Invert effect is audible?

I didn’t think the Invert effect would make an audible difference, but for the first time today I noticed that it did, in some speech. Nasals like m sound a bit boomier and otherwise it is hard to describe.

If one track of a dual-mono pair is inverted it sounds different.
On headphones the invert-effect is just spatial,
via loudspeakers it could have an effect on frequency-content because of interference.

Good point, but this was just plain mono.

Inverting a sine wave shouldn’t make a difference, that is just a phase shift. Inverting something more complicated does sound different, and I am a bit surprised.

How were you listening? Headphones? Ear buds? Computer speakers? Studio monitor speakers?


I could imagine that plosives might sound different inverted when listening through headphones, but I’m surprised that “nasals like m” should sound different. However, I’ve tried it myself and I do agree that there is a clearly audible difference.

The attached is one cycle from an “Mmm” sound, repeated 100 times, then inverted. Through speakers there is an audible click each time it changes, but not much difference that I can hear, whereas through my headphones there is a distinct difference in timbre.

A similar difference can be heard (in headphones) when inverting a sawtooth waveform.

Here’s a low-pass filtered sawtooth waveform. The only difference between the two tones is that one is inverted.

To ensure that was the case, I generated one waveform, then filtered it, then used Normalize to ensure no DC offset, then duplicated it to make an exact copy, then inverted the copy.

Even if you just play “filtered-saw.wav” in one earphone the inversion produces an audible difference : the ratio of the harmonics change.

Maybe the transducers in the earphones have an asymmetry* : so they have different resonant properties when pulled than when pushed.

[ * diaphragms not flat ? …

Headphones - Wikipedia ]

Quite possibly the diaphragm reproducing the sound might be affected by its asymmetries. Could the same be true of the human eardrum?

Some waveforms have much taller and narrower excursions on one side and wide and short ones opposite, still with no DC offset, so that it is only the positive or the negative side that affects peak amplitude while rms is very different. This happens much in speech. It could be too that highpass filtering might shift phases of some passed components, changing the degree of this asymmetry.

My guess is that it is due to asymmetrical response of ears.

Some drums have one skin that you hit and are open at the other end. Other drums have two skins, one that you hit and one closing, or partially closing the other end. The second skin has a noticeable effect on the resonance of the drum, and skilled drummers will tune both heads (drum skins) to create the timbre that they want.

“Ear drums” are “designed” to work as single headed drums, but placing a speaker in very close proximity to the pinna (external ear) will act like a second drum skin. If I move the headphone away from my ear, I find that the effect reduces rapidly with distance, becoming inaudible at around 30 cm distance.

Whatever the actual reason, it’s an interesting phenomenon.