LESSON 20 · SPEAKERS

Why does a speaker sound different across the couch?

Move away from the middle seat and the sound can change even though the volume knob hasn't moved. Speakers send different amounts of sound in different directions, and the difference can change with pitch.

On axis (0°)
Listening straight ahead from the speaker's stated reference direction, usually near its tweeter or acoustic axis.
Off axis
Listening at an angle away from that forward direction, either sideways or above and below.
Directivity
How strongly a speaker concentrates sound in certain directions instead of spreading it everywhere.
FICTIONAL HORIZONTAL RESPONSE SPECIFICATION8 kHz · 30° horizontal: −6 dB relative to 0°

One frequency, one angle, same measurement distance and signal. See exactly what this rating means ↓

Start with the measurement angle

Microphone positions for a horizontal speaker measurement Viewed from above, a speaker faces right. The on-axis microphone sits straight ahead. The off-axis microphone sits 30 degrees to the side, at the same height and distance from the speaker. Top view · horizontal plane Speaker 30° 0° on axis 30° off axis Samedistance
Where the measurement is taken. The two positions show where one microphone would be relative to the speaker at 0° and 30°. The angle changes while height, distance and input signal stay the same. A vertical measurement changes the angle above or below the reference axis. At home, toe-in changes the horizontal angle between each speaker and your seat.

DECODE THE EXACT STAT

What does 30° horizontal, −6 dB at 8 kHz mean?

In this example, 8 kHz sound is 6 dB lower at 30° sideways than directly in front.

Imagine measuring one speaker at the same distance in two directions while feeding it the same test signal. Straight in front is called 0°. Move the measurement microphone 30° sideways, and in this made-up example the 8 kHz sound pressure level drops by 6 dB.

A 6 dB drop in sound pressure level is about half the pressure amplitude. That's what the two bars show. It is not a claim that music suddenly sounds half as loud; perceived loudness is more complicated.

What this number cannot tell you: what happens at 1 kHz, whether the speaker is balanced on axis, how the vertical response behaves, how the room reflects sound, or how wide the best listening area is overall.

At 8 kHz · same distance and input signal

0° directly in front0 dB difference

100% normalized pressure amplitude · our reference

30° to the side−6 dB

About 50% of the reference pressure amplitude at 8 kHz

The bars show relative sound pressure at this one frequency, not perceived loudness, total music volume, full speaker frequency response or sound power.

The two bars plot the exact fictional rating. Normalized pressure = 10(level difference / 20), so 10−6/20 ≈ 0.501. No real speaker was measured.

SEE WHAT CHANGES WITH DIRECTION

A speaker doesn't spread every frequency equally.

Switch the angle. Each synthetic curve shows how much weaker that direction is than the speaker's own on-axis response at the same frequency. The scale always stays at 0 to −18 dB. The 30° curve includes the exact −6 dB at 8 kHz from the opening example.

30° to the side

Our exact example is marked at 8 kHz: −6 dB compared with the speaker’s own 0° response. Different frequencies spread differently; this is why off-axis tonal balance can change.

0 dB-6 dB-12 dB-18 dB 500 Hz1 kHz2 kHz4 kHz8 kHz16 kHz Frequency (octave spacing) · vertical axis = dB relative to 0° at each frequency
8 kHz difference-6 dB16 kHz difference-10 dB

Original imaginary responses, not a manufacturer's measurement. The 0 dB reference at each frequency means “same as on axis there.” It does not mean the speaker's on-axis frequency response is perfectly flat.

Read two real speakers on the same graph

Compare the two pairs in the same positions, at the same distance and with their on-axis levels aligned. The solid lines show what reaches you straight ahead. The dashed lines show the direct sound if you turn each speaker so your seat is 30° from its axis.

KEF LS50 and Genelec 8030A · measured horizontal response

Level-matched KEF LS50 and Genelec 8030A responses at 0° and 30° Both on-axis traces are aligned at 1 kHz. At 5 kHz the KEF is about 3 dB above the Genelec on axis and the two are nearly equal at 30°. At 10 kHz they are nearly equal on axis and the KEF is about 2 dB above the Genelec at 30°. Level relative to each speaker’s 0° level at 1 kHz (dB) +4 +2 0 −2 −4 −6 500 1k 2k 5k 10k 20k Frequency (Hz · log scale)
  • Cyan · KEF LS50
  • Amber · Genelec 8030A
  • Solid · 0° on axis
  • Dashed · 30° off axis
Digitized from Princeton University’s 3D3A Lab charts for the KEF LS50 and Genelec 8030A, CC BY 4.0. We shifted both curves for each speaker by the same amount to align their on-axis levels at 1 kHz. The gaps within each pair are preserved. Readings are approximate.

What changes when you turn the speakers?

Start with the voice and cymbals. Follow one trace from left to right to see tonal balance. At any one frequency, the vertical gap between a speaker’s solid and dashed lines shows what changes with angle. To compare the two models, first align their on-axis levels at a reference frequency. The examples below use 1 kHz and apply the same level adjustment to both traces for each speaker.

Facing the speakers: Around 5 kHz the LS50 has roughly 3 dB more output relative to its 1 kHz level than the 8030A. That can make consonants and the initial edge of a cymbal hit stand out more on the KEF when the recording has energy there. A more prominent edge can sound like extra detail. The graph shows a difference in emphasis.

At 30°: The LS50 loses about 4 dB around 5 kHz compared with its own straight-ahead response. The 8030A loses about 1 dB there. After the 1 kHz alignment, the two speakers are close to the same level at 5 kHz at this angle. At 10 kHz the situation changes. The KEF’s traces nearly meet, while the Genelec’s off-axis trace is about 2 dB below its on-axis trace. At 30° the KEF is roughly 2 dB ahead there after alignment. So a listener may hear the KEF’s consonant edge soften as the speakers turn away, while a little more of the upper-treble shimmer remains audible compared with the Genelec. The next section uses a fixed room and seat to compare what comes back from the sidewalls.

How the manufacturer’s ±3 dB spec fits in

KEF lists the original LS50 at 79 Hz–28 kHz (±3 dB) in its manufacturer brochure. The tolerance describes a 6 dB-high window around the spec’s reference level over the stated range. The independent plot here spans 500 Hz–20 kHz, so the brochure supplies the wider frequency limits. The vertical gap at 5 kHz compares two angles at one frequency.

How much of the couch falls within the coverage?

Wider coverage

The cyan coverage areas of both speakers span all three marked seats.

Narrower coverage

The center seat falls within both cyan coverage areas. The outer seats sit beyond both drawn boundaries.
SPK-01 · Coverage at one illustrative frequency (8 kHz). The room, aiming and on-axis level are matched. Dotted lines mark the speaker axes. The cyan areas show wider and narrower coverage contours. Output changes gradually across the contour, and the shaded area does not encode level. This example compares direct sound only.

Hear the difference in a 12 × 15 ft room

Keep the same seat and swap the two pairs into the same spots. The speakers are 6 ft apart and 2 ft from the front wall, aimed at a seat centered 9.71 ft behind the speaker line. Both sidewalls are painted drywall. The direct sound comes from the 0° axis. The first bounce leaves each speaker at 60°, so this example uses the 60° traces from Princeton’s original charts for the KEF and Genelec.

The direct sound and the first sidewall bounce

Direct and first sidewall paths in a 12 by 15 foot room Two speakers are six feet apart and aimed at the centered seat. The solid paths reach the seat directly. The dashed paths leave each speaker at 60 degrees from its forward axis and reflect from the nearest sidewall. 12 ft15 ft6 ft apartCentered seat Wallbounce
  • Solid · direct sound
  • Dashed · first sidewall bounce
The direct path is 10.16 ft and the bounce travels 13.24 ft, arriving about 2.74 ms later at 1,125 ft/s. This is a plan view with the speaker and ear heights matched.

Listen for the voice first, then the cymbals

A centered voice around 4 kHz. With the same 1 kHz alignment, the KEF’s direct sound is about 2–3 dB stronger here, while the Genelec’s first sidewall bounce is about 2 dB stronger. The KEF may give the edges of consonants more bite straight ahead. The Genelec sends more of that part of the voice back from the sides, so the consonants could feel a little less tightly gathered around the center.

Cymbal shimmer around 10 kHz. The direct levels are close, and the KEF’s first bounce is only about 1 dB stronger in this layout. Expect a smaller difference here than in the vocal example. With the left and right paths symmetric, a centered vocal can stay centered on both pairs even if its edges feel a little wider on one.

How the room estimate was calculated

The comparison models one first bounce per speaker. We assume 3 dB of loss at each wall as an illustrative value, plus about 2.3 dB from the longer path. Around 4 kHz, that puts each first bounce approximately 17 dB below the KEF’s direct sound and 12 dB below the Genelec’s. These are separate path levels estimated from digitized measurements. The example leaves out phase summation and other reflections, so the listening impressions are hypotheses for this layout.

Explore how the angle can change the tone of a reflection

Off-axis sound can change tone as well as level

Synthetic off-axis frequency responses at the same angle dB relative to on-axis0-3-6-9-122505001k2k4k8k16kFrequency (Hz)
  • Solid cyan · smooth tilt
  • Dashed lilac · added notch
SPK-02 · Synthetic responses at the same horizontal angle, 30°. Both traces are relative to each speaker’s own on-axis level at each frequency. The examples match at 8 kHz to the displayed precision. Around 4 kHz, one radiates less energy at the chosen angle. A surface in that direction can receive a different tonal balance even when a single coverage rating looks similar. The plot shows speaker output at one angle. A room measurement would also capture the surface, distance and other paths.
What are “listening window,” “sound power” and “spinorama”? Optional detail

A formal off-axis test can measure many angles in both horizontal and vertical planes. ANSI/CTA-2034, commonly called a spinorama, groups these measurements into summaries. On-axis is one reference direction. The listening window is an average of nearby response directions, including angles out to ±30° horizontally and ±10° vertically. Early reflections estimates first-bounce reflected energy using a defined angle-weighted procedure. Sound power estimates radiated acoustic output averaged over directions.

Directivity index compares the level of a forward/listening-window response with a wider-angle response. Its trend can tell you how radiation becomes more focused with frequency. A smooth trend is often more informative than a single advertised “dispersion angle.” These are standardized measurement summaries, not a guarantee of a particular room's exact response.

Read more about soundstage, imaging and separation in the glossary.

ANOTHER DIRECTION TO THINK ABOUT

Sideways isn't the same as above or below.

A published horizontal plot tells you what happens when you move left or right. A vertical plot describes moving above or below the speaker's reference axis, such as standing up from the sofa or sitting lower than the tweeter.

A speaker often has a woofer for lower notes and a tweeter for higher ones. Around the notes where both contribute, their sound can add together or partly cancel depending on whether your ears are higher or lower. Engineers call this pattern vertical lobing. It helps explain why standing up can change the tone even if you stay the same distance away.

Practical example:

A speaker may have smooth horizontal coverage across your couch while exhibiting a narrow vertical sweet spot near the crossover. Another design may behave differently. A horizontal “±30°” figure says nothing reliable about its vertical coverage.

When reviewing speaker measurements, look for both directions, especially if your speakers sit high, low, on a shelf, or cover people at different seated and standing heights.

What should you check when choosing or positioning speakers?

  1. Look for on-axis and off-axis frequency response. A single “frequency range” or “120° coverage” label is far less informative than curves across multiple angles.
  2. Think about seats. For a wide couch, check response around the angles your seats make relative to each speaker, rather than assuming wider published coverage is always an advantage.
  3. Check horizontal and vertical behavior separately. Aim the acoustic reference axis at a sensible listening height and look for troublesome crossover-region dips away from it.
  4. Try positioning before replacing equipment. Small adjustments to speaker height, spacing, toe-in and distance from reflective walls can change the direct and reflected balance. Keep left/right positioning reasonably symmetrical.
  5. Compare measurements from known methods. CTA-2034-style data, directivity maps and measured listening windows can reveal much more than a marketing adjective.
How do I interpret a single “−6 dB beamwidth” number? Optional detail

For a named frequency and measurement plane, a −6 dB coverage or beamwidth can mean the angular width between directions where sound pressure has dropped 6 dB relative to the relevant reference or maximum. Check whether the published value means full included beamwidth, or an angle from the centre line, and whether it refers to horizontal or vertical coverage. The example here specifies one offset angle: 30° horizontally. It does not assert that the full beamwidth is 30° or 60°.

Sources and further reading

Integrity note: The angle selector and smooth-versus-uneven examples use illustrative curves. The LS50 and 8030A overlay is digitized from published anechoic measurements.