Heres cockney gangster 'Bricktop' from the Guy Richie movie Snatch, talking about birdwatching https://www.youtube.com/channel/UCRRrlrMDAd-a2sn1d81lTKA
Unauthorized "David Attenborough" AI clone narrates developer's life, goes viral
61–70 of 195 posts
Re: Unauthorized "David Attenborough" AI clone narrates developer's life, goes viral
#62Loved it, I thought that was quite hilarious! I can see how people get a little upset about it though, but in my mind this kind of content is just unavoidable.
I myself have been playing a lot with this kind of tech lately by creating realistic clones (voice + videos) of YouTubers and tiktokers. I think it has potential to go viral quite easily. I never published any of those though, because of ethical concerns. Just a question of time before other creators do this at scale though.
I almost did that with cloning real TV newscasters from bbc / fox / cnn, etc... but decided not to. But I did create a 100% "fake" journalist for an app that lets you create breaking news on the fly [1]. The result is still a little rough, obviously, but I can already tell how in just a few months it'll be impossible to distinguish a clone from a real person.
Re: Unauthorized "David Attenborough" AI clone narrates developer's life, goes viral
#63Earlier quoted context omitted.
> Requires a 3 second (!) clip of the voice you want to clone. Sure, if you want a guaranteed uncanny valley experience. There is no way a few seconds are enough to cover all the ways a specific person pronounces things. A person's voice is much more than just the pitch and with a 3 second sample anyone who knows them will be able to tell something's off within 3 seconds.
No I suspect 3 seconds is often enough. It's not learning how you pronounce each word, it's seeing some pronunciations you use and from that guessing how you would pronounce other things based on how those pronunciations are clustered in its training data. In other words if in that 3 seconds it hears you say y'all it has a pretty good shot at inferring how you say a lot of other things.
> it hears you say y'all
The accuracy or realism of the emulated voice probably depends on how "rich" the recorded material is within those 3 seconds? I mean, if it includes a single long syllable, or some phrase that contains a variety of vowels and consonants.
Re: Unauthorized "David Attenborough" AI clone narrates developer's life, goes viral
#64Earlier quoted context omitted.
Have you tried to get a GPT to write a joke? It’s got a ways to go.
It’s not funny because it’s “aligned”. You can’t be funny with all those restrictions.
It’s not funny because it doesn’t know what it’s talking about and there’s a big gap between “follows the rules of jokes” and “actually amusing”
Re: Unauthorized "David Attenborough" AI clone narrates developer's life, goes viral
#65When I see a lot of this voice replication I am reminded of something. (Dune series spoiler) For example, Duncan Idaho keeps being brought back with cloning tech, over and over, for centuries, because Leto II just likes him. Over and over, the same man keeps being put by the (nearly) immortal emperor's side... I ask: are we stuck as a monoculture, not making new, interesting things, but simply replicating older ones?…
"Not making new things" is moreso the default, see: all mythology, religion, etc.
Re: Unauthorized "David Attenborough" AI clone narrates developer's life, goes viral
#66When I see a lot of this voice replication I am reminded of something. (Dune series spoiler) For example, Duncan Idaho keeps being brought back with cloning tech, over and over, for centuries, because Leto II just likes him. Over and over, the same man keeps being put by the (nearly) immortal emperor's side... I ask: are we stuck as a monoculture, not making new, interesting things, but simply replicating older ones?…
This reminds me of something I noticed: disproportionately, the classics in a given field (ie, the work agreed upon to be particularly important, influential, etc) are disproportionately the ones that were made shortly after a new medium or technique was invented. For example,
It's almost as if we default to tending towards imitation and recycling, and only each time a totally new technology changes what is possible do we really try new things -- but the well of newness is rapidly exhausted and so those artisans who are fortunate to be able to be the first to work with a new medium get to be the ones that create its classics.
The question is why? Does a medium combined with a culture totally determine the kind of work that can be made? Or is the influence of the original classics impossible to escape? In other words, is this situation path dependent?
Re: Unauthorized "David Attenborough" AI clone narrates developer's life, goes viral
#67Someone at work shared this and I thought it was hilarious and awesome. A number of other people found it disrespectful or gross. I'm sort of fascinated at the values divide. I can see why SAG/ASTRA, or voice actors, aren't going to like this technology, but it doesn't feel all that different to any other industry swept away by tech, whether that's farming, horses and carriages, mining, etc. In other words, I can see…
This demonstration is respectful. It’s funny. It’s obvious parody. But, the tech can be used to make disrespectful content. Look forward to: David Attenborough narrates porn. I don’t know what we can do. Ain’t no easy answers. But do we want courts and $Lawyers deciding what is respectful?
I believe Tom Hanks already had some one deep fake him into endorsing some medicare scam product. I'd be more worried about that than the unsavory narration which most people probably would assume was fake.
Re: Unauthorized "David Attenborough" AI clone narrates developer's life, goes viral
#68Earlier quoted context omitted.
> Requires a 3 second (!) clip of the voice you want to clone. Sure, if you want a guaranteed uncanny valley experience. There is no way a few seconds are enough to cover all the ways a specific person pronounces things. A person's voice is much more than just the pitch and with a 3 second sample anyone who knows them will be able to tell something's off within 3 seconds.
No I suspect 3 seconds is often enough. It's not learning how you pronounce each word, it's seeing some pronunciations you use and from that guessing how you would pronounce other things based on how those pronunciations are clustered in its training data. In other words if in that 3 seconds it hears you say y'all it has a pretty good shot at inferring how you say a lot of other things.
People speak differently in different situations, based on their mood and who they are talking to. They may change their voice when quoting another person, they may alter their accent if talking to a stranger, and there's many other details you don't even think about because you're used to them but it'll immediately sound wrong if it's not replicated correctly.
Apple's announced voice imitation feature requires a 15 minute sample if I remember correctly, and you can bet it would be shorter if they were satisfied with the results.
> if in that 3 seconds it hears you say y'all
I can guarantee that no 3 seconds of text of your choosing will ever be enough to reliably differentiate between all existing dialects and accents, let alone differences between individual people.
There's a reason all the decent fake voice clips on the web are based on public figures with hours of training material - and even with those you can tell within seconds that it's fake, even if you don't know how you noticed it.
Re: Unauthorized "David Attenborough" AI clone narrates developer's life, goes viral
#69Earlier quoted context omitted.
This demonstration is respectful. It’s funny. It’s obvious parody. But, the tech can be used to make disrespectful content. Look forward to: David Attenborough narrates porn. I don’t know what we can do. Ain’t no easy answers. But do we want courts and $Lawyers deciding what is respectful?
Courts and lawyers will do a bad job, and also the best job out of every option available. As it has been for 100 years.
I'm sure the Brits would claim since 1066.
Re: Unauthorized "David Attenborough" AI clone narrates developer's life, goes viral
#70One area where this kind of technique would be a game changer, I imagine, is for blind people to have their surroundings narrated in real time. For example, through a pair of glasses with a camera for input, and an earphone to listen to the continuous narration.
I heard about an online service where people can sign up to help the blind by describing visual scenes that they capture on their mobile. There must be companies out there developing a similar service entirely run on machine learning.