In my previous post, I discussed target curves and gave examples of curves you might want to consider when implementing DSP correction in your sound room.
While the use of full-range target curves is not uncommon, as discussed, there are those who advocate against doing so and would insist that only partial corrections in the bass frequencies up to the Modal+Transition Zones should be applied. There are of course some good reasons for this line of thinking which will be addressed.
As you can see in the comments last time, I had a long discussion with Joe Mariano (aka Joe N Tell) of Magic Beans Audio about his product called True Target. I continued my discussions with Joe via E-mail and he provided access for me to the Pro version which allows export of the curves to Room EQ Wizard in order to have a deeper look under the hood at how the system works.
The picture above was taken right before one of my measurement sessions with Magic Beans highlighting some "tools of the trade" used to perform the tuning.
Let's talk about the Magic Beans* True Target app.
[Standard version regularly priced at US$249 should suffice for the vast majority of users, Pro version US$399 allows up to 128 channels and unlimited daily calibrations.]
* As audiophiles, we know of course that there is no supernatural magic necessary to understand engineered hardware or software. A person or company claiming otherwise or unable to show meaningful effect with their product would obviously be engaging in selling snake oil, should be avoided, and it's only fair to warn others.
In this case, the company is simply using the "magic beans" name to generate interest through colorful humor. No need to let the name be a stumbling block for those of us phobic of audio scams in the 21st Century. 😉
I. What is the Magic Beans True Target app doing?!
As I watched the YouTube videos, read reviews, and checked out the online discussions, I remained confused about this product. It wasn't until I started to have that long discussion with Joe and then tangibly examining the output from the software that things started to make sense.
The suggestion against using full-spectrum EQ/DSP to a "generic" target curve revolves around the fact that beyond bass modes in small domestic listening rooms, what we hear at the main listening position (MLP) is always a combination of the on-axis/direct sound from speakers plus the off-axis/indirect radiations that end up being reflected around the room.
Consider if we had a perfect loudspeaker, we would find that this hypothetical ideal transducer would have what's called a smooth "directivity index" (DI), basically the ratio of direct power to omnidirectional (or subset of full 360°) power across the audible range at the listening position which can be plotted across the frequency spectrum. For a front-radiating speaker, the ratio of front on-axis sound compared to the omnidirectional power would trend towards an upward slope as frequencies rise, wavelengths shrink, and the driver behaves like a directional radiator, concentrating energy forward. The more the speaker "beams" forward at higher frequencies, the more acute that directivity slope becomes.
For such an ideal loudspeaker, even if there is beaming at higher frequencies, because the DI graph is smooth, perhaps linearly tilted in appearance, we can say that as a whole, it's still spectrally balanced. Applying an EQ to a spectrally balanced loudspeaker will generally predictably color the sound in the room as intended.
Unfortunately, loudspeakers are not perfect and this means that we will not see those nice, smooth DI lines as above in the majority of products. Rather, "directivity errors" show up in the radiation patterns at different points in the spectrum. The dominant reason for directivity issues are irregularities at loudspeaker crossover points where the outgoing (larger) driver is operating at the higher end of its useful bandwidth and starting to narrow as it's handing over to a smaller driver with more omnidirectional radiation pattern. Beyond crossover regions, there are comb-filtering directivity errors (related to driver spacing), baffle diffractions, cabinet resonances, among others captured by thorough speaker measurements. We need to refrain from using EQ on these spectrally unbalanced frequencies.
We can decide that above the bass zone it's too complex and just leave it alone. While there's nothing wrong with just correcting for the bass frequencies, this can however miss genuinely correctable direction-independent midrange/treble anomalies.
So then, how would we construct an 'optimal' target curve to feed into a DSP system like Dirac Live that can handle full-frequency correction? This, friends, is where we get into the rationale behind the Magic Beans True Target app.
Acoustic research (Floyd Toole - 1980's NRC Canada, and with Sean Olive - Harman) showed that listeners preferred the sound of speakers with anechoically flat or gently tilted on-axis frequency responses and smooth directivity with no dips, peaks, and abrupt discontinuities. The 'Spinorama' (ANSI/CEA-2034) was designed to measure speakers more completely, including capturing the off-axis power radiation and directivity index as part of those calculations when we look at these graphs (see here).
As hobbyists, unless a company were to publish detailed data or we can find independent results, we are unlikely to have access to measurements for the on-axis frequency response nor directivity curve for speakers we own.
However, we can capture a reasonable on-axis response by taking a nearfield measurement (NFD) making sure the mic is far enough away for multi-driver integration, but close enough to avoid all major room reflections. We can also get an idea of the room transfer function if we compare this NFD with a separate MLP measurement. This is what Magic Beans does.
Let's illustrate the concept like this:
From this, I trust you can already imagine how we might want to create an optimized target curve that is a hybrid which smooths out the bass frequencies (as per the grey 1/2-octave smooth trend line), then in the midrange and treble regions, apply EQ to flatten the speaker's on-axis response so it mirrors our ideal loudspeaker, but only where there are no major quasi-DI anomalies.
[It's worth thinking about what happens to the sound at the MLP if we force correction into directivity dips and peaks.
From what I've read and understood, the worst situation is to boost the on-axis response into a DI dip because this will "correct" the direct sound but the reflected/reverberant field will not follow. It is this reflected energy that gives instruments and voices a sense of natural "body", "presence" in the room. If we push that direct sound with EQ, even though we might experience extra detail because we can hear the frequencies clearer, without the "body", the subjective experience would be one of a "thin", "forward", unnatural sound that some might also call "overly hi-fi" - in the bad connotation way. 😐
Alternatively, if we boost into a DI peak, we're accentuating an overly-strong reflected energy zone. The result would be excess "warmth", "bloat", "thickness", "boom", or "muddy" sound.
Obviously we do not want either of these things to happen, especially at the frequency zones where our hearing is the most sensitive - roughly 1-5kHz, peaking around 3-4kHz. These frequencies are also often where multi-way loudspeakers crossover from the midrange driver to tweeter.]
Putting all that together through discussions with Joe, here's a visual aid to roughly model what the software is doing for a non-ideal speaker in a non-ideal room:
The "True Target" as Magic Beans calls it (plotted mostly in white) is generated with a smoothed bass response as you can see from 20Hz to 500Hz. While this loudspeaker rolls off around 30-40Hz, we can imagine that if the system has subs, it would extend down towards 20Hz as per the grey segment.
Above where the software detects the end of the bass/transition zone (say 500Hz in this example), it will try to apply gentle EQ to flatten the NFD frequency response - but only in zones where the quasi-DI is reasonably smooth. This is why Joe calls it "speaker response correction" rather than "room correction" - the room response is the consequence. We can see regions of directivity error in the pinkish-colored zones, identified as "low authority" regions where less or zero EQ should be applied. The effect of the strategic application of EQ is reflected in the yellow dotted "DI-conditioned NFD" curve where these zones are allowed to flow with the measured NFD without attempting to strongly correct them.
And so, once we have that "DI-conditioned NFD" curve, we can create the "True Target" which is a combination of the smoothed bass response spliced with a calculated MLP÷DI-conditioned NFD curve - the final white tracing with grey sub-bass extension.
We could create such an optimized hybrid bass correction plus speaker midrange-treble flat target curve by hand but it would take a lot of time manually correcting, cutting, and splicing the frequency responses together; especially laborious if you have a multichannel system! The efficient automation of this process is the "magic" in the True Target app.
II. Magic Beans, in use.
For reference, already there are many videos on YouTube about how the app is used practically, so I will not repeat much of that. Check out these official videos from Joe Mariano if you're not already familiar:
For those who prefer the written word, here's Steven Stone's review of MB in TAS from March 2026 targeted at "old-skool" mere 2-channel audiophiles. 😉
For my testing, I used the Windows version of the app (currently v0.4.0610) which ran without any issues across a few computers here at home. Here are a few screenshots of the app in use (across platforms -Windows, Mac, Android, iOS - it looks the same):
The app guides you through the steps in preparation for the measurements. Notice that it's able to see my miniDSP UMIK-2 USB microphone and can grab the calibration data online when I entered the serial number. Alternatively, you can load a local calibration file (use the 0° on-axis one). To be efficient, make sure to heed the "Getting Started" and "Pre-Calibration Checklist" items.
It's very important to make sure your AVR/processor/DAC/amp isn't applying any EQ or other DSP including basic bass management. Make sure all speakers are seen as "large" or "full band" in your home theater crossover settings like this before getting started.
![]() |
| Integra DRX-8.4: speakers all full band and Dirac Live off before running measurements. |
Also make sure the playback is "Direct":
![]() |
| "Direct" playback of channels, not using things like DSP upmixing such as Auro3D or DSU. |
The recommended test signal for the app is 16k-length periodic pink noise (measure ~75-80dB SPL). You can get this from various sources including creating your own in REW for a simple stereo 2-channel set-up. If you have a multichannel/Atmos system, you can grab the Spatial Audio Calibration Toolkit BluRay and there's also a complementary periodic pink noise pack available for Magic Beans customers.
[For my testing, I created my own 1-minute long TrueHD-Atmos, 16/48 lossless, 16k pink noise files. They were bit-streamed off my fanless MiniPC to receiver for each speaker; similar to how I put together the Atmos Walkthrough with DaVinci Resolve Studio.
If you're wondering, no, I was not able to have the same computer run the audio output and Magic Beans at the same time. If MB is unable to secure audio out, the microphone also does not capture.]
The app will show a video of how to perform the Moving Microphone Measurement (MMM) for NFD and MLP. Notice a range of speaker layouts is available from 2.0 to 9.1.6 and 128 channels in the Pro version. My main system is 5.1.4; single LFE, the Integra receiver manages the dual subs.
You then one-by-one capture the NFD and MLP curves for each speaker. Once I got used to the process after a couple of practice trials, it took me less than 45 minutes to measure all 10 channels carefully. After that, voilà, the app will calculate your "True Target".
Here's the final result of my system, single curve averaged among the speakers:
Notice that my final curve looks a lot like the Harman curve doesn't it? That's to be expected since the general shape in a hi-fidelity listening room should look like what was found in some of the Harman/NRC papers using good speakers in reference rooms. However there are little wiggles in the curve customized for my speakers and my room which we'll look at in more detail below.
[Regarding that bass elevation to +6dB around 20Hz; while I think this sounds good especially in movies, for hi-fi music, IMO that's for bass-heads and untrained listeners. I'd want to pull that down a couple dB's for my audiophile ears. 😏]
Magic Beans is a 3rd party custom target curve generator - think of it like a "pre-processor". While it can export graphic and parametric EQ settings and convolution WAV files (can be used in Roon, JRiver, Audirvāna, etc.) and will sound great, it's not a full DSP correction package like say Dirac Live, Audyssey MultEQ-X, or Trinnov with advanced features like time-domain correction and complex crossover settings especially for multi-sub set-ups.
Notice that the app allows for different types of exports; Per-channel where each speaker/channel is calculated individually, Unified Sound Field where the bass frequencies are the same for each speaker, and Global which exports a single averaged frequency response for all speakers/channels.
For my set-up, I went with the default Unified Sound Field, exported as "DLBC (Separate Height Sets)" with separate front and back height channels, then went into Dirac Live Bass Control to create the filters (see DLBC write-up).
III. So is there evidence this works?
Before talking about measured outcomes that result in numbers and visually in graphs, I was challenged to think about what is a "good looking" outcome in light of the above discussions in Section I.
Sensory perception through human ears and brain is much more complex than what a single microphone captures at one spot. So while we can measure something like frequency response and recognize that this will correlate with what is heard, and we can compare that result to an ideal, in the real world, complexities like the in-room directivity issues suggest that what is "best" for our speakers, room, and sitting position may not - and should not - look like idealized flat lines. Instead, the best calibration we should be aiming for is one that synergizes with the best that our speakers and room can offer, and it's unlikely that any of this will "look" ideal! Ultimately, the science and measurements are to serve in optimizing subjective experience.
When we're talking about the whole sound system in a room, we need to recognize that this is more nuanced than the comparatively straight forward measurements we take when objectively reviewing isolated electronics like DACs, or amplifiers (or God-forbid worry about cables and even more meaningless minutiae tweaks) where "accuracy" and "transparency" are much easier to understand and achieve.
Having said this, let's address a few questions I had about measurement reliability and to demonstrate change with some tests as I got to know the Magic Beans True Target app.
1. How consistent are the Moving Microphone Measurements (MMM) for NFD?
Though rarely spoken of in this blog, the MMM technique is not new and I discussed it a little bit back in 2020. It's a way to quickly capture spatially averaged frequency response but since the microphone is moving around, not for capturing temporal data accurately.
I'm not as concerned about the MLP MMM measurement as I am about consistency and speaker integration when doing the NFD with my large Paradigm Signature S8 speakers spanning ~30" between tweeter and furthest woofer. So, let's do 4 measurements using the Magic Beans procedure based on their instruction video. I'll start around 25" from the speaker at tweeter axis and move back to 35" slowly over 10 seconds, mic pointed at the tweeter, using my meter stick for approximate guidance of distance.
Not bad; even without trying hard, 3 of 4 measurements were highly concordant across the whole spectrum, achieving ~1dB variance above 500Hz. The one outlier (light blue trace) showed mostly variation <100Hz.
With a little care, it should not be difficult to achieve good accuracy with the MMM method using the MB app.
2. Does MB flatten the NFD measurement of the loudspeakers as intended and how is it compared to standard Dirac Live Bass Control with Dolby Atmos Music Target Curve?
![]() |
| Curves normalized at 200Hz point. While the True Target for these front speakers is nuanced, notice it's still within the +/-3dB tolerance for the Dolby Atmos Music Curve. |
3. How about comparing Magic Beans True Target to Dirac Live Bass Control with Dolby Atmos Music Target at the farfield MLP?
Alright, now that we've had a look at the nearfield, let's look at the effects at the MLP which is my seating "sweet spot" about 9-10 feet away:
4. What about tonal consistency between different speakers if True Target changes the curves for each speaker (or pair of speakers as in Dirac Live Bass Control)?
For this, let me set FDW to 10 cycles for finer frequency resolution while still focused on mainly the direct sound (purely direct sound above 2.8kHz) and examine the variance in frequency response at the MLP. Let me capture my 5 bed channels; front speakers are a pair of Paradigm Signature S8 v.3 towers, center is the Signature C3 v.3, and the rears are Studio 80 v.3 towers:
IV. So how does the "True Target" sound in the system?
Above, you see the final "Unified" True Target curves created for my system as I settled in for some evening listening sessions. Since I am using Dirac Live Bass Control, the left and right channels would actually be averaged and folded down into a pair for fronts, surrounds, front heights, and rear heights. The only significant customization I made was to tweak the LFE to roll off by 120Hz at the upper end with a peak of +3dB around 30Hz, down to 0db by 20Hz, and very steep infra-bass roll-off with something like -25dB into 10Hz - I don't like too much bass and infrasonic rumbling.
Beyond the graphs and looking for differences between curves, ultimately it is about how this sounds subjectively and whether we enjoy it.
I have to admit, the True Target sounded great.
As seen in some of the Section III graphs, the tonality with the Magic Beans True Target is brighter than the Dolby Atmos Music Target or "General Audiophile" curves discussed previously, and on the whole perceived as having more "air" when A/B compared. To make sure it doesn't sound overly bright though, I had a listen to some early digital recordings like Tower of Power's "Squib Cakes" off Direct (1981, DR13). This was the only Sheffield Labs direct digital recording released, captured with the old Mitsubishi X-80 running at 16-bit/50.4kHz. The sound of this CD has always been a bit on the sonically crisp side for my taste, and with the MB True Target it still sounded reasonable without making my ears "bleed".
One of the first familiar tracks I tried was the classic Muddy Waters "My Home is in the Delta" (Analogue Productions SACD rip of Folk Singer, 1964, DR16) which I find can sound overly forward in some systems I've heard. This track sounded really good using these MB settings with a realistic sense of spatial "presence" as Mr. Waters sang and fingerpicked his guitar in my room. The guitar in particular sounded realistic, and not overly harsh.
I'm using the Sony SS-CS5 bookshelf variants for my height channels and noticed that Magic Beans has accentuated the upper frequencies above 5kHz on those. This had an interesting impact with enhancing the impression of height in a number of tracks. For example, one of the demos I like to play for visitors is Yosi Horikawa's "Bubbles" using Dolby Surround Upmixer (DSU) to show off the ability for a multichannel system to extract an expanded soundstage from a 2-channel source in a way that's much more "solid" than aligning your head between 2 speakers. In particular, I thought the Magic Beans target curve made a positive impact on placement of those bouncing ping pong balls not just in the flat ear-level surround plane but also vertically at multiple levels. To be clear, I was able to hear this height effect as well with the generic Dolby Atmos Music Target, but not as well separated when I compared by flipping back and forth between the Dirac settings.
Bass depth and spatial envelopment remains excellent - as expected given the frequency response of the system. Realistic bass pressure is easily felt in the room from pipe organ tracks like Anna Lapwood's "Chevaliers de Sangreal" on Midnight Sessions at the Royal Albert Hall (2023); a very different experience than the low-bass thump of R&B or the dramatic bass-drops of synthetic electronica.
While organ music can sound good in stereo, IMO they really shine in multichannel! A reader recently suggested that I have a listen to Cameron Carpenter's album Revolutionary (Telarc SACD, 2008, 5.1 DR13). The reimagined "Bach: 'Evolutionary' Toccata and Fugue in D Minor, BWV 565" track sounds amazing with Auro-3D "Auro-Matic" upmixing with the MB True Target, lifting the sound up into the height channels like in a real cathedral. Although the 2-channel mix sounds great as well, this is a nice example of how multichannel can utilize the extra speakers to expand the 3D soundstage and better render the nuances of such complex sounds instead of squishing everything into just two front channels!
[Fun fact, Carpenter's album was recorded at Trinity Church Wall Street not with an analogue pipe organ, but using a Marshall & Ogletree Virtual Pipe Organ - the OPUS 1.]
The folks at SDE have done it again with another excellent multichannel mix, this time of Bryan Ferry's Boys And Girls (2025 release, 1985) in full TrueHD-Atmos by Bob Clearmountain who did the original stereo mix all those years ago. Simply excellent. One interesting thing I noticed was what sounded like a more solid localization of content at the rear phantom center with the MB target. This was also noticeable with the multichannel version of Roger Waters' "Too Much Rope" off Amused To Death (5.1 24/96, 2015 BluRay mix, 1992) as the horse-drawn carriage/sleigh with bells passes behind the listener.
I found nothing to complain about with vocals when using the True Target, like George Michael's "Brother Can You Spare A Dime?" off Songs From The Last Century (1999, multichannel/Atmos mix also available) or ladies like Alison Krauss on "When You Say Nothing At All" from Now That I Found You: A Collection (1995, love the DR12 dynamics). Excellent stable front-and-center vocal placements, clarity without shrillness or sibilance, and separation of the voice from the instrumentation.
V. Summary
I can imagine that a portion of the technical stuff in this post probably went above the heads of some audiophile readers. This is perhaps expected since what we're looking at here is intended to be an extra step beyond typical DSP correction; requiring a level of investment for perfectionist audiophiles and home theater enthusiasts looking for even more from their sound systems. In the audiophile world, I think it remains all too rare that even "high-end" systems have not embraced the clear benefits of DSP correction.
The Magic Beans True Target app obviously does not possess some kind of metaphysical 'magic' once we understand how it works: it creates a target that smooths out the bass modes, flattens the direct sound from your speakers in the midrange and treble frequencies to be closer to an ideal transducer while sensitive to speaker/room/sitting position anomalies, thus avoiding over-correction. If this process makes sense to you or you've been thinking about doing something like it in your system, the app will certainly save you a lot of time compared to manual tweaking, especially with a multichannel system!
In use, the app does a good job walking the user through the process. Although it's not difficult, I would suggest having a moderate level of comfort performing measurements, familiarity with your DSP system like Dirac Live, and experience with post-calibration verification using REW to make sure nothing went awry.
Magic Beans True Target is not snake oil and now having had a chance to consider what it's doing and having experienced the effects for myself, I am of the opinion that the asking price of US$249 for the standard edition is not unreasonable for an enthusiast especially with multiple systems to optimize and over time as the room and components change. Let's just say that for personal music enjoyment, the Magic Beans True Target setting will take a place of prominence among the DSP options in my system!
Having interacted with Joe Mariano on the blog and through E-mails, I appreciate the man's passion in supporting the product. There's also a Discord forum where users can further discuss issues, share tips, and talk about more advanced topics. Highly recommended to check it out even before purchasing the app.
A big thanks to Joe Mariano again for the gentlemanly discussions and opportunity to given the app a whirl on my system!
Whew, long post. 😅
Here's some music to take us out - David Bowie "Sound and Vision" originally from Low (1977):













.png)







%20%5BFINAL%5D.png)

While not the exact same speaker, similar design from the same vendor.
ReplyDeleteCompare your NF measurements to the manufacturer's measurements. They look pretty close above 500 Hz.
https://www.spinorama.org/speakers/Paradigm%20Persona%203F/Paradigm/index_vendor.html
I have compared mine with Klippel measurements from Erin and they line up pretty well.
Another advantage of MB NF measurements, is they can quickly show if there are issues with your speakers pretty quickly.
In our emails, we've gone over anechoic measurements of the Paradigm S8 taken by SoundStage Network. Teal and purple measurements were nearfield measurements from MB, and orange is the anechoic LW measurement of the Paradigm S8. https://photos.app.goo.gl/71w26pGQffR5XXoF9
DeleteYup, even without detailed Klippel and anechoic measurements, the NF measurements are I think very adequate. In Joe's link above with the SoundStage! measurements, good correlation even with an older version of my Paradigm S8's.
DeleteWay to go Joe! Combining quasi/pseudo-spinorama data with room effects results in a practical, real-world tool that would appear to blow away Dirac and Audyssey (and Phil's lazy MLP measurements.) Your subjective comments of course carry quite a bit of weight with me. Might just bite on this one after work on my home is finished. Thanks to both of you!
ReplyDeleteHey there Phil,
DeleteMake sure to come back and let us know how things work out for you in the new space!
Hi Arch
ReplyDeleteThanks for the great article showing how the Magic Beans are working. I have read a bit on their website and safed the tap, maybe for a later use / play with it. But your articles descripes it very well.
All the best
Juergen
BTW: On a side note, yesterday I was at the friends home, that have a WiiM Ultra streamer in his „mid fi“ stereo system and asked me to help him with the WiiM Room Fit with better microfone. So I brought my Earthworks M30 with Preamp with me, connected to my iPad and run the WiiM Room Fit in the individual channel mode and was postively surpriced about the results. Yes, this was no „High End“ system, but for an „average“ Mid Fi customer, for sure of great help (with using the Dayton USB Mic).
Cool Juergen,
DeleteIt's quite amazing how much function is built into devices these days, even on something like the WiiM Ultra and their RoomFit (especially I imagine using the much more accurate Earthworks mic+pre instead of the iPad mic!). Sounds like good results from what I believe is an automated PEQ system - reminds me of the HiFiBerry DSP add-on I had a look into back in 2021 using the Raspberry Pi.
In time, I hope these kinds of technologies will permeate into the audiophile mindset and culture as more of us get used to the power of speaker/room correction!
Cheers.
It is a bit of pity, that RoomFit does (right now) use only one point for the measurement, but even though, it is better than without using Room Fit. Here you can see the basic FRQ measurements in 2 Meter distance.
Deletehttps://magentacloud.de/s/X4GSzXrXNjRbSGD
And as for the audiophile mindset, I can't change their thinkings, but in my over 4 decades of experience, the interaction of the loudspeaker with the room, is always, always the biggest elephant of the complete hifi system.
Nice, even with that single reference, results are better and I trust that correlates with the sound.
DeleteWell, let's hope in time audiophiles can will *finally* appreciate the wisdom and importance of that room-loudspeaker interaction! All the best Juergen.
Hi Arch
DeleteYesterday I was at another friend, also with a WiiM unit, and I have tried the „Beta version of MMM = Moving Microphone Measurements). This works out pretty nice (can also be used by „regular customers“, without too much knowledge).
I have used a very simple Dayton UMM6 mic, and the Mac OS installation of the WiiM Home app. So you can move 60 seconds with the mic around a deticated listening area.
Then you can also adjust the maximum boost and cut in dB, and the frequency range of interest that should corrected. This all is a nice add on in the Beta.
I have attached some screen shots of those feature, in the above given download link (yes, they are in Germany language but still).
Have a nice weekend
All the best,
Juergen
How does Magic Beans compare with Acourate and Audiolense? I have both programs and am wondering whether Magic Beans adds anything that those two programs don't already offer. Am I right in understanding that Magic Beans doesn't offer time correction?
ReplyDeleteMagic Beans is not a room correction system, it works with your correction/DSP system of choice. Magic Beans is all about creating the ideal target for each speaker for you DSP application to implement and correct to.
DeleteOnce you run MB, you have a choice of export options. The most robust is a 31 band PEQ. But there is also a generic export that lets you define the number of bands, max gain and max q for your target system.
I won’t completely speak for JoeNTell, but the idea is that you are not correcting for the room, you are correcting the speaker. A generic target curve is trying to prescribe an outcome, which may not be achievable with your system or your room. Magic Beans is creating a target with the thought that room is going to do what the room is going to do, based on speaker placement and the MLP. It gives you a bass (below the transition region) target based on what the room is providing in terms of gain. This bass target can be per speaker, or common among the speakers which is recommended. Above the transition region Magic Beans is going to try to correct the speakers so the on-axis (also referred as NF in MB) is flat. It does this within the directivity of the speakers and a 6 db boost limit.
DeleteThe interesting thing is that if you had a perfectly measuring (flat NF) set of speakers, in a perfectly treated room, the True Target curve would look very close, if not identical to the Harman target curve for speakers.
That is because that is what Dr. Toole's research concluded. The Harman curve is not a prescription, rather it is a result. It is the result of very good speakers (very flat NF measurements) in a very good room (very little sound impact) is what the trained listeners preferred in the end.
Magic Beans is a different approach to a target. It tries to correct a speaker based on the difference between the in-place speaker and the impact of your room when at your MLP vs prescribing an outcome based on assumptions about your room and your system. (Generic targets)
If you have a tried-and-true method for DSP correcting your system, that is great. This is another option that most likely takes into consideration things most do not. Specifically, your speakers NF response in your room to help inform the process.
Hi Archimago and Joe,
ReplyDeleteThank you so much for all the work you have done on this subject! I have a quick question regarding listening distance and the shaping of the target curve in the HF region. From my experience, I have noticed that the shape affects the perception of height of virtual sources (I'm talking purely about classic stereo systems without height channels). I'm curious if the derivation of the True Target assumes that the NFD-to-MLP magnitude ratio automatically produces the psychoacoustically correct HF tilt for this particular MLP. Or do you still have to perform manual adjustment afterwards—I think Archimago used "Direct" track for this evaluation—do I understand this correctly?
Hi Mikhail. JoeMacJr's response above might help answer your question as well.
DeleteYes, the idea is the "true target" produces the correct HF tilt based on your room, speaker dispersion characteristics, and listening distance. How do we know? Because we take both nearfield and MLP measurements, so we can see exactly what the room does to high frequencies. If you have an ideal speaker with a flat direct response and smooth off-axis response, then the effect of the room on the high frequencies is the exact tilt and roll-off that's required to sound natural in that room.
Here's another way to look at what MB does and doesn't do. We call it speaker response correction and not room correction. When you measure a speaker in a room the typical way, it's really hard for the software to determine whether it's the issue is with the speaker's inherent response, the nearby surfaces changing the direct response, or the rest of the room's surfaces changing the steady state response which includes all the reflections. The term room correction is a misleading term. Only physical treatment can truly affect the room response. Anechoic measurements are useful to understand what the speaker itself is doing, but there's production variations, degradation of parts, and things like nearby surfaces like AT screens that affect the direct sound. MB fixes the direct sound and gives you a smooth and natural bass response for your room.
Trying to make sure I'm following the measurement techniques properly. Am I correct that the measurements for the Magic Beans software is a near field measurement varying from 25" to 35" from tweeter? For measurements I used with REW to create convolution files for Roon for room EQ, I've always measured at the listening position using a moving microphone technique. I like the idea that the Magic Beans method reduces the room effects in the measuring process to get a bit more pure speaker performance profile. If I'm understanding how you performed your tests then it seems that both methods can get to very similar results. My room is not optimized and I like the idea of doing speaker correction and then living with the consequences of the room. I can add some EQ where I need it but hopefully not over cook things with the convolution filters.
ReplyDeleteHey Doug,
DeleteMy comment about 25"-35" was my way of capturing the nearfield/direct sound from my Paradigm S8 speakers which are quite tall with the center distance between the tweeter up top down to the woofer below being somewhere around 27-30".
Magic Beans recommends measuring by slowly moving the microphone back over 10 seconds from a distance around that of the span between drivers in order to capture good multi-driver integration; hence I chose those distances for my specific speakers. For smaller speakers, the distance should be closer - for example the Sony height speakers I'd probably start from about 10"-15" over 10 seconds.
Anyone researching IPTV providers may eventually come across IPTV LIVE USA. Taking a closer look at the service can help viewers understand what it currently offers and whether it deserves consideration. IPTV LIVE USA
ReplyDelete