crossover was essentially untampered although impedance changed (looks like minimum impedance actually dropped from stock 8Ω with his new crossover):

He used these parts and the kit price is as shown: 

New, the SourcePoint 8 sells for $2000/pair without stands currently direct from MoFi. So the crossover kit is somewhere between 22-33% of the speakers' price without accounting for the time and other materials to build. Let's not forget the effect on the warranty and the potential to mess things up and end up with worse speaker performance (among other cautions below). Personally, I would never jeopardize existing warranty by doing these mods, and DIY work is unlikely to be good for the speaker's resale value.

Since there is little frequency response change, and probably insignificant variance due to impedance for most amps, it would be very interesting to run a volume-controlled blind test of the stock vs. upgraded SourcePoint 8. At this price point, the driver and enclosure should be at an excellent level to listen for significant changes. Evidence of listener preferences towards the GR-Research crossover would be nice to prove given the +25% in price, plus time, plus effort!

Danny's take is that the speakers now sound to be of "much higher quality detail levels", it's a "really worthwhile jump", "you need this upgrade", "very substantial". Blind listening identification should be an absolute cakewalk if all this is true, right?!

[Here's a video of a Wharfedale Diamond 10.1 comparison (older version of modern Diamond 12.1, typically $400-500). If you want to, grab the uncompressed 24/48 .wav file in the YouTube description, maybe cut up the segments into individual files and try ABX testing. There are amplitude differences with the crossover-upgraded recordings ($358 kit, almost price of stock speakers), louder by >0.25dB, along with recency bias where the upgraded sample always plays after stock speaker, likely would increase preference for the upgrade in casual listening. Make sure to apply volume normalization and randomize which sample you listen to first for a more fair comparison. Try with good headphones and amp.

It would not be difficult to take this audio sample from GR-Research/New Record Day and create an Internet Blind Listening Test out of that 24/48 file. See if you can "easily" identify the difference! Let me know if you're interested in doing this because supposedly, despite being a recording, we can "still hear a big difference... in clarity, and detail, and resolution... a lot different... because these (stock) parts smear the signals..." according to Danny.]

Believe...

III. Other specious ideas

Beyond the crossover and parts upgrades, if you watch the GR-Research videos, Danny makes many questionable claims about things like ferromagnetic parts (eg. steel nuts and binding posts) in the audio chain degrading sound significantly. He claims that one will hear dramatic sonic improvements using his "Electra Cable Tube Connectors" (US$59/set) as opposed to typical binding posts.

It's one thing to acknowledge differences of opinions when others disagree with you, but he seems to think that those who do not believe are "Flat-Earthers". This is a totally ridiculous charge since he has no evidence for his claims of superior sound quality (beyond differences due to crossover design). Flat-Earthers reject scientific evidence which is not what's happening here! The scientific evidence does not support his beliefs about the components, nor his belief in cable lifters using Tinkertoys, nor his ideas about cables ("Not Flat" T-shirt, bizarre rambling nothing-video 🤣), nor this video of cryogenically treated cables (apparently invariably preferred by his customers!). I might have missed other items of speculation, faith, and gross exaggeration over his 4-5 years of video posts.

While he tries to frame it that way, Danny is not fighting against unscientific "Flat-Earth" audiophiles in his videos. He's actually fighting against healthy skepticism that his claims are simply too good to be true based on rational experiences in life and scientific understanding of audiophiles living in a technological age! He seems intent to almost religiously "convert" everyone to believe his stories about these parts elevating sound quality. He claims to observe gains from cable lifters and cryogenic freezing techniques which for rational hobbyists belong not in scientific discussions, but as the speculations of sci-fi or fantasy enthusiasts.

In audiophilia, we see other forms of this mystical "woo" - such as quack medical benefits (also see here). His claims therefore are more akin to those of an extra-terrestrial "UFO Believer" (discussed before) or one who claims to have witnessed a spiritual event with nothing beyond words to convince others. Danny and these GR-Research parts-quality videos feed into obsessive-compulsive neuroticism which may not be particularly healthy (related to more extreme audiophilic psychological tendencies).

Needless to say, beware of salesmen. For Danny, there is an obvious profit and maybe reputational motive here. Enough talk: measure, demonstrate, prove.

IV. Caution on those kits

I think it's important to be cautious about these GR-Research crossover upgrade kits. It's not unusual that speaker brands will change parts like the drivers and crossover components over the years, whether small tweaks or larger product refreshes. I've been using Paradigm speakers for awhile and have seen/heard their changes with the versions. The Studio speakers for example have had 5 significant revisions in their lifetime and it's good that the company clearly indicates this. I can imagine especially smaller companies making changes on-the-fly without notice or supply chain issues might force a company to adjust (especially more recently through the pandemic).

There are situations where his DIY kit is significantly more expensive than the speaker he's modifying. For example, the US$200 or less per pair Sony SS-CS5 has a GR-Research kit that costs $300+! Before bringing out the credit card and rushing to replace the crossovers, given the very low price point for these speakers, are we even sure a pair of the Sonys bought in 2025 perform the same as earlier ones from 2015? Does Danny and the company check if the speakers might have changed over time?

I specifically bring these Sony speakers up because I have a pair here and noticed Zero Fidelity's review from a few years ago suggested there may have been a change in treble quality with newer units which is quite possible.

[My SS-CS5 has a production date of July 2024 so it might be interesting to measure this compared to ASR's 2019 production year.]

Another issue is that as far as I can tell (based on this Jay's Iyagi video), Danny designs these kits typically using loudspeakers sent in by customers. As such, he's not using "golden samples" representative of intended performance nor even necessarily of average performance units compared to the thousands that a company might have made of that product. Fundamentally, unofficial crossovers designed based on a single speaker (apparently he only gets 1 speaker sent in, not even the pair!) is I believe highly problematic. Even more so, the speaker could be sent in because the owner was dissatisfied with the sound from that unit. Without even 2 samples, how does GR Research know if one of them is failing, or the stereo matching is poor?!

This might be even more important when he's sent in older speakers like this video of the Klipsh Forté Version I (1985-1989!). Presumably those are 35+ year old drivers - well "broken in" I'm sure, but how do we know if they're not in any way "broken down" compared to the day they were born?

Let's have a look at the frequency response from the Klipsch Forté brochure from the mid-'80s compared to what Danny measured in his video:

Notice that difference between Klipsch vs. GR-Research measurement
at 2-4kHz. Ears sensitive here, could be quite significant.

Hmmm. All on-axis measurements, 1m meter away I believe. Is the GR-Research measurement done decades later performing like the same speaker that Klipsch produced or intended to produce? Did the other Forté I the person owns measure the same and would the crossover upgrade improve that frequency response in a similarly beneficial way?

[I see that Danny made a response video to address the Jay's Iyagi video where he expresses his belief that "we know" already all these upgrades in parts and connectors make a difference so there's no need to further confirm that his work is beneficial for each upgrade. I really don't believe "we know" is adequate.

He also tries to contrast "design work" claiming this can be done with one speaker, but AudioScienceReview's "reviewing" using 1 speaker is a no-no? His answer unfortunately deviates back to his usual simplistic unsubstantiated idea that cheap parts damage sound (ie. inaccurate harmonics and diminished natural decay) but does not address the fact that he simply does not bother, nor can, hear his crossovers reproducing the 3D soundstage. 

Sure, ASR's review with one speaker might be inaccurate due to getting a poor unit to test out, but Amir is not trying to sell anything, it's a review, a mostly measurement-based data point for consumers to consider. As for GR-Research, he's selling a kit for a few hundred bucks billed as a substantial improvement. What loudspeaker manufacturer would not have a stereo set-up to listen to the product in the intended fashion to confirm good "design work" before selling?

Anyhow, I find it all a bit dissonant to "call out" reviewers who at least are listening with stereo pairs and can comment on soundstage when he can't even do that in designing these kits regardless of how well-treated his room is.]

Another case in point might be my ferrofluid replacement article last week. Suppose I sent in something perhaps not as severe as my right speaker before ferrofluid replacement telling Danny that these speakers sounded a bit dull, a little "dark" with that frequency response dip, not knowing that the fix can be done for CDN$25 and a little work with the tweeter.

Without the left speaker to compare, we might imagine Danny making a video about the "cheesy" crossover (he did a video awhile back on the smaller Paradigm Signature S6 so I assume it's similar), how it sucks life out of the music, then huffs and puffs about how poor the frequency response is. Assuming I gave him the green light and offered to pay for a fix, he'll likely create something with premium parts that compensates for that dip with his usual procedure and software and he presumably will have an improved before-and-after frequency graph to show me of that one speaker. Of course it would be disastrous to sell such a design because the issue was not with the crossover. If I then changed the crossover of the other speaker to this, the sound still suffers with stereo pair mismatch and would not be good for anyone else buying the kit! With already >50 kits on sale, how do we know that some of these are not already designed to correct for the performance of a non-representative speaker?

As much as he claims to be "all about the measurements", he's obviously more in the subjective camp in many of his beliefs. Since he believes measurements can't tell him everything about how speakers sound (I don't necessarily disagree), if Danny doesn't even have two speakers to design his crossover kits with, how does he even know that his upgrades will result in good subjective spatial stereo reproduction?

Audiophile guys want "beautiful fantasy 3-dimensional images floating in space", right? Surely, he can't just base it on the highly smoothed, gated measured frequency response at 1 meter of one speaker like what Electrical Engineers might do? 🤣 Therefore, it seems like these kit designs have not really been fully tested to subjective "audiophile" standards, yet they're sold as upgrades attracting those who are most interested in the perfectionistic pursuit of audio quality - enough to rip out parts and put more money down!

IMO, he's trying to convince us that he is both:

1. A masterclass passive crossover designer who uses elementary measurements to create crossovers that surpass the efforts of basically every company he has come across. Using just one speaker.

2. An audiophile "golden ear" who vehemently claims to hear unmeasurable, likely insignificant when blind tested, audible benefits from expensive parts, connectors, wires, etc. Which he happens to also sell.

I'll leave it to you to decide whether you're convinced that he is an embodiment of both, one, or neither.

Showing off 3 crossovers claiming that the big one "doesn't degrade the signal", allowing full spatial cues, sounding "big", "deep", "layered", soundstage described as "this is like being there".
How do you ensure these characteristics when designing "upgrade" crossovers with just 1 speaker?

This has the cheapest parts, wires & connectors: "Measures exactly the same... dull as hell..."

Notice that he makes lots of "just so" unsophisticated correlations between sound quality and types of components (eg. electrolytic caps = audible smearing, really so simple?). How much of that is actually true and can be empirically measured or better yet blind tested? Probably almost none.

This part on hearing cable differences (any cable - power, analog or digital!) because of idealization of his room and gear is simply not believable and puts into question a lot of what else he "hears".

Given Danny/GR-Research's criticisms of so many other brands, it is only fair for his products to be tested also. I see this assembled X-LS Encore Kit measured quite well. They're going for US$360+ these days (without cabinets, add $275 for DIY MDF flatpack), and US$1235+ fully built ($50 for those "tube connectors"). Looks reasonable although there are many competitors at the $1200-$1700 price point of the fully assembled pair.

On the other hand, the GR-Research LGK 2.0 kit (US$200+) did not fare well at all in this ASR review. A completed pair of these single "wide-band" (not true full range) driver speakers costs US$1124+. Notice the severe distortion in the measurements at higher levels, easily audible through a YouTube video! That's simply very bad. (Plus the air coil inductor apparently didn't look perfect in the kit.)

[Here's GR-Research's video on the LGK 2.0 with admission that these need low-frequency filtering. Since when did fully-finished >US$1100 desktop speakers, referred to as "giant killers" no less, not handle genres with deep bass, or require that customers EQ out low frequencies to avoid severe distortions?

To rectify this, I see that the 3" driver is paired with a small 5" woofer for the LGK 2.1 ("Whats (sic) The Best Speaker Ever?"). 

He's also using this driver with the GR-Research/New Record Day's upcoming line of Aria speakers paired with his 5¼" M130 woofer. Notice the off-centered front placement, and centered rear dipole/quasi-omni effect I presume, interesting. The claim is excellent imaging and vocals are "un-be-lievable". However, they are messing around with the Caelin Gabriel/Shunyata unmeasurable woo-woo stuff like the HARP module, and KPIP "technology" (see this). 🫣 I wonder how much that stuff adds to the price. Will be interesting to see what the reviewers say and measurements look like when released.

BTW, speaking of distortion from that LGK 2.0, notice that GR-Research's videos and website lack distortion measurements. I wonder when he designs crossovers for speakers from other brands whether there is a change to distortion before and after, and also changes in speaker sensitivity.]

There are other objective reviews of GR-Research stuff like here and here (I believe Danny designed the crossovers for AV123 with its checkered past).

I hope that all the DIY guys buying crossover kits already know how to measure speakers and can confirm that their sample at home correlates closely with what Danny's showing in his videos. Only if there's close correlation in frequency response should one consider buying the kits. Furthermore, after installing the kit crossovers, make sure to re-measure to confirm your DIY abilities and that there indeed has been an improvement. It's a lot of work, and at the price of US$250 on the lower end to sometimes more than US$1000 higher up, one would hate to end up with a worse performing loudspeaker! I'm not even sure what one would do if the outcome turns out badly. 😢

If this still sounds good to you, great! And good luck. 🫡

I hope you're enjoying the music, friends.

--------------------

New music from FKA twigs - "Childlike Things" from EUSEXUA (2025) - as often the case, the multichannel/Atmos mix sounds quite interesting:

Old music from John 'Cougar' Mellencamp - "Small Town" (Acoustic Version) from Scarecrow (1985):

There's a nice multichannel remix available for "Small Town" (Acoustic) which enhanced vocal/instrument separation. However, I wish they used the center channel to reduce the dependence on phantom-center vocal imaging, expanding to the front left/right channels as needed for image width. I've heard chatter that the center channel is often not used purposely because there's fear that people will rip the artist's a cappella vocals. Not sure how much of that is true.

It would be a shame if out of concerns like this, multichannel potential becomes incompletely utilized!