Maybe im paranoid but all this technology with video manipulation and audio manipulation worries me.
evo8
quote:
Originally posted by MSZ
Maybe im paranoid but all this technology with video manipulation and audio manipulation worries me.
there was a vid on youtube recently where they were dubbing over famous peoples voices but also matching facial movements/expressions to the new dialogue, using videos of Dubya and Putin as examples, scary indeed.....
DJ RANN
quote:
Originally posted by MSZ
Maybe im paranoid but all this technology with video manipulation and audio manipulation worries me.
Yeah, I'm not OK with this.
First there's a video tech:
Then that Adobe's new software can do real time voice manipulation to make you sound like anyone else as long as you have about 20 mins of them speaking for a "voice map".
This could be used for some really ed up uses. Adobe initially said they're not going to release it but now it seems they are.
SystematicX1
Delved deeper and found out that this was a 4th generation application,which says to me that they have had this technology for quite awhile. Call it government contracts or whatever,but I am curious,if this has been around for awhile already..what is next
tehlord
In other news, Nvidia now has an AI super computer that will start designing products for Nvidia.
Looney4Clooney
quote:
Originally posted by DJ RANN
Yeah, I'm not OK with this.
First there's a video tech:
Then that Adobe's new software can do real time voice manipulation to make you sound like anyone else as long as you have about 20 mins of them speaking for a "voice map".
This could be used for some really ed up uses. Adobe initially said they're not going to release it but now it seems they are.
you think ?
It sounds a little too good to be true. Usually the term sound the ssame mean almost the same and that usually means close enough for a 420 youtube video.
Any actual example probably has the bests voice over actor doing the map with good mics and a dead room .
but ya, i would love this , The fact that i haven't heard about this from post prod people makes me wonder.
DJ RANN
quote:
Originally posted by Looney4Clooney
you think ?
It sounds a little too good to be true. Usually the term sound the ssame mean almost the same and that usually means close enough for a 420 youtube video.
Any actual example probably has the bests voice over actor doing the map with good mics and a dead room .
but ya, i would love this , The fact that i haven't heard about this from post prod people makes me wonder.
Almost the same is right, but considering our news cycle these days literally does no fact checking or due diligence, a 420 Youtube is basically the equivalent of DNA these days.
Also this is the consumer grade . Imagine what the security agencies or google are working with. Tbh, I've seen a couple of pro tools ops who were already able to bend voices in to shape the hard way (robbie williams, selena), and this just automates that process.
The voiceover thing is interesting though; you could imagine that a lot of actors could just create a perfect map and the studio could keep it on file for ADR.
They'd never have to go back in the studio for any voiceover (and 99% of films these days are dubbed afterwards). They could literally sell voiceovers for doing no work.
Just type in the script and let the software do the talking. Bit of articulation by an engineer and you probably wouldn't be able to tell unless you did deep forensic analysis.
FruitTumySlam
I want to get this.
Looney4Clooney
quote:
Originally posted by DJ RANN
Almost the same is right, but considering our news cycle these days literally does no fact checking or due diligence, a 420 Youtube is basically the equivalent of DNA these days.
Also this is the consumer grade . Imagine what the security agencies or google are working with. Tbh, I've seen a couple of pro tools ops who were already able to bend voices in to shape the hard way (robbie williams, selena), and this just automates that process.
The voiceover thing is interesting though; you could imagine that a lot of actors could just create a perfect map and the studio could keep it on file for ADR.
They'd never have to go back in the studio for any voiceover (and 99% of films these days are dubbed afterwards). They could literally sell voiceovers for doing no work.
Just type in the script and let the software do the talking. Bit of articulation by an engineer and you probably wouldn't be able to tell unless you did deep forensic analysis.
it shouldn't be really that hard given that the science behind it isn't theoretically that hard but then you have to wonder, why do automated systems all sound so robotic.
The applications are really endless. You can do your own map for days where your sick. The licensing of voice maps would be a huge extra source of income.
There is just something that seems too good. For example, i'm assuming this is english ? Well what if you speak in french. The basic formant adjustments should translate given that the voice map is more based on phonetic data.
I don't know. I think it is just the fact it is adobe. I also don't see how it will allow for small inflections just based on the text example. I actually thought it was more using one voice to morph and sound like an another.
if it is just text to audio, less exciting and given the research in linguistics, this is not going to be happening any time soon at least most languages until they can account for the actual important part of speech which is the tone and inflection. 7- 38 - 55 rule.
anyways, i shoould watch the entire video lol
tehlord
I would expect that at some point (if it hasn't already been shown) that it could work like a vocoder where a voice actor (modulator) can talk into a magic box and the stored voice (carrier) will come out sounding like the famous person the stored voice was 'sampled' from.
The Running Man predicted this. :tongue2
evo8
quote:
Originally posted by tehlord
I would expect that at some point (if it hasn't already been shown) that it could work like a vocoder where a voice actor (modulator) can talk into a magic box and the stored voice (carrier) will come out sounding like the famous person the stored voice was 'sampled' from.
