Stereo low end is a translation decision

The bass that breathes in mono

You layer a sub under a synth bass, open the unison and spread the voices for a wide, warm low end. On headphones it is huge. Then the track plays on a single Bluetooth speaker and the bass swells and fades on its own, loud on one bar, nearly gone on the next, although every note was programmed at the same velocity.

A single speaker plays left plus right. The detuned voices drift in and out of step between your two channels, so their sum rises and falls from bar to bar.

Why it matters

Low end gets summed in many places

The usual advice is to keep everything under about 120 Hz in mono, and the lesson on mono explains why some width tricks fail there. That advice is about translation, so it helps to know exactly what it protects against.

Plenty of playback adds the channels in the low end, even when it plays the rest in stereo. Single-speaker phones and smart speakers sum everything. A 2.1 system or a home cinema with bass management sends the lows of both channels to one sub. Many club rigs feed their subs a mono sum. Vinyl does it mechanically: in the 45/45 stereo groove that Blumlein's patent described, the sum of the channels moves the stylus sideways and the difference moves it up and down, so large out-of-phase bass makes for a deep, hard-to-track groove. Each of these hears $L + R = 2M$, twice the mid signal, and anything that exists only in the side signal is not there.

The science

Four kinds of stereo bass

Whether stereo bass survives depends on what differs between the channels.

A level difference is safe. If the right channel carries the same waveform as the left at a lower level, the sum only adds. A bass panned off-centre never cancels.

A small time difference is nearly safe in the sub. The phase shift is $360^\circ \times f \times \Delta t$, so 1 ms between the channels is only 20 degrees at 55 Hz, and the sum loses about 0.1 dB. The same 1 ms is 180 degrees at 500 Hz, which is why time offsets hurt the harmonics and the click long before the sub.

A detune is not safe. Unison voices, a chorus, or a different oscillator in each channel drift in and out of step. Two equal voices cancel completely in mono once every $1/\Delta f$ seconds, where $\Delta f$ is the frequency difference. With more voices the dips are less regular, but they are still there. That is the breathing in the hook.

Anti-phase content is the worst case. A side boost, a polarity flip on one channel or a phase-rotating widener puts some or all of the bass in opposite polarity on each side, and that part cancels in any sum, as the mid/side lesson works out.

Mono bass is often justified by saying the ear cannot locate low frequencies. That is only partly true. For sounds that contain low frequencies, the timing difference between the two ears is the cue listeners follow for direction; with the low frequencies removed, level and pinna cues take over (Wightman and Kistler, 1992). The ear can use those timing cues below roughly 1.5 kHz (Moore, 2012), which is where the harmonics of a bass sit. In the deepest octave, a small room adds its own pattern on top: the room modes mix the output of both speakers, so a stereo difference down there tends to show up as a level change at your chair.

That gives a practical split. The sub carries the weight and most of the risk. The harmonics carry the character, the ear can place them, and they cost far less in mono. With a detune measured in cents, the nth harmonic beats n times as fast as the fundamental, so the harmonics drift in and out of step at different moments instead of all vanishing together.

The demo plays two copies of one bass note. Adding them is exactly what a mono fold does to a bass whose channels differ by a delay or a flip, so listen for how fast the low end thins out.