Car Audio Explained, Part 3: One Seat vs Two, Center Channels, Rear Fill and Upmixing (Andy Wehmeyer's Lessons)

4 days ago

41 min read

This is Part 3 of 19 in the series Andy Wehmeyer on Car Audio, Explained for Beginners. Start here · Download the full guide as a PDF

Unofficial guide. Andy Wehmeyer's points are paraphrased from his public forum posts, blog, tuning guides, videos and podcasts, with a link to each source; the plain-English explanations were written for this guide. Not written, reviewed or endorsed by Andy Wehmeyer or Audiofrog.

1.9 One seat versus two seats

What this part is about: In a car nobody sits in the middle, so a tune that is perfect for the driver can sound lopsided for the passenger. This part covers his recipes for one seat versus two seats, why time delay helps one seat and hurts the other, and the tools he uses instead: all-pass filters (which shift timing without changing loudness) and a center speaker.

1.9 #1 · His ranked recipes: one seat, two seats, and "go nuts".

One seat: a 3-way in each door plus delay. Two seats: the same doors with sail tweeters, a big center and processing that extracts the center (the whole point of a center is that both front seats hear stereo). One seat, "go nuts": add a center, crosstalk cancellation and synthesized reflections. [RARE] Sources: DIYMA MS-8 FAQ p5 #98 · DIYMA 2013-08-02 · DIYMA 2022-02-24

In plain English: A "3-way" splits each side's sound across three speakers: a woofer for low notes (midbass), a midrange for the middle, and a tweeter for high notes. "Delay" (time alignment) means a processor holds back the nearer speakers so all the sound reaches your ears at the same moment. Sail panels are the small triangular panels at the front corner of each door, by the mirror. A center speaker sits in the middle of the dash. "Extracting the center" means a processor pulls out only the sound that left and right share and sends it there. He ranks three recipes. For one listener: a 3-way in each door plus delay. For two front seats: the same doors with sail tweeters, a big center and center-extracting processing, because the whole point of a center is that both seats hear stereo. For one seat, going all out ("go nuts"): add a center, crosstalk cancellation (processing that stops each ear from hearing the opposite speaker) and synthesized (artificially created) reflections.

1.9 #2 · Delay fixes one seat and doubles the error in the other.

The distance error is equal and opposite for the two front seats. His rule: "the passenger's seat can't suck"; a tuner who leaves other seats sounding like garbage is "a learner", not a "tuning God". Carmakers back the two-seat goal with heavy market research. [RARE] Sources: DIYMA 2016-03-16 · D&F live @26:03 · DIYMA 2021-09-09

In plain English: You always sit closer to one front speaker than the other, so its sound arrives first. Delay holds back the near speaker so both arrive together. The catch: the passenger sits in the mirror-image spot. The speaker that is "near" for the driver is "far" for the passenger, so delaying it makes it arrive even later there. The passenger's timing error was already the same size as the driver's, just the other way round ("equal and opposite"); after the delay it is twice as big. His rule is that "the passenger's seat can't suck". A tuner who leaves the other seats sounding like garbage is, in his words, "a learner", not a "tuning God". He adds that carmakers back the two-seat goal with heavy market research. Many people believe a tune is finished once the driver's seat sounds perfect; he disagrees.

1.9 #3 · A-pillar cars without a center are always one-seat cars.

He has never heard one that wasn't; centered vocals need no center speaker only in a car optimized for one seat. [RARE] Sources: DIYMA 2014-06-16 · DIYMA 2021-12-07

In plain English: A-pillars are the posts on each side of the windshield, a popular spot for speakers. Many people believe that speakers up there, well tuned, give everyone a centered singer without a center speaker. He says he has never heard an A-pillar car without a center that wasn't a one-seat car. A singer can sound centered with no center speaker only when the car is optimized for one seat; move to the other seat and the center falls apart.

1.9 #4 · What breaks the image for both seats is phase near 250 Hz, and an all-pass fixes it without delay.

With door speakers, left and right sit near −180° at ~250 Hz in both seats; an all-pass reverses one speaker's phase over a narrow band there, helping both seats at once: not perfect in either, but better. [RARE] [Changed view: 2013–15 he judged phase EQ inaudible next to delay for one seat; in Jan 2016 he said new DSPs should drop L/R delay for phase EQ.] Sources: D&F live @21:00 · DIYMA 2016-01-25 · DIYMA 2016-03-16

In plain English: Sound is a wave, and "phase" describes where in its up-and-down cycle the wave is at a given moment. If two speakers play the same note but one is pushing while the other is pulling (180 degrees apart), they cancel, like two people pushing a swing from opposite sides. Because of the different distances to the door speakers, left and right arrive about half a cycle apart near 250 Hz (a lower-midrange note) in both front seats. Half a cycle early or half a cycle late cancels the same way, which is why both seats suffer. An all-pass filter is a processing tool that leaves loudness alone and only shifts the phase of a chosen band. He says putting one on one speaker reverses its phase over a narrow band there, which helps both seats at once: not perfect in either, but better. Delay cannot do that. His view changed: in 2013–15 he judged phase EQ (adjusting phase by frequency) inaudible next to delay for one seat; in January 2016 he said new processors should drop left/right delay in favor of phase EQ.

1.9 #5 · Carmakers already do it: Toyota, Subaru, Bose and a Harman patent.

Toyota uses all-pass filters from basic systems up to JBL; Bose's Fiat 500 images in both seats with plain stereo. In such cars add no front delay and leave "reversed" factory polarity alone: the filters can't be inverted, and added delay makes them work poorly. [RARE] [Conflict: he recalls Harman patent US6876748 naming 250 Hz; its text cites about 200 Hz.] Sources: D&F live @23:05 · DIYMA 2021-05-13 · US6876748 (Google Patents)

In plain English: He says carmakers already use this all-pass approach, naming Toyota, Subaru and Bose. Toyota uses all-pass filters from its basic systems up to its JBL premium systems, and Bose's Fiat 500 system makes a good image in both seats with plain stereo (no center). A Harman patent covers the idea. Polarity is which way a speaker's cone moves for a given signal; a speaker wired "reversed" moves in when it should move out. If you upgrade such a car, he says: add no front delay, and leave factory polarity that looks "reversed" alone. The filters can't be inverted (undone), and added delay makes them work poorly. The [Conflict] tag: he remembers the patent naming 250 Hz, but the patent's own text cites about 200 Hz. Both figures are given here; this explanation doesn't choose between them.

1.9 #6 · Two quick tests show whether a factory system uses all-pass filters or plain delay.

From each front seat, listen to a centered vocal: with all-pass it stays roughly central for the driver and both sides stay audible from the passenger seat; with delay it collapses into the passenger door. Or feed mono pink noise and view L+R: one big dip at 200–500 Hz means all-pass; an even comb means delay. [RARE] Sources: DIYMA 2021-10-26

In plain English: These tests tell you how a factory system handles the two seats. Test one: play a song with a singer mixed in the center and listen from each front seat. With all-pass filters, the singer stays roughly central (not pinpoint) from the driver's seat, and from the passenger seat you can still hear both sides. With plain delay, everything collapses into the passenger door. Test two: play mono pink noise (a steady hiss containing all frequencies, identical in both channels) and view left plus right combined (L+R) on your processor's input analyzer. One big dip somewhere between 200 and 500 Hz means all-pass. An even "comb" (a row of regularly spaced dips, like the teeth of a comb) means delay. That is because a delay shifts timing at every frequency, so cancellations repeat up the range, while an all-pass shifts only one region.

1.9 #7 · His 2016 starting recipe: one 2nd-order all-pass near 250 Hz, Q about 1, on one side.

Pick where half a wavelength equals the path difference; it needn't be exact. Size it with his phase "viewer": left/right phase differences of 130° or more count as 180°, and Q ≈ center frequency ÷ the region's width. Use phase EQ only from ~80 Hz to 1 kHz. [RARE] Sources: DIYMA 2016-03-16/18 · DIYMA 2016-03-17/04-03

In plain English: This is his 2016 recipe for setting up that all-pass filter. A "2nd-order" all-pass rotates the phase through a full cycle across a band of frequencies, reaching a half-cycle (180 degrees) at its center frequency. "Q" sets how wide or narrow the filter's effect is: a higher Q is narrower, and a Q of about 1 is fairly broad. Put it on one side only, near 250 Hz, at the frequency where half a wavelength equals the path difference (the difference between your distances to the left and right speakers), because that is where the two sides arrive exactly half a cycle apart. He says it needn't be exact. To size it, he used a phase "viewer" he built: subtract one side's measured phase from the other's, count any difference of 130 degrees or more as 180 (a cancellation zone), then set Q to roughly the center frequency divided by the zone's width. For example, a zone centered on 250 Hz and 250 Hz wide gives a Q of about 1. Use phase EQ only from about 80 Hz to 1 kHz.

1.9 #8 · The FIR version: phase correction below 1 kHz only, and no left/right delay.

FIR is a filter type that can shape phase freely. Stitch a single-mic measurement below 1 kHz to a spatial average above, leaving the top minimum-phase; with no L/R delay this may give a two-seat stereo car without a center ("watch 250 Hz"). Spending 64k taps on 10–20 kHz is excessive. [RARE] Sources: DIYMA 2016-03-16

In plain English: FIR (finite impulse response) is a kind of digital filter that can shape phase freely, separately from loudness. Its "taps" are the number of calculation steps; more taps means finer control and more processing. His recipe: correct phase only below 1 kHz. Measure below 1 kHz with the mic at a single spot, measure above 1 kHz as a spatial average (several mic spots averaged), and stitch the two together. Above 1 kHz, leave the phase "minimum-phase", meaning no deliberate phase correction there. Set no left/right delay. Done this way, he says, it may give a two-seat stereo car without a center speaker, with the caution to "watch 250 Hz". In the same 2016 thread he explains the 1 kHz split: above it, phase matters little, and an averaged measurement better represents what we hear. Separately, he notes that a 128k-tap filter puts about 64,000 (64k) of its taps between 10 and 20 kHz, the top octave. He calls that excessive, because most of the resolution lands there at the cost of processing power and latency (processing delay).

1.9 #9 · Sub-to-midbass phase differs by seat even with a center: fix it with all-pass, not delay.

In his car the sub-to-near-midbass distance was half a wavelength at the crossover, and delaying the near midbass pulled the bass toward that side for both seats. Fix: an all-pass on the sub plus one half an octave higher on one midbass, so no pair sits at −180°. [RARE] Sources: DIYMA 2016-01-22 · DIYMA 2016-01-28

In plain English: A subwoofer plays the deepest bass; the midbass woofers (here, in the doors) play the next range up. The crossover is the frequency where one hands over to the other. His point is that adding a center speaker does not fix this hand-over, because sub-to-midbass phase still differs between the seats. In his own car, the paths from the sub and from the nearer door midbass to the listener differed by half a wavelength at the crossover frequency, so the two cancelled there. Flipping polarity did not help, and delaying the near midbass pulled the bass toward that side in both seats. His fix: an all-pass filter on the sub, plus another set half an octave higher on one midbass (an octave is a doubling of frequency). That way no pair of these speakers sits at −180 degrees (cancelling), and he reports that the bass then sat centered in front of the dash in both front seats.

1.9 #10 · All-pass imaging sounds bigger and less pinpoint, and by 2021 he wouldn't teach it.

Driver's seat: a real but larger center than a delay-tuned "diorama of tiny army men on the dash"; passenger seat: more diffuse, still spread across the dash. By 2021 he called it fiddly (an all-pass on the later channel can create a new dip; try the other side) and teaches only repeatable processes. [RARE] [Changed view: 2016 enthusiasm → 2021 "helps two seats, but only with fiddling".] Sources: D&F live @23:05 · DIYMA 2021-09-07 · DIYMA 2021-08-27

In plain English: He compares the two approaches by ear. A delay tune for one seat gives a small, precise picture, which he likens to a "diorama of tiny army men on the dash" (a diorama is a miniature model scene). With all-pass filters, the driver's seat gets a real but larger, less pinpoint center; the passenger seat gets a more diffuse sound that still spreads across the dash. The [Changed view] tag: in 2016 he was enthusiastic, but by 2021 he called the method fiddly. An all-pass on the later-arriving channel can create a new dip, and then you try the other side. Because he teaches only repeatable processes, he would no longer teach it, though he still says it helps two seats with fiddling.

1.9 #11 · MS-8 seat modes: Driver, Passenger and "Front".

Driver or Passenger sets time alignment for that seat; "Front" drops L/R alignment but keeps it within each side. Without a center, Front balances tone but won't image; with one, the center gives good imaging in all seats. Gary Biggs won IASCA with stock BMW speakers, "Front" mode and the center up "a notch or two". [Era: MS-8, 2008–2012] Sources: DIYMA MS-8 FAQ #7 · DIYMA MS-8 FAQ #6 · DIYMA MS-8 FAQ p2

In plain English: The MS-8 was a JBL car processor sold around 2008–2012 that tuned a car automatically from microphone measurements; this point is about its settings. "Driver" or "Passenger" mode sets time alignment (delay) for that seat. "Front" mode drops the left/right alignment but keeps alignment within each side, for example between the tweeter and woofer on the same side. Without a center speaker, Front mode gives an even tonal balance but no focused image. With a center, the center provides good imaging in all seats. His example: Gary Biggs won IASCA (a car-audio sound-quality competition) with the stock BMW speakers, "Front" mode, and the center turned up "a notch or two".

1.9 #12 · The polarity-flip trick is a pre-DSP half-fix; never combine it with delay.

Reversing one mid fixes the ~250 Hz center cancellation but moves the problem to ~500 Hz and kills midbass; it dates from when polarity was the only tool. [RARE] Sources: Time Alignment Part 2 · DIYMA MS-8 FAQ #26 · D&F live @33:05

In plain English: Reversing polarity means swapping a speaker's plus and minus wires, so its cone moves the opposite way. An old trick was to reverse one front midrange to cure the cancellation near 250 Hz that ruins the center image. He calls it a half-fix from before processors existed, when polarity was the only tool. It moves the problem to about 500 Hz, where the two sides had been back in step, and it "kills" the midbass, because at low notes the two sides were in step and now cancel. His rule: never combine the flip with delay.

1.9 #13 · Blur the center on purpose for listeners who hate it, without losing coherent bass.

Ask in the parking lot where they hear the singer. In an active 3-way, delay the midbass but not mids and highs; in a 2-way plus sub, back delay off until drivers are in phase only around the ~100 Hz crossover. Never just switch delays off ("nobody hates coherent bass"); club-style listeners need a DSP for midbass-to-sub impact too. [RARE] Sources: OFCAP Ep. 38 @57:06 · OFCAP Ep. 38 @1:00:03 · CMA Connected Oct 2021 @46:57

In plain English: Some listeners don't like a sharply focused center image. He suggests asking them in the parking lot where they hear the singer. To blur the center on purpose without losing tight, unified bass: in an active 3-way (each speaker on its own amplifier channel, with a processor doing the crossovers), delay the midbass but not the mids and highs. In a 2-way plus sub, reduce the delay settings until the speakers are in phase only around the ~100 Hz crossover. His explanation: in phase, you hear an image between the speakers; out of phase, you hear the speakers themselves. Never just switch the delays off, because "nobody hates coherent bass". He adds that club-style listeners need a processor too, so the punch from midbass to sub still lines up.

1.9 #14 · Judged cars game imaging; customers deserve "close to correct".

Competitors tune per track: without per-track EQ he'd skip delay and use an all-pass for the center-vocal track, leaving everything above the vocal fundamental out of phase, which judges hear as width. Score sheets should reward two seats; until a car sounds larger than itself and alike in all seats, nobody should score a 10. [RARE] Sources: OFCAP Ep. 38 @54:01 · SQOLOGY P2 @21:02

In plain English: In judged competitions, he says, competitors tune for the specific test tracks. If he couldn't change EQ per track, he would skip delay and use an all-pass filter aimed at the center-vocal track, fixing phase around the singer's fundamental (the lowest main note of the voice). Everything above that would stay out of phase, which lets you hear where the speakers are, and judges hear that as width. His point is that judged cars game imaging this way. He thinks score sheets should reward cars that work in two seats, and that nobody should score a 10 until a car sounds larger than itself and alike in all seats. Winning a contest, he says, isn't the same as great audio; for customers, the goal is to get "close to correct".

1.10 The center channel

What this part is about: A center channel is a speaker in the middle of the dashboard. This part covers what signal a proper center should get, what kind of speaker it needs and where it goes, and how having one changes the way the rest of the system is tuned.

1.10 #1 · "L+R is not a center": a summed mono center hurts more than it helps.

There's "no happy medium" between loud enough to form a center and quiet enough to keep width, and passive L+R/L−R matrices "do not work". For a summed center: "turn it off and take the speaker out". Stereo music needs an upmixer, not a Dolby decoder. [RARE] Sources: DIYMA MS-8 FAQ #10 · DIYMA MS-8 FAQ p5 #82 · DIYMA 2018-04-06

In plain English: "L+R" means simply mixing the left and right channels together into one mono signal and sending that to a center speaker. Many people believe any center speaker fed this way helps. He says it hurts more than it helps. There is "no happy medium": loud enough to create a center image, it squeezes the stage narrow; quiet enough to keep the width, it doesn't create a center. Passive L+R/L−R matrices (simple wiring networks with no processor) "do not work" either. His advice for a summed center: "turn it off and take the speaker out". For stereo music you need an upmixer (a processor that splits stereo across more speakers), not a Dolby decoder (built for movie soundtracks).

1.10 #2 · A real center gets only what L and R share, and that content stays in L and R about 6 dB down.

A "hard center" of (L+R)−(L−R) is just 2R; extraction needs absolute values, which analog can't do. Off-center images form between L+C and C+R, so any polarity error spoils them. Remove center content from L/R entirely and images shrink; don't attenuate it and the stage narrows. [RARE] Sources: DIYMA MS-8 FAQ #11 · DIYMA 2017-07-01 · DIYMA 2018-03-29

In plain English: In a stereo recording, a sound "in the center", such as the lead singer, is identical in left and right. A proper center channel gets only that shared content, and the same content stays in left and right about 6 dB quieter (a clearly audible step down). He explains why simple math can't do this: a "hard center" of (L+R)−(L−R) works out to just 2R, a doubled right channel. True extraction needs absolute values (the size of a difference, ignoring its sign), which analog circuits can't calculate. Sounds slightly off-center, like a guitar just left of the singer, form between the left and center speakers or between the center and right ones, so any polarity error (a speaker wired backwards) spoils them. Getting the amount right matters: remove the center content from left and right entirely and images shrink; leave it at full level and the stage narrows.

1.10 #3 · With real center processing, the center becomes the main front speaker.

Most popular music is mixed center, so it must play as loud as L/R with similar response, include a tweeter, and cover 200 Hz–20 kHz at minimum (100 Hz is better), on main-channel power. Tweeters at the mirror are not a center. [RARE] Sources: DIYMA MS-8 FAQ p2 · DIYMA MS-8 FAQ p2 (tweeters) · DIYMA 2018-03-28

In plain English: Once a processor sends true center content to a center speaker, he says, that speaker becomes the main front speaker, because most popular music is mixed to the center. So it must play as loud as left and right, with a similar frequency response (similar tonal balance from low to high), and include a tweeter. It must cover 200 Hz to 20 kHz at a minimum, and reaching down to 100 Hz is better. It should run on main-channel power, the same kind of amplifier power the left and right get. In his view, tweeters at the mirror are not a center.

1.10 #4 · Bass management follows the crossover you SET, so set the center's high-pass where it really plays.

Content between what the center can reproduce and the high-pass you entered "is lost". Start with center high-pass 200 Hz and doors-to-sub 80 Hz; below the center's range the image depends on L/R time alignment again, so a 3" center with 8" doors means "there's gonna be trouble". [RARE] [Era: MS-8] Sources: DIYMA MS-8 FAQ p2 · DIYMA MS-8 FAQ p5 #93

In plain English: "Bass management" means the processor sends the low notes a speaker can't handle to other speakers: the center's lows to the left and right doors, and the doors' lows to the sub. A high-pass filter lets through only the notes above a set point. His warning is that the processor routes sound by the crossover numbers you type in, not by what the speaker can actually play. Content between the center's real lower limit and the high-pass you entered "is lost". His starting values: center high-pass at 200 Hz, doors-to-sub at 80 Hz. Below the center's range, the content plays from the left and right doors again, so the image depends on left/right time alignment again. That is why a 3-inch (about 75 mm) center with 8-inch (about 200 mm) door speakers means "there's gonna be trouble". This was MS-8-era advice.

1.10 #5 · His preferred center: the biggest mid that fits plus a tweeter, on the dash near the windshield junction, firing up.

Below ~1 kHz an 8" radiates evenly, so it's effectively on-axis whichever way it points, and the image sits between speaker and glass: higher, with "a nice, big center image". Seal its baffle. Its dip near 2 kHz is a feature, because ear spacing puts a similar dip in a phantom center (Shively and House, Harman AES papers, late 1990s). [RARE] Sources: DIYMA MS-8 FAQ p4 #71 · DIYMA MS-8 FAQ p5 #100 · DIYMA 2021-09-22

In plain English: His preferred center is the biggest midrange speaker that fits, plus a tweeter, mounted on the dash close to where it meets the windshield, firing upward. Below about 1 kHz, an 8-inch (about 200 mm) speaker spreads sound evenly in all directions, so it acts as though it is pointed at you whichever way it faces. The image then forms between the speaker and the glass, which places it higher and gives "a nice, big center image". Seal its baffle (the panel it is mounted in) so no air leaks around it. Such a speaker shows a dip near 2 kHz, which, he explains, comes from the reflection off the windshield. He says the dip is a feature, not a flaw: because of the spacing of our ears, a phantom center (the illusion of a center made by left and right speakers) has a similar dip. He credits this to late-1990s Harman AES papers by Shively and House.

1.10 #6 · A center's line of sight barely matters.

Up-firing and on-axis centers differ only above ~6 kHz without a tweeter (~10 kHz with one), like hearing someone yell from the kitchen. Sources: DIYMA 2013-12-05

In plain English: "On-axis" means the speaker points straight at you. He says it barely matters whether a center fires up at the windshield or points at you. The two sound different only above about 6 kHz (a high treble note) without a tweeter, or about 10 kHz with one. His comparison is hearing someone yell from the kitchen: you can't see them, but you hear them fine.

1.10 #7 · No timbre matching needed: EQ matches the center to L/R.

"No timbre matching by identical speakers is required"; EQ, not identical drivers, matches L/C/R. A coaxial works there, even a 2-way with its midbass in a small box behind the console. [RARE] Sources: DIYMA 2010-04-01 · DIYMA 2021-01-02 · DIYMA 2021-09-22

In plain English: Timbre is the tonal character of a sound. Many people believe the center must use the same model of speaker as the left and right so they sound alike. He says "No timbre matching by identical speakers is required": equalization (EQ, adjusting the loudness of different frequencies) is what makes left, center and right match. So a coaxial (a speaker with a tweeter mounted in its middle) works as a center, and so does a 2-way whose midbass sits in a small box behind the console.

1.10 #8 · He has "NEVER heard" a good center in a Harman factory system.

The factory center he replaced in 2016 made the whole car sound bad except the midbass, and he said a new algorithm's center target should have a deeper dip near 2 kHz. [RARE] Sources: DIYMA 2016-01-18 · DIYMA 2016-01-25

In plain English: He worked at Harman, which makes factory sound systems for many carmakers, yet he says he has "NEVER heard" a good center in a Harman factory system. The factory center he replaced in 2016 made the whole car sound bad except the midbass. He also said a new automatic-tuning algorithm should aim the center at a target with a deeper dip near 2 kHz (compare 1.10 #5).

1.10 #9 · With a real center, L/R delay barely matters; place images with level.

Steering exists "to minimize the importance of PLDs" (path-length differences). Give the center little or no delay (it's nearly equidistant), match its response to L/R, then aim for a mid-dash center for every seat: turn it down a bit to pull the image toward each listener, or up "a notch or two" for two seats. [RARE] Sources: DIYMA MS-8 FAQ #17 · DIYMA 2010-04-04 · EDUCAR

In plain English: "Steering" is when a processor moves each sound to the speaker where it belongs. He says it exists "to minimize the importance of PLDs", the path-length differences (your unequal distances to the left and right speakers) that delay normally has to fix. So with a real center, left/right delay barely matters, and you place images with level instead. The center needs little or no delay because it is nearly the same distance from both front seats. Match its response to left and right, then aim for a center image in the middle of the dash for every seat. Turning the center down a bit pulls the image toward each listener (some factory designers want it straight in front of each person); turning it up "a notch or two" suits two seats.

1.10 #10 · Center polarity trick for windshield centers.

With an up-firing center and L/R mids low in the doors or kicks, reversing the center's polarity and recalibrating sometimes focuses the center; first confirm the L/R mids are wired correctly. Sources: DIYMA 2013-03-05

In plain English: This is a trick for an up-firing center that bounces off the windshield while the left and right mids sit low, in the doors or in the kick panels (by your feet). Reversing the center speaker's polarity (swapping its plus and minus wires) and then recalibrating (re-running the automatic tune) sometimes makes the center image more focused. He says to first confirm that the left and right mids are wired correctly. He says "sometimes", so it is something to try, not a guaranteed fix.

1.11 Rear and side speakers

What this part is about: Many cars have speakers in the rear doors, on the rear shelf or in the back pillars. This part explains what he thinks those speakers are for (a sense of space, not sounds from behind you) and how to feed, place and set them, both with and without an upmixer.

1.11 #1 · Rears are for envelopment, not rear events.

The cabin's early reflections tell us we're in a car; later sound from the sides and rears lets us believe in a bigger space. With an upmixer, rears put you in the room instead of watching it through a window; a car can't sound like the Rose Bowl but can suggest a space larger than itself. [RARE] Sources: DIYMA MS-8 FAQ p3 · OFCAP Ep. 19 @3:28:03 · EDUCAR

In plain English: Many people believe rear speakers exist so you can hear sounds coming from behind, or that a serious sound-quality car should switch them off. He says rears are for envelopment, the feeling of being surrounded by the space of the recording. His reasoning: the early reflections in a car (sound bouncing off nearby surfaces, arriving just after the direct sound) tell your brain you are in a small space, a car. Sound that arrives later from the sides and rear lets you believe in a bigger space. With an upmixer, rears put you in the room instead of watching it through a window. A car can't sound like the Rose Bowl (a huge stadium), but it can suggest a space larger than itself.

1.11 #2 · Rears are where extra width starts, if used carefully.

Older recordings put out-of-phase "width" content in the opposite channel, which Logic7/PLII steer to the rears; rears at the right delay and level form phantom images between front and rear on each side. Rolling off their highs makes them harder to localize, and they sit far from the windshield. [RARE] Sources: DIYMA 2011-11-17

In plain English: He says rears are where extra width begins, if handled carefully. Older recordings often created "width" by placing a sound, phase-reversed, in the opposite channel. Matrix upmixers such as Logic7 and PLII (Pro Logic II) detect that out-of-phase content and steer it to the rears. Set at the right delay and level, the rears then form phantom images between the front and rear speakers on each side, so sounds can appear beside you. Rolling off their highs (turning down the treble) makes the rear speakers harder to pinpoint. He also notes they sit far from the windshield; in the same discussion he names windshield reflections as something that narrows the stage.

1.11 #3 · Rears must be stereo, never mono or just "front speakers mounted in back".

Mono rear fill is one of the "bizarro" layouts. Short of channels? Run the front mid and tweeter on one channel with a passive rather than give up stereo rears. [RARE] Sources: DIYMA MS-8 FAQ p2 · DIYMA 2010-05-27 · DIYMA 2015-08-23

In plain English: Rear fill is sound from rear speakers that supports the front stage. He insists rears be stereo (separate left and right), never mono, and never just "front speakers mounted in back", meaning unprocessed copies of the front sound. He calls mono rear fill one of the "bizarro" layouts. If you are short of amplifier or processor channels, his answer is to run the front midrange and tweeter on one channel with a passive crossover (a small filter that splits one channel between the two speakers) rather than give up stereo rears.

1.11 #4 · His rear-fill recipe without an upmixer: plain stereo, high-passed, raised, then delayed until the stage returns to the front.

Tune the front first. Feed the rears normal stereo with only a high-pass (at least 100 Hz) and EQ out big peaks. Raise them until the stage starts moving rearward, then add delay (~10–20 ms) until it's anchored in front; more level needs more delay, and ~35 ms becomes an echo. [RARE] Sources: DIYMA 2021-07-25 · DIYMA 2021-12-07 · OFCAP Ep. 19 @3:33:00

In plain English: This is his step-by-step recipe for rear fill without an upmixer. First tune the front speakers. Then feed the rears normal stereo with only a high-pass filter (which blocks the bass) at 100 Hz or higher, and use EQ to remove any big peaks. Raise the rear level until the stage (where the music seems to be) starts moving toward the back. Then add delay, around 10–20 ms (thousandths of a second), until the stage is anchored in front again. The idea rests on a basic fact of hearing: we tend to place a sound where it first arrives from, so slightly late rear sound can add space without pulling the image backward. The more level you give the rears, the more delay they need, and at about 35 ms the rear sound becomes a distinct echo.

1.11 #5 · He dropped differential (L−R) rear fill for plain stereo.

In 2017 he gave a miniDSP recipe for L−R rears delayed 12–20 ms, and says he may have helped give L−R fill its "semi-mythical" status (a member's phrase). By the 2020s he called it nasal, honky and "phasy"; stereo changes the response less and works on nearly all music. Wiring rears out of phase, the pre-DSP trick, cancels bass and center. [RARE] [Changed view: 2017 L−R → 2020s in-phase stereo] Sources: DIYMA 2017-10-06 · D&F live @33:05 · DIYMA 2021-12-07

In plain English: "Differential" or L−R rear fill feeds the rears the left channel minus the right, which keeps only what differs between them and removes centered sounds. In 2017 he gave a recipe for doing this on a miniDSP (an inexpensive processor) with the rears delayed 12–20 ms. He says he may have helped give L−R fill its "semi-mythical" status (a forum member's phrase). By the 2020s he had dropped it, calling it nasal, honky and "phasy", and he now prefers plain stereo, which changes the tonal response less and works on nearly all music. The old pre-processor trick of wiring rears out of phase cancels bass and center content. The [Changed view] tag marks this shift from L−R (2017) to normal in-phase stereo (2020s).

1.11 #6 · If you still build L−R fill: it's one signal, and leftover center usually means mismatched inputs.

For L−R on one side and R−L on the other, just reverse one rear speaker's polarity (a bridgeable amp already inverts one channel inside). Center "leaking" into L−R usually comes from test noise that isn't truly mono (or is lossy), unequal L/R EQ or level, or phase differences between head-unit outputs. Sources: DIYMA 2021-11-24 · DIYMA 2022-01-15

In plain English: If you still want L−R rear fill, he gives two practical notes. First, L−R on one side and R−L on the other are the same signal, one simply upside down, so you can just reverse one rear speaker's polarity (swap its wires). A bridgeable amplifier (one that can join two channels into one) already inverts one channel inside. Second, if centered sound "leaks" into the L−R rears, where it should cancel out, the usual causes are mismatched inputs: test noise that isn't truly mono (or is lossy, meaning compressed like an MP3), unequal left/right EQ or level, or phase differences between the head unit's outputs.

1.11 #7 · Logic7 rule: high-pass rears and sides at 100 Hz, 24 dB/oct, because one of them gets polarity-flipped.

To anchor front-steered sounds in front, Logic7 inverts one rear/side channel, which would cancel midbass (which belongs up front anyway). Rears should play 100 Hz to at least 10 kHz with a tweeter ("Coaxials are great for rears"). [RARE] [Era: MS-8/Logic7] Sources: DIYMA MS-8 FAQ #11 · DIYMA MS-8 FAQ #16 · DIYMA MS-8 FAQ p2

In plain English: This rule is for Logic7, the Harman upmixer used in the MS-8 era. High-pass the rears and sides at 100 Hz with a steep 24 dB-per-octave slope. The reason: to keep sounds that belong up front anchored in front, Logic7 flips the polarity of one rear or side channel, and if those speakers played midbass, the flipped one would cancel it. The midbass belongs up front anyway. He says rears should play from 100 Hz to at least 10 kHz with a tweeter: "Coaxials are great for rears" (a coaxial has its tweeter mounted in the middle of the woofer).

1.11 #8 · Sides at 12 ms, rears at 20 ms, tweeters at ear level.

Rear-door speakers act as sides and package-tray speakers as rears; mount tweeters high with a clear path to the ears (no aiming needed). He prefers sides over rears (they widen the stage a bit) and ear-level spots like C/D-pillars; 5.1 versus 7.1 is "really hard to hear". [RARE] Sources: DIYMA 2015-03-03 · DIYMA MS-8 FAQ p3

In plain English: These are his delay and placement figures. In his scheme, the rear-door speakers act as "sides" and the package-tray speakers (on the shelf behind the back seat) act as "rears". Delay the sides by 12 ms and the rears by 20 ms. Mount tweeters high, with a clear path to the ears; no aiming is needed. He prefers sides over rears because they widen the stage a bit, and likes ear-level spots such as the C- or D-pillars (the body posts behind the rear doors). He finds 5.1 versus 7.1 (five versus seven main speaker channels, plus a sub) "really hard to hear".

1.11 #9 · Don't buy fancy rear speakers, but do use rear tweeters.

"Use what you have": matching rear size or timbre to the fronts isn't critical, since rears play mostly processed content at low level. Rear tweeters help even in an "SQ" car; he mocked treating them as a competition secret. [RARE] Sources: DIYMA 2010-06-01 · DIYMA 2010-05-22 · DIYMA 2021-07-24

In plain English: Many people assume rear speakers must match the fronts in size or sound. He says "Use what you have": because rears play mostly processed content at low level, matching their size or timbre (tonal character) to the fronts isn't critical, so there's no need to buy fancy ones. He does recommend rear tweeters, even in an "SQ" (sound-quality competition) car, and he mocked the idea of treating them as a competition secret.

1.11 #10 · Rears are the condiment.

Front L/R are the bread, the center the meat and cheese, the rears the condiment. Under Logic7 rear level depends entirely on the recording's out-of-phase content, and soloed rears swell and fade "like a wave", so solo them only to diagnose. [RARE] Sources: DIYMA MS-8 FAQ p3 · DIYMA MS-8 FAQ p3 (solo)

In plain English: His sandwich comparison: the front left and right speakers are the bread, the center is the meat and cheese, and the rears are the condiment. They add flavor but are not the meal. Under Logic7, how loud the rears play depends entirely on how much out-of-phase content the recording has. If you solo them (play them alone), they swell and fade "like a wave". So solo the rears only to diagnose a problem.

1.11 #11 · One stage for everyone, not a second stage for the back seat.

OEM Logic7 gives rear passengers their own stage at the front seatbacks, which he calls "seriously difficult"; he prefers one stage at the dash for every seat. For the back seat, delay the sides to arrive with the fronts there, and raise them only until the front images start moving rearward. Sources: DIYMA MS-8 FAQ #11 · DIYMA 2016-01-16

In plain English: Factory Logic7 systems give back-seat passengers their own stage at the backs of the front seats. He calls that approach "seriously difficult" and prefers one stage, at the dash, for every seat. To make that work for the back seat, delay the side speakers so their sound arrives at the rear seats together with the sound from the fronts. Then raise the sides only until the front images start moving backward.

1.12 Upmixing: matrix decoders, factory upmixers and music upmixers

What this part is about: An upmixer is a processor that takes ordinary two-channel stereo and spreads it across more speakers: a center, sides and rears. This part explains the different kinds (older "matrix" decoders made for movies, the ones built into factory systems, and newer music-focused ones), how they work and fail, how to check and set them, and his long experience living with them.

1.12 #1 · How matrix upmixers (Pro Logic II, Logic7) steer.

An upmixer spreads stereo over more channels. Matrix types compute a steering angle thousands of times a second and ride whole-channel gains: left/center/right from level differences, front/rear from phase (in-phase forward, out-of-phase rearward). They were built to decode surround hidden in 2-channel film media like VHS. [RARE] Sources: DIYMA MS-8 FAQ #11 · DIYMA 2018-04-06 · DIYMA 2017-07-01

In plain English: An upmixer spreads stereo over more channels. "Matrix" types, such as Dolby Pro Logic II and Harman's Logic7, work out a "steering angle" (an estimate of where the main sound is coming from) thousands of times a second. They then ride, meaning turn up and down, the volume of whole channels. Left/center/right is decided from level differences: a sound louder in the left channel goes left, and one equal in both goes to the center. Front/rear is decided from phase: in-phase sound (left and right moving together) goes forward, out-of-phase sound (moving opposite) goes rearward. He notes they were built to decode surround sound hidden inside two-channel film media such as VHS tapes.

1.12 #2 · Check the steering with pink noise.

Mono pink noise should come mostly from the center (L/R about 6 dB down); invert one channel and it moves to the rears; balance hard right should play only the right front. Stereo pink noise, with the center level set by the mono share, is probably the best check. [RARE] Sources: DIYMA 2017-08-05 · DIYMA MS-8 FAQ #25

In plain English: Pink noise is a steady hiss containing all frequencies, used for testing. He gives simple checks that a matrix upmixer is steering properly. Mono pink noise (identical in both channels) should come mostly from the center, with left and right about 6 dB down (clearly quieter). Invert one channel (flip its polarity) and the noise should move to the rears. Turn the balance hard right and only the right front should play. Stereo pink noise (different noise in each channel) plays through all the speakers, with the center's level depending on how much of the content is mono; he calls this probably the best check.

1.12 #3 · Where matrix upmixing fails, and what breaks it.

It can't steer a transient one way while a steady sound goes another (speech in front kills steady rain in the rears). Weak FM, factory per-band delays and FIR processing look like phase shifts and steer strangely, as do producers' out-of-phase tricks and lossy files. Remove factory delay, not just EQ, and never stack two upmixers. [RARE] Sources: DIYMA MS-8 FAQ #16 · DIYMA 2018-04-07 · EDUCAR

In plain English: Because a matrix upmixer turns whole channels up and down, it can't send a short, sharp sound one way while a steady sound goes another. His example: speech in front kills steady rain in the rears, because the speech pulls the steering forward. Anything that looks like a phase shift fools it and makes it steer strangely: weak FM radio reception, factory delays applied per frequency band, and FIR processing. So do producers' out-of-phase studio tricks and lossy files (compressed formats like MP3). His practical rules: when feeding it a factory signal, remove the factory delay, not just the factory EQ, and never stack two upmixers one after the other.

1.12 #4 · Feed an upmixer two channels, not discrete 5.1, and use 5.1 outputs rather than 3.1.

Logic7 builds center, sides and rears from stereo; for DVDs pick the 2-channel downmix. Home 5.1 assumes a TV, dialogue and near-equilateral seating, none of which fits a car; upmixing is not just for movies. [RARE] Sources: DIYMA MS-8 FAQ #8 · DIYMA 2018-06-04

In plain English: "Discrete 5.1" is a movie soundtrack already split into separate channels: front left, center, front right, two surrounds and a sub. He says to feed an upmixer two-channel stereo instead, and to use its 5.1 outputs (including sides or rears) rather than only 3.1 (left, center, right and sub). Logic7 builds the center, sides and rears from stereo, so for DVDs pick the two-channel downmix. Home 5.1 assumes a TV screen, dialogue, and seating that forms a near-equilateral triangle with the speakers, and none of that fits a car. His point is that upmixing is not just for movies.

1.12 #5 · When to switch Logic7 on.

No center and no rears: off ("it does nothing for you"). Rears but no center: try it. Center but no rears: add rears. In MS-8 the center plays only with Logic7 on, so if you dislike rear steering, leave it on and fade forward. [Era: MS-8] Sources: DIYMA MS-8 FAQ p2 · DIYMA MS-8 FAQ p3

In plain English: This is MS-8-era advice on when to switch on its Logic7 upmixer. No center and no rears: leave it off ("it does nothing for you"). Rears but no center: try it. A center but no rears: add rears. In the MS-8 the center speaker plays only when Logic7 is on, so if you dislike sound being steered to the rears, leave Logic7 on and use the fader to move the sound forward.

1.12 #6 · Matrix upmixers for music: from "the ticket" to poor.

MS-8 era: for 2-channel music a matrix processor "is the ticket", more believable than Audyssey's "more clinical" sound. Later: PL, PLII and Logic7 suit video but steer mixers' tricks rearward in music; music-oriented upmixers (Circle Surround, Harman QLS, Audiofrog Multiseat) extract common content instead. [RARE] [Changed view: 2008–10 → 2017 onward] Sources: DIYMA MS-8 FAQ #10 · DIYMA 2017-10-12 · D&F live @35:06

In plain English: His view of matrix upmixers for music changed over time, which the [Changed view] tag marks. In the MS-8 era (2008–10) he said that for two-channel music a matrix processor "is the ticket", more believable than the "more clinical" sound of Audyssey (the technology in Alpine's Imprint processors then). From 2017 on, he said PL, PLII and Logic7 suit video but, with music, steer mixing engineers' out-of-phase tricks to the rear. He contrasted them with music-oriented upmixers (Circle Surround, Harman QLS, Audiofrog Multiseat), which instead pull out the content common to both channels.

1.12 #7 · What a music upmixer should do.

Use clues in the recording to make it more believable without altering it, only spreading it across channels; the one standard he knows (an ITU broadcast rule) requires the upmix to fold back to stereo, which matrix processors can't. A good one matches or beats stereo in more than one seat but rarely equals stereo in its one ideal seat. [RARE] Sources: D&F live @41:05 · DIYMA 2017-10-12 · EDUCAR

In plain English: He sets out the job description. A music upmixer should use clues already in the recording to make it more believable, without altering the recording, only spreading it across more channels. The one standard he knows of, a broadcasting rule from the ITU (International Telecommunication Union), requires that the upmix fold back to the original stereo if you mix all its channels together again. Matrix processors can't do that. In his view a good upmixer matches or beats stereo in more than one seat, but rarely equals stereo in stereo's one ideal seat.

1.12 #8 · Audiofrog Multiseat Stereo splits rather than steers.

Built on Penteo (a studio upmixer) and reworked for cars, it splits stereo into mono (to the center), similar-but-not-identical L/R, and L-only/R-only content, spreading the last two over L/R, sides and rears "like a horseshoe". With no phase-based steering there's no coloration, bass stays punchy, and every seat works without L/R delays. [RARE] Sources: DIYMA 2017-11-13 · DIYMA 2018-04-06 · Mobile Electronics Aug 2016

In plain English: Audiofrog is the company Andy co-founded, and Multiseat Stereo is its upmixer. It is built on Penteo (an upmixer made for recording studios) and reworked for cars. Instead of steering whole channels, it splits stereo into three parts: mono content (identical in both channels), sent to the center; content that is similar but not identical in left and right; and content found only in the left or only in the right. The last two are spread over the left and right fronts, sides and rears "like a horseshoe". He says that because no phase-based steering is used, there's no coloration (no change to the tonal balance), bass stays punchy, and every seat works without left/right delays.

1.12 #9 · He has lived with upmixers for 20+ years.

His cars have run Logic7, PLII, DTS Neural, a custom room-synthesis algorithm and Audiofrog's upmixer; Audiofrog's CES 2016 car (DTS Neural plus a real center) so that "the driver's seat sounds just like the passenger seat". At home his JBL M2 monitors play multichannel 95% of the time. [RARE] Sources: CES 2016 @3:03 · DIYMA 2015-08-09 · DIYMA 2021-09-08

In plain English: This point is about his own experience. Over more than 20 years his cars have run Logic7, PLII, DTS Neural, a custom room-synthesis algorithm (processing that creates the sense of a room) and Audiofrog's own upmixer. Audiofrog's demo car at CES 2016 (a big electronics trade show) used DTS Neural plus a real center so that "the driver's seat sounds just like the passenger seat". At home, his JBL M2 studio monitors play multichannel 95% of the time.

1.12 #10 · Factory upmixers (e.g. Harman QLS): use them instead of ripping them out.

QLS (Hyundai/Kia, Maserati, Lincoln) separates center, left, right and in-between content with about seven filters, and its outputs may not sum back to stereo. Wire each factory output to a matching amp channel, add power and better speakers, then just re-EQ and set levels; the delays and crossovers are already there. [RARE] Sources: EDUCAR · Absolute Electronix 2023 @58:34 · Absolute Electronix 2023 @56:27

In plain English: Many installers believe the first step in upgrading a premium factory system is to rip out or bypass its processing. For factory upmixers, he says the opposite. Harman's QLS (found in Hyundai/Kia, Maserati and Lincoln systems) separates center, left, right and in-between content with about seven filters, and its outputs may not add back up to plain stereo. So connect each factory output to its own matching amplifier channel, add power and better speakers, then simply re-EQ and set levels. The delays and crossovers are already in place.

1.12 #11 · Upmixed cars are harder to tune: L+R must sum like center plus L−R.

Harman's factory tuners use mic arrays and in-house software to meet that. [RARE] Sources: DIYMA 2012-11-02

In plain English: He says a car with an upmixer is harder to tune than a plain stereo car. The combined sound of the left and right speakers (L+R) has to match closely the combined sound of the center (the mono part) plus the L−R (difference) part. Harman's factory tuners use arrays of several microphones and in-house software to meet that. The source states the requirement without spelling out the reasoning further.

1.12 #12 · His two-seat (and back-seat) process with an upmixer (2016).

Front delays to a point midway between the front headrests; crossovers by dispersion and safety; measure each side plus sub in both front seats (six mic spots on a 7" circle), average and auto-EQ; repeat for center and surrounds; delay sides to arrive with the fronts at the rear seats; add 12–20 ms to the rears; set the upmixer for the widest stage with a stable center. [RARE] Sources: DIYMA 2016-01-16

In plain English: This is his 2016 process for tuning two front seats, plus the back seat, with an upmixer. Step by step: (1) Set the front delays to a point midway between the two front headrests, so neither seat is favored. (2) Choose crossovers by dispersion (how widely each speaker spreads its sound) and by safety (keeping each driver away from notes that could damage it). (3) Measure each side plus the sub from both front seats, at six mic spots on a 7-inch (about 18 cm) circle, average the results and let the auto-EQ correct them. (4) Repeat for the center and the surround speakers. (5) Delay the side speakers so their sound reaches the rear seats together with the fronts. (6) Add 12–20 ms (thousandths of a second) of delay to the rears. (7) Set the upmixer for the widest stage that still keeps a stable center.

1.12 #13 · Room synthesis fooled the judges.

At his last IASCA/MECA show judges said they'd never heard a car reproduce a recording's room so accurately; his DSP generated it. He tolerates "HUGE errors in frequency response" when spatial qualities are good. [RARE] Sources: DIYMA MS-8 FAQ p4 #70 · DIYMA MS-8 FAQ p5 #99

In plain English: IASCA and MECA are car-audio competitions. At his last such show, he says, judges told him they had never heard a car reproduce a recording's room (the sense of the space it was recorded in) so accurately. In fact his processor was generating that room, not just playing it back. He says he tolerates "HUGE errors in frequency response" (unevenness in tonal balance) as long as the spatial qualities (width, depth, sense of space) are good.

1.12 #14 · Aftermarket DSPs lack upmixers for business reasons, not demand.

Licenses are cheap, but a product manager must understand upmixing and win it in a meeting that includes finance; many are steeped in 2-channel audio or judged it from bad setups. Nearly every new premium factory system has one. [RARE] Sources: DIYMA 2017-07-07 · DIYMA 2021-08-01

In plain English: Aftermarket DSPs are the add-on sound processors sold to car-audio enthusiasts. Most don't include an upmixer, and he says that is for business reasons, not lack of demand. Licenses for the technology are cheap, but a product manager must understand upmixing and win the argument for it in a meeting that includes finance. Many product managers are steeped in two-channel audio or judged upmixing from badly set-up systems. Meanwhile, he notes, nearly every new premium factory system has an upmixer.

1.13 Believability over accuracy

What this part is about: This part sums up his philosophy. Because a recording is not a true copy of a real event and a car is not a concert hall, he aims for sound you can believe in rather than "accuracy", but built on getting the basics right first.

1.13 #1 · "Think of it as believability rather than accuracy."

The guitar never travels through the circuit board, only a signal, which must be changed to make up for the acoustic system; the reward for choosing believability "might be greater than you expect". [RARE] Sources: DIYMA MS-8 FAQ #13 · DIYMA Apr 2010

In plain English: Many people believe the goal of a good system is "accuracy": passing the recorded signal through untouched. He suggests thinking of it as believability instead. His reasoning: the guitar itself never travels through the circuit board, only an electrical signal that stands for it. That signal has to be changed (equalized, delayed and so on) to make up for the acoustic system, meaning the speakers and the car interior they play into. He says the reward for choosing believability "might be greater than you expect".

1.13 #2 · There's no accurate reference to be accurate to.

Most records are many events sped up, pitch-shifted, compressed and given a synthesized room; a studio disc is a "deep fake" of an event, so aim for suspension of disbelief. Producers want the studio experience, not "flat": with his DSP off they asked why their tracks sounded "mono"; with it on, "it needs more bass". [RARE] Sources: DIYMA MS-8 FAQ p5 #83 · DIYMA MS-8 FAQ p4 #70 · CMA Networks 2023 @35:05

In plain English: He argues that most records have no real original to be accurate to. They are built from many separate recordings that have been sped up, pitch-shifted, compressed (their loud and quiet parts squeezed closer together) and given a synthesized, artificial room. He calls a studio disc a "deep fake" of an event, so the aim is suspension of disbelief: letting you accept the illusion. He adds that producers want to hear the studio experience, not "flat" (equal loudness at every frequency). When he played producers their tracks with his processor off, they asked why the tracks sounded "mono"; with it on, they said "it needs more bass".

1.13 #3 · Two-channel sounds like a diorama on the dash; multichannel sounds like a space bigger than the car, for two people at once.

He points to Apple spatial audio and Dolby Atmos, and to a 120-plus-speaker L-ISA concert he heard as "a comforter of sound": a non-stereo presentation can be beautiful. Aftermarket listeners judging factory premium systems only on bass and loudness miss exactly this. [RARE] Sources: Absolute Electronix 2023 @55:28 · CMA Connected Oct 2021 @48:18 · CMA Networks 2023 @39:03

In plain English: A diorama is a small model scene. He says plain two-channel stereo in a car sounds like a diorama sitting on the dash, while multichannel sound can feel like a space bigger than the car, and for two people at once. He points to Apple spatial audio and Dolby Atmos (modern surround formats), and to a concert using L-ISA (a multi-speaker concert sound system) with more than 120 speakers, which he heard as "a comforter of sound", like a thick quilt wrapped around you. His point is that a non-stereo presentation can be beautiful. Aftermarket listeners who judge premium factory systems only on bass and loudness miss exactly this.

1.13 #4 · Believability rests on the basics, not instead of them.

He values width most, then ambience; a peaky midrange is forgivable, a thick midbass is his pet peeve. Experienced listeners judge a demo car in seconds: full, impactful bass from the front, a centered singer, then width and depth. For customers, first "get close to correct". Sources: OFCAP Ep. 19 @3:31:01 · Absolute Electronix 2023 @53:21 · OFCAP Ep. 38 @54:01

In plain English: He stresses that believability is built on the basics, not a replacement for them. Width matters most to him, then ambience (the sense of the recording's space). He can forgive a peaky midrange (some middle notes sticking out), but a thick midbass (too much low-middle sound) is his pet peeve. He says experienced listeners judge a demo car within seconds: first full, impactful bass that seems to come from the front, then a centered singer, then width and depth. So in practice, for customers, the first job is to "get close to correct".

1.14 Top rarely-discussed points in this section

Door speakers sit 180° out of phase near 250 Hz in both front seats, right in the vocal band (1.1 #3, 1.9 #4).
Delay never moves a speaker; center an off-center image with level, never with more delay (1.1 #7, 1.1 #8).
Set delays to the midrange's distance, and let crossover phase, not pillar mounting, set vocal height (1.1 #6, 1.1 #12, 1.7 #7).
Above the midbass, left plus right sums to only about +3 dB in a car; leave anything within ±90° alone (1.1 #13).
A car has no reverberant field: you hear the power response, and no EQ or delay fixes a reflection comb (1.3 #1, 1.3 #4).
Predict reflection nulls with a tape measure; push mids into corners and move tweeters away from surfaces (1.3 #5, 1.6 #8).
Dash mats and foam do nothing in the midrange; ~200 Hz would need ~1.5 ft of absorber (1.3 #8).
Dispersion is diameter versus wavelength, and his circumference ÷ wavelength rule decides where each driver can go (1.4 #1, 1.4 #2).
A 6" + 1" 2-way has an off-axis hole that aiming only moves, which is why cars need 3-ways when studios don't (1.4 #4, 1.4 #6).
Aim only the tweeter; door midbass aiming changes "zip" (1.5 #1, 1.5 #2).
A small mid high in the door beats A-pillar pods, which comb the whole midrange and are one-seat-only without a center (1.6 #2, 1.6 #6, 1.9 #3).
"HRTF height EQ" works only while you're adjusting it (1.2 #5).
Images can't be made reliably outside the speaker triangle; "width" from delay games is comb filtering (1.7 #1, 1.7 #2).
Two-seat stereo without a center: an all-pass near 250 Hz instead of left/right delay, as Toyota, Subaru and Bose already do (1.9 #4, 1.9 #5).
Two listening tests reveal whether a factory system uses all-pass filters or delay (1.9 #6).
L+R is not a center; real extraction leaves center content in L/R about 6 dB down (1.10 #1, 1.10 #2).
A dash center's ~2 kHz windshield dip is a feature that mimics a phantom center (1.10 #5).
Keep a factory upmixer and go one-to-one instead of "getting back to flat" (1.12 #10).

Changed views and conflicts in this section

Windshield corner: 2012 "keep speakers away from the junction" → 2013/2021 "tight in the corner is best; a pod hanging nearby is the problem" (1.6 #8).
Firing mids into the glass: did it himself in 2009 → from 2011 it narrows the stage, and in Sep 2021 up-firing mids make the densest comb; conflict with his Oct 2021–Jan 2022 praise of windshield-aimed factory dash "middlers" (1.6 #9, 1.6 #23).
Tweeterless widebanders: 2012–14 "must be on-axis" → 2018–21 "aim at the windshield; on-axis sounds clinical" (1.8 #4).
Phase EQ for two seats: 2013–15 inaudible next to delay → Jan 2016 "drop L/R delay for phase EQ" → 2021 helps "only with fiddling", so he won't teach it (1.9 #4, 1.9 #10).
Rear fill: 2017 differential (L−R) recipe → 2020s plain in-phase stereo, high-passed and delayed (1.11 #5).
Matrix upmixers for music: 2008–10 "the ticket" → from 2017 poor for music; music-oriented upmixers instead (1.12 #6).
Stage height: 2016 "from physical location" → 2020–21 adds that crossover phase holds everything at tweeter height (1.7 #7).
Conflict, A-pillars: "a HORRIBLE place for a midrange" versus Audiofrog copy calling GB10+GB25+GB60 "the ultimate active system for A-pillars" and a GB25 designed for pillar pods (1.6 #6, 1.6 #7).
Conflict, crossfired pillar tweeters: his "band-aid" verdict versus a 2015 Harman/JBL/Infinity tip sheet that recommends it and the Bose Fiat 500 he praised (1.5 #6).
Conflict, Harman patent: he recalls US6876748 naming 250 Hz; its text cites about 200 Hz (1.9 #5).
Unchanged: "HRTF EQ" can't raise the stage, from 2008 through 2021 (1.2 #5).

The full series

Download the complete guide as one PDF (351 pages, every point with its explanation).