I have always wondered whether there is a clear distinction between (speech) sounds and segments. The two words appear to be used interchangeably in some places while in others they are considered to be different.
Although this is a "usage" question, it's about a potential technical distinction. There is no technical distinction between "a sound" and "a segment", and the thing is most often referred to in the phonological literature as a segment. On the other hand, outside of linguistics, the concept of "segment" is alien, instead people talk about "sounds". This means that there is some tendency for non-technical writing by linguists to use "sound" rather than "segment", to cut down on the amount of arcana that has to be introduced to the general audience.
"Segment" has a somewhat technical connotation, just as "atom" is a technical term in physics, in that an atom is not onlyelectrons, protons and neutrons, it also involves relations between these primitive units. In older SPE phonology, a segment was the totality of features delimited by the first feature "segment", and in autosegmental accounts, it more or less corresponds to "everything dominated by a single Root node", but in fact the existence of the segment as a special representational unit was not universally accepted (see for example Grounded phonology, and similar decompositional efforts by Steriade and Selkirk).
The term "sound" is more widely used by phoneticians, probably because "segment" has certain ontological implications that phoneticians may not want to commit to. However, when phoneticians approach the problem of cutting up continuous speech into discrete units, they talk of segmenting an utterance and the results will be segments (though their choices are often different from those made by phonologists, who are not traditionally concerned with segment-internal structure, such as closure versus release).