I don’t believe that the text to sound is accurate here… in particular i don’t believe that there is as much liaison between the two first words as I hear here.
Thanks for reaching out! Is it just this phrase or you’re referring to the accuracy of the text to speech for Romanian, in general?
How do i find the Romanian text to speech? It would be better for you to get a response from a native speaker, but i would say it’s generally pretty good, maybe about 95% of the time, and that’s not too surprising because Romanian, compared to some languages, gives stress marks and the spelling leads pretty logically to pronunciation. But the weaknesses i have heard are not usually within a word, but generally relate to the liaison between words and the stress of the whole sentence. Occasionally i hear pronouns get a big stress in the sentence, and this should normally not be the case. In this sentence, in my opinion, the biggest stress should be the “LEG” syllable, with a secondary stress in the next word, pronounced “re - ti - CHEN - tsa” . The way I heard it on July 25 threw me off totally. I’m not sure how your text to speech programming works, but if nouns and verbs were globally given a stress higher than pronouns and prepositions, it might improve things a little.