Hyper-Realistic Artifical Voices

[.matthew.]

So I just read about this and the tech demo is really quite impressive. My first thought was being able to use it for making your own 'radio play' or audiobook without relying on voice actors. I suspect this software will be absurdly expensive at first, but with it getting more advanced by the year, I imagine that some time soon we'll all be able to self publish our writing in even more fantastic ways.

Sonantic

 

[CupofJoe]

I'd be more impressed if there was no background music.
That is covering up a lot of the nuances and cues I want to hear. As far as I hear it sounds like badly mic-ed artists and a not very good audio engineer.
Oh... and lets fake dead actors again...

 

[Astro Pen]

Hmm Although I know tech is innevitable I am very ambivalent about this because I have just spent a couple of weeks recording some of my stories and poems into 5 or 10 minute audio pieces.
For the stories I have put considerable effort into mastering various accents, RP, Bronx or Belfast. depending. And then there is the poetry, timing, pauses, emphasis etc'.
The prospect of any teenager in their bedroom simply lobbing text into an 'app' and replicating that hard vocal work at the push of a button and getting credit as a supposed 'creator' of the final piece is plain nauseating.
Like when a kid presses the demo button on a synth and mom and dad fawn over his "musical talent".

 

[-K2-]

I find it inconceivable that until AI learns to understand emotion, the mood and tone of much more than a single phrase (the entire story around it), individual inflection and style, you'll never have a quality self generating voice. Using one of my favorites--about as monotone and bland as you can get--compare this to that (that, the example above, most likely considerably worked over to generate the pauses and very minor inflections).

Whoops, I guess we don't allow soundcloud any more, check here: 1984 By George Orwell FULL Audiobook : Free Download, Borrow, and Streaming : Internet Archive , Frank Muller begins speaking at 90-seconds in.

K2

 

[.matthew.]

Yea, there is the argument to be made that it's a bad thing, but will allow a lot more people to put their work out there (once the technology matures) without the insane expense and effort of current production.

Frankly, when you get all "people shouldn't be able to do what I do easily," it is the same as saying writers shouldn't use word processors because they're easier to use than typewriters... Or that ebook self publishing is bad because it devalues the work of typesetters

Technology moves on and opens doors, and while I wouldn't expect 'perfect' performances, or for the software to 'know' what inflections to use at what time, it will still be a huge boon to creators (for example writers who don't have the vocal range or money to work in other mediums).