I don't really think that changing speed and pitch is highlevel audio processing, but it might still be a little out of SFMLs scope (although it would be really nice!) and there is the licensing issue.
Processing on sf::SoundBuffer is quiet easy, due to the getBuffer() method. I have done that before.
But I have not succeeded to manipulate sf::Music in real time. So like suggested it would be an awesome addition if there was a way to access the musics samples for processing. This would give everybody the freedom to do what they like (FFT analysis, pitch, distort...)