The only "natural" (i.e. "using only the data in the sample") way to change the pitch of a recorded sample is to change the speed of its playback. So a sample that sounds an octave lower should take twice as long to play back.
But you have a tool that somehow manages to play that sample back octave lower, while somehow keeping the playback time the same. How can this be? It is achieved through clever algorithmic modification of the sample's playback. However, this modification is a distortion of what it would "normally" sound like, and you get these undesirable extra sounds as a consequence. In general, these extras are called "artifacts."
The good news is that while you can't avoid artifacts entirely, you can optimize for different scenarios. There are many techniques for pitch and time modification, based on concepts like chopping the sample up temporablly into small loopable slices, or deconstructing it harmonically, and reassembling the transformed parts into a new recording. Many tools let you choose from several algorithms, with different tradeoffs. For example, chopping up over time preserves transients, which is very nice for percussion and drums where the transient "hits" are more prominent than the decays, but is problematic for tonal sounds where smoothness is more important (since you can hear the slices).
You may want to look into your Time and Pitch Machine and see if you can choose a different algorithm that might be better suited to your sample.