[Physicist]

All of what you say is true, but perhaps not for commercial VOIP systems. These depend on sending as little information as possible. If customers started using an encryption scheme that maximized the amount of data sent, rather than just rearranging it, the provider will quickly put a stop to it one way or another. They'll either prevent the practice or drop those customers. Bandwidth is money.

[RadioAstronomer]

However, if you broke it up into tiny packets with fill bits filling in the blanks and sent the fixed length packets across at a reasonable bit rate, it would mask the phonetics.

Just my two cents. :-)