I'm interpreting your question literally in my answer, as I think you intended it: "Which actually reconstructable proto-languages languages are believed to have been spoken around 7500 BCE?". PIE might have been spoken around that time (so Hittite was not spoken, nor Germanic) – this could actually push PIE back too far, see below for discussion of dating. Bantu was not spoken then – that's a later development. Austronesian doesn't date back that far either (it's put more at the 3500 BCE). AFAIK, there is no reconstructed proto-language from which Austronesian descended, where we can say "at 7500 BCE the language that was spoken is Proto-Austronic", so Austronesian drops out of the count. Austro-Asiatic is also thought to reconstruct to much later than that period, so scratch that family. You should note that I'm not mentioning "Austric" which could be an ancestor of Austro-Asiatic and Austronesian, because Austric isn't a reconstructed language (which might have a time depth), it's a conjecture that certain proto-languages might be related.
For pre-Bantu, we would probably conclude that Niger-Congo was spoken around that time (whence Bantu and related languages). The difference between Bantu and Austronesian is that for Austronesian, we can only say "Something, which must have led to Austronesian, was spoken", but for Bantu we have to make a choice as to which specific level of reconstruction we are talking about. Likewise, Semitic probably reconstructs to 3500 BCE so that would not be a contender, but maybe Afro-Asiatic (but it may be that Afro-Asiatic actually is much older: yet we don't have a descendant branch in Afro-Asiatic that confidently reconstructs to 7500 BCE, in case Afro-Asiatic is too old).
Because of lack of time-depth in reconstructions, that lets out much of the New World, though Dené–Yeniseian could qualify. (A caveat: Dené–Yeniseian is not actually a reconstructed language, it's a family some parts of which are kind of reconstructable). We can probably add Uralic to the list. However, given the imprecision in dating reconstructed languages, it might be that Uralic and Austronesian have comparable time-depths (so Uralic should not be included?). Pama-Nyungan might reconstruct to that time depth.
Again, many languages will fall out of the count simply because there is no reconstructed language (viz. there are many language isolates) – Basque, Burushaski, Etruscan, Korean, Sumerian. I may have mentioned the entire set of proto-languages spoken at the target date, as long as you don't mean 7500 BCE plus or minus 4000. Clearly very many languages were spoken in 7500 BCE, but they mostly do not correspond to actually reconstructed languages.
One cannot look at structural changes from a proto-language and compute a time-depth based on a theory that languages change at a constant rate. Instead, you really need to look at archaeological and word-appearance facts (where a particular word referring to a significant cultural item, such as plant names or technology, correlates with language subgrouping, and this can be pinned to archaeological events like the emergence of Urewe ware maybe around 500 BCE). It may be that Indo-European archaeology is in a more advanced or, simply, lucky state compared to Niger-Congo. In fact, looking for "7500 BCE" is probably not reasonable, instead you should look for a range spanning at least 3 millenia, definitely not centered around 7500 BCE, since that is at the far limits of reconstructive technology.