Skip to main content

Indie game storeFree gamesFun gamesHorror games
Game developmentAssetsComics
SalesBundles
Jobs
TagsGame Engines
(+1)

Yeah, the thing is, people don't talk with spaces. We anticipate the second word and this affects how we say the first one. The proof for this is- if you record someone talking and then move the words around in Audacity or whatever- it sounds completely unnatural- like an automated train station announcement.

Also, people often slur speech in a way that sounds perfectly natural, but is rarely thought about. So, the word 'ladies' often sounds more like 'lays' if you analyse it in natural speech.

So- 'hot dog' might be prounounced 'hoddog' and so on.

I like linguistics chat on itchio- this was my degree XD

Yup, despite not having a linguistics background or fluency in more than one (human) language, I find this stuff pretty interesting. Text-to-speech like just using Siri can expose some of these issues (I like to set Siri to non-American English voices to hear how different it sounds).

My dim sum app uses text-to-speech (and speech to text) for Chinese, and when it's fed one word at a time, the compound words sometimes sound unnatural (and in Chinese sometimes a word's tone might change if it's combined with another word, I think for a similar reason, so it's easier to say the combined words).

Out of curiosity, I just tried google translate, and it pronounces hot dog pretty naturally, but for a contrast, the Japanese loanword throws in an intervening vowel.