As I have written some of the Nyquist code / plug-ins related to “hidden messages”, I should probably chip in at this point.
For me, “silent subliminals” are a curious form of steganography (hidden messages). Just like children frequently have fun inventing “code writing”, so other forms of steganography can be fun and intriguing to play with (a curious example by Aphex Twin: Hey, Who's That Face in My Song? | WIRED).
However, I also tend to agree with Wikipedia that “subliminal messaging” has little or no effect.
A much more effective way to remember something is through repetition - sometimes called “learning by rote”.
Learning by rote is not good for developing understanding, but it is still considered one of the most effective ways to memorize and recall basic facts.
Sub-sonic frequencies are not very suitable for carrying messages. In simple terms, a 1 Hz frequency carries 1 piece of information per second, so it will take a long time to transmit a meaningful message. The higher the carrier frequency, the more information it can carry per second - which is why optical broadband networks run at extremely high speeds.
About the simplest way to embed a “hidden message” in audio, is to modulate a high frequency carrier tone.
In it’s simplest form, this can be done by applying this code to a mono track via the Nyquist Prompt effect:
;version 4
(mult (hzosc 14000) *track*)
After applying that code to a mono voice recording, the sound becomes a high pitched squeak / swishing sound.
To decode the track, apply the same code again.
For a more sophisticated version, and a long discussion about “silent subliminals”, read this topic: Silent Subliminals [solved?]