This is another of those questions that people debate endlessly. There really is no single (correct) answer - it depends on your ears, your equipment, the quality and complexity of the input sound data, etc. It's probably fair to say that 128 Kbps is a bit on the low side for music, while 160 Kbps encoded well should sound pretty decent to most people. Which MP3 encoder you use really does make a difference, and the LAME MP3 encoder (the default in CoffeeCup CD & MP3 Ripper) or the Fraunhofer MP3 encoder work very well.

Also note that simple speech is generally much less acoustically complex than music, and that people generally care less about the audio quality of speech, as long as it's understandable. This means that bitrates below 128 Kbps can be just fine for compressing speech, even though those bitrates are usually not even worth considering for music.