The Society went with the "backwards masking" rumors back in the '80s:
*** w83 10/15 p. 11 par. 8 Beware of Music That Debases! ***
There have also been recordings of unscriptural and even demonistic messages by means of a technique called backward masking, used by a number of musical groups. When the recording of one very popular song is played backward, it repeatedly says, "Decide to smoke marijuana." If played backward, another popular recording carries this message: "I will sing because I live with Satan. . . . There’s no escaping it, my sweet Satan."
The first song is Queen's "Another One Bites the Dust" (the actual misheard line is "It's fun to smoke marijuana"). The second is of course Led Zeppelin's "Stairway to Heaven". Here is a Youtube video of the Queen bit:
http://www.youtube.com/watch?v=xj2dJsVpgCY
In the middle of the Queen song, the refrain "another one bites the dust" is repeated over and over. Because the words are repeated quickly over and over, phonotactic simplification occurs. "Another one bits the dust" articulated carefully would be [an ∂ th ∂ wan baytsð ∂ dυst], but what you hear is more like [an ∂ Γ w am b ¦ays t∂ dυst]; "another one" is more like "an'oer one" (with a flapped "r" sound), the alveolar [n] of "one" assimilates to bilabial [m] before the [b] of "bites" (this seems to vary a bit between the four repetitions), etc. Also there is a drum machine sound occurring simultaneously with the word "bites", which is going to strongly color how that word sounds backwards (the onset, i.e. ¦, of the drum sound occurs at the same time as the "b" and the rest of the sound overwhelms the vowel, hence the strike-through). And if you run that backwards, you get what sounds like [t sud¬ ∂ ht∂ s w≈ ¬ maΓ w ∂na]. "Dust" gives [t sud¬ ] which has a sudden hard offset (¬ ), and that is misheard as "fun"; [t s] is a fricative sound like "f", the vowel is the same as "fun", and [d¬ ] doesn't sound like a "d" so it is heard as an "n". Then "the" plus a little bit of the trailing off drum sound gives something like [∂ ht∂]; the drum sound gives the impression of a vowel at the end where none exists, so that is heard as "to", thus we so far have: "fun to". Now comes "bites", and the backwards drum sound has completely drowned out the actual word and strongly colors what you hear backwards. The drum sound normally has a strong sudden onset and a loud reverberation (≈) that, when mixed with the vowel of "bite", sounds more like the "o" of "smoke". Also, there is a sweeping quality to the reserved drum sound (if you've ever heard drums backwards you'd recognize that sweeping sound) that has a hint of a "w". And the sudden onset of the drum sound, drowning out the "b", is heard as a hard sudden offset, which gives the impression of an unreleased "k", e.g. say "kick" without releasing the final "k"). So [s w≈ ¬ ] comes out sounding rather like [s wo k], or "smoke". So now the mind trying to make sense of these nonsense sounds may hear "fun to smoke...". And finally the first phrase [an ∂ Γ w am] "ano'er one" reversed simply gives [maΓ w ∂na] (it should be noted that [Γ w ] is not a sequence of sounds but is more like a single sound, the lips round to produce the "w" even during the [Γ]), which sounds almost exactly like "marijuana", or rather "mar'juana". So "another one bites the dust" yields what some may interpret to be "fun to smoke marijuana".
One can just as easily do the same kind of analysis with "Stairway to Heaven". Claims of backwards messages in audio that is clearly not intentionally reversed in production (like the intentional backwards message in Pink Floyd's "Empty Spaces", which does sound like gibberish when played normally) are usually nothing more than aural pareidolia.