The Wikidata feature in Phonos (T319270) is built upon a distorted idea of what the IPA is, that there is such a thing as "the IPA for German", "the IPA for Hindi", etc., when in fact the IPA is simply a tool, a set of shorthands, for linguists to efficiently communicate whatever it is that they want to convey in the given context. As the Handbook of the IPA (1999: 30) puts it:
There can be many systems of phonemic transcription for the same variety of a language, all of which conform fully to the principles of the IPA. ... In English, for example, the contrast between the words bead and bid has phonetic correlates in both vowel quality and vowel duration. A phonemic representation which explicitly notes this might use the symbols /iː/ and /ɪ/ ... But it is equally possible unambiguously to represent these phonemes as /iː/ and /i/ ..., or as /i/ and /ɪ/ ... All three pairs of symbols are in accord with the principles of the IPA ... The IPA does not provide a phonological analysis for a particular language, let alone a single 'correct' transcription, but rather the resources to express any analysis so that it is widely understood.
Notice /i/ represents the vowel in bid in the second pair, but the one in bead in the third. This means you can't know whether the transcription bid is supposed to be pronounced like bead or bid just by looking at it—you have to know the conventions underlying it.
So the idea that you can extract a transcription from Wikidata and expect a TTS to render it the way the transcriber intended is fundamentally bankrupt. If users still want it, they can do it by putting {{#statements:P898}} inside {{#tag:phonos}} themselves. But it shouldn't be in the Phonos code. It is not even clear what the use cases are or what problem it is trying to solve to begin with.