BlackOut wrote: ↑1 year ago
Been working on this today - started it off with a text to video prompt and have been taking the end frames from each clip to continue the animation further - voices added in via Eleven Labs - touch up editing done in Clipchamp.
Did you use the voice changer in 11Labs, or the text-to-speech? Generally speaking if you choose one of the pro quality voices that have been highly trained (there are quite a few available to free accounts) you'll get much more interesting results with voice changer. With t2s you're relying on the algorithm to guess the cadence from the text, but with voice changer you can record yourself saying the dialogue exactly how you'd like the AI voice to say it. The pro voices will even do giggles, laughs, grunts, gasps, exclamations, sarcasm, whispers, etc. -- which the t2s absolutely cannot do!
There's a handful of useful pro voices that I've found, available to free accounts. For example, "Jessica Anne Bogart" has a slightly husky American female accent; "Brittney - Young, Peppy" has a youthful American female accent, light but not too girly; "Donny" is a male super deep movie trailer voice; etc... There's just far too many to list.
EDIT: Some further notes... The first video I posted uses Jessica Anne Bogart, which is an American voice but I recorded the audio with a posh British accent as an experiment (to try to match the Victorian setting.) What came out is a kind of hybrid trans-Atlantic voice, which sounds quite nice. But I've noted, while experimenting, that it helps if you can mimic the speech patterns of the AI voice accent you're using. For example, if you speak English but choose a Chinese voice you get something that sounds really rather racist. Although the accent is authentically Chinese, the cadence of the voice and the way certain sounds are pronounced (or mispronounced) is inauthentic. So it doesn't sound like a native Chinese speaker struggling to get their tongue around English pronunciations, rather it sounds like Benny Hill doing a 'comedy Chinaman' skit. (Out of curiosity I watched some YouTube videos by a Hollywood voice coach, and he explained the secret to an authentic Asian-speaking-English voice is how double consonants at the ends of words, and repeating consonants between words, are vocalised. And also the voice timbre. There's more to it than mispronouncing your R's like Benny Hill.
)
R5
