A security log entry by Schneier on eavesdropping compelled me to write a haiku:

Compress your bitrate
And expose the key to sound;
VoIP flows insecure.

I really like the attack vector he points us to. In short, when you compress voice on phone systems it creates a predictable key of sounds that can be used to unlock the encryption. In other words, sounds have patterns that the encryption does not hide. Even though the sounds themselves might be encrypted, they still have the appearance of known words and can therefore be guessed. For example “cow” will appear different, due to the length of the word when spoken, compared with “cat”. It might look something like ASDFADSFADSF versus ASDF.

Ok, second attempt:

Compress your bitrate
Hear the keys to sound exposed;
VoIP flows insecure.