For the sake of balance, I feel I ought to note that synthesising English speech is significantly harder than synthesising Japanese speech, due to the structure of the language. Japanese is one of the most straighforward languages to synthesise, due to the relatively simple pronunciation and structure.
That said, Vocaloid sounds shit. There's no getting round it. It is slightly better than Microsoft Sam, since there are fewer complexities to deal with than in English, but that's it. I make no comment on the songs themselves - some are doubtless very good (the Bakemonogatari ED theme, which I can never remember the name of, was originally written for vocaloid, by Supercell, but was then sung by an actual singer instead, and it is amazing). But the synthesiser itself is just terrible, from both an aesthetic (which is admittedly somewhat subjective) and from a technical viewpoint. It's not even designed to be technically great in the first place though - it was always intended to be a low-cost, relatively simple consumer product, purely for the consumer market (ie. random people who want to play around with making music in their spare time), rather than any serious artists or companies.
I also don't think it is hugely popular with anyone other than otaku. The character Hatsune Miku is more popular - I think people find her cute (my local Family Mart sells little fluffy Hatsune Miku cushions) - but the actual vocaloid is not widely listened to, as far as I can tell.
