Image
Comments
| cyber95 #824820 6 months ago |
Because voices are a tad more complicated than musical instruments? |
| cyber95 #824823 6 months ago |
also my train of thought was travelling through this and stopped at "Pinkie Pie voicing my GPS" |
| DingoSteve #824825 6 months ago |
Voices are much more complex than musical instruments. While instruments are essentially the same sound at different pitches and such, voices have sounds, accents, other features, and all that. It's not that easy. |
| BUzer #824829 6 months ago |
Musical instruments are prerecorded.
But yeah, pony voices sounds like music to me, indeed. |
| cyber95 #824830 6 months ago |
That said, voice emulation technology is getting pretty good. Still very limited, but I was listening to a vocalization of a news article on a website once, and it took me an entire paragraph to realize it wasn't a real person.
Some day. |
| Anonymous #824853 6 months ago |
When I'm fighting Pokemon in Black/White the sound the game makes when selecting an action resembles a certain yellow pegasus making a certain cute cheer.
I don't know if that counts, but still... yay. |
| Anonymous #824855 6 months ago |
If someone can get a voice sample of every syllable used in english then someone could make something like a pony vocaloid. |
| DeeperMadness #824903 6 months ago |
Do you mean something like this combined with this?
Honestly, you have a whole Internet at your disposal. Just pick your favourite search engine and type in a few key words. It's easier than making a text image and waiting for results on here. |
| Wheezie_Moonflower #825117 6 months ago |
There are quite a handful of "text to speech" tools out there. Many are owned by Nuance. It's in the Vocalizer package. |
| MintBerryCrunch #825119 6 months ago |
There are YouTube videos with mixed vocals all over the place. Problem is it takes a hell of a lot of work, and still sounds really terrible. |
| Anonymous #825155 6 months ago |
Simple. |
| HereComesGummy #825192 6 months ago |
Simulating a generic voice is simple. Simulating a SPECIFIC voice is very difficult. And it's doubtful we'll ever reach the point where you can just ask a voice simulator to mimic the person of your choice.
Voices are like fingerprints; every single person's is a bit different. |
| P0ny_News #825206 6 months ago |
Beside, even if the voice is perfectly imitated, one can still "feel" that the voice owner isn't human. How sentence are formed, for example...
It's impossible for a computer to talk like a human. |
| Anonymous #825774 6 months ago |
Because Jews control the music industry. |
| Xuncu #825930 6 months ago |
Because digital is digital, and a mechanical analogue that's as precise as a biological being (the VAs) would be exeedinlgy fragile and expensive.
The digital part, it's part of how digital works; it's all binary, so it's either 1 or 0, no .5, or the infinite points in between any two intergers. And the kbps of a sound file is how much the file is programmed to read in (kilobytes) bits (one instance of 0 or 1); so while you can make it sorta sound close, utlimatley, you can't quite make it compressed enough to accuratley... well, the analogy is rasterizing to the resolution of "molecule"-- same way that today you can still tell the CGI plants in Avatar, because it just doesn't "look right". 
Meanwhile, a Vinyl record does record on the molecular level of the plastic, hence why most audiophiles still swear by LPs rather than MP3s. And those fancy stereo systems like Bolse radios; there's an air channel inside of them than, in a very loose anaolgy 'wears down' the jaggedness of the sound waves to make it into a more naturalistic curve. |
| Xuncu #825950 6 months ago |
And yeh, as mentioned above, an instrument is a tuned tool meant to produce a particular tone; on a sonogram, no matter what note or key, a properly functioning instrument will produce a clear sine wave, while the human voice (or (nearly) any animal sound) is a range of overlapping buzzes and hisses, which looks like a big scribble on the same sonogram.
What's more, when you have music going on, too; your speakers are trying to not also play the sines of the music, but the erratic scribble, as well; so, that terraced line above has to be rendered were all these lines overlap, so it's innacurate to both (at least two) sonogram lines. |