AI Clones Your Voice After Listening for 5 Seconds (2018) • Hacker News Board

Re: AI Clones Your Voice After Listening for 5 Seconds (2018)

#231

Post by thedaemon » Wed, Nov 13, 2019, 11:18 PM UTC

r00fus wrote: post #195

throw1234651234 wrote: post #190

Earlier quoted context omitted.

"They were not delighted to hear that." - I used to do that, be a smartass. Then I realized at worst they get more info, at best, you are training them and wasting your time. Then there is that whole thing where they are getting your voice.

POTS just needs to die unless they allow for authenticated CID. One could wonder if this was some sort of conspiracy to break one of the most successful protocols in the world (or at least not update it so it dies by neglect) to increase profits by other means.

POTS is pretty much dead almost everywhere in the USA, most are VOiP these days. They are not replacing the copper wire.

Re: AI Clones Your Voice After Listening for 5 Seconds (2018)

#232

Post by Lucadg » Wed, Nov 13, 2019, 11:18 PM UTC

anextomp wrote: post #125

Lucadg wrote: post #100

At first glance there seems to be more more malicious uses than good ones with voice. Yes, hearing someone dear to me say things he/she never said maybe comforting. Anything else? Maybe some movies with the deceased actor's voice? But what if someone who wants to hurt me sends me files (or phone calls) from the deceased person saying horrible things like: - "I am still alive but left as I was tired of you" - "oh Jan,…

I'm looking forward to an AI assistant that can use my voice to make phone calls I'm sure some people with selective mutism that would like to use text to speech with their own voice > How will we adapt? Digitally signing audio clips

Ok mutism is a great use case, thanks. Still, the problem is solved also with normal text to speech software

Re: AI Clones Your Voice After Listening for 5 Seconds (2018)

#233

Post by JaRail » Wed, Nov 13, 2019, 11:21 PM UTC

Riccardo_G wrote: post #199

What it is doing is not really cloning, but because it was trained on 18k different voices, it actually finds one that is closest to yours, and uses that one. It can do a bit of interpolation to create an embedding which is closer to your own, but only if it is well represented by a mix of other voices. Real voice cloning like at https://replicastudios.com/ can take just a minute or two of audio, and it does a fairly…

I'm not really sure where you're getting this. It doesn't pick a specific voice from a database to use.

From their introduction: "Our approach is to decouple speaker modeling from speech synthesis by independently training a speaker-discriminative embedding network that captures the space of speaker characteristics and training a high quality TTS model on a smaller dataset conditioned on the representation learned by the first network."

Section 2 of the paper explains how it works. Two minute papers also goes through it if you'd prefer a video. Link: https://youtu.be/0sR1rU3gLzQ

Re: AI Clones Your Voice After Listening for 5 Seconds (2018)

#234

Post by thomascgalvin » Wed, Nov 13, 2019, 11:25 PM UTC

jfengel wrote: post #187

thomascgalvin wrote: post #113

Earlier quoted context omitted.

I legitimately think this could be huge for self-published authors. It takes a skilled professional about forty hours of work to produce an audiobook from a novel-length manuscript. Tacotron could do it in minutes.

I don't see that coming soon. The voice is one thing, but the performance goes far, far beyond that. Without understanding the text, you can't get good prosody out of a single sentence, much less developing a character for a whole performance. You'd have to "direct" this on a word-by-word basis: "Put the emphasis here. Speed up 10% here. Decrease vocal intensity 25%". You'd end up producing a whole "score", and it wo…

Audio books with genuinely good performances seem rare to me. There are a handful of voice actors that stand out, but many of the titles I've listened to have very flat delivery; the first sample in the original article has as much inflection as they do.

Re: AI Clones Your Voice After Listening for 5 Seconds (2018)

#235

Post by danaris » Wed, Nov 13, 2019, 11:26 PM UTC

Lucadg wrote: post #229

danaris wrote: post #123

Earlier quoted context omitted.

It doesn't have to be about creating audio of specific text with 's voice; it can be much more about creating audio of specific text with a wide variety of believable people's voices. I could see this, if it becomes commercially viable, potentially being a huge boon to indie game creation, for instance, since hiring a load of voice actors to record the dialogue for an entire game is vastly more expensive than, say, h…

Again this is a good use scenario which seems to bring less benefit than the damage a bad use case scenario. Cheaper games vs distressing phone calls in this case. I'm open to better use cases but for now I haven't heard any.

I mean...it's not like we're getting a choice here. The technology is possible, therefore it will happen (and is now starting to happen).

Like so much else in life, we have to take the bad with the good, but not looking for good in it doesn't mean we're less likely to get the bad.

Re: AI Clones Your Voice After Listening for 5 Seconds (2018)

#236

Post by x2f10 » Wed, Nov 13, 2019, 11:27 PM UTC

arkades wrote: post #159

I once got a call while I was lecturing some students. It was repeated three times in three minutes - I assumed it was an emergency and stepped out. I was greeted by someone explaining that my father had caused a car accident, and they were calling on his behalf. That someone would need to send over some money for repairs or they’d call the police. Sure. They added that their cousin, the driver, is a parolee now hold…

How were you to send money?

Re: AI Clones Your Voice After Listening for 5 Seconds (2018)

#237

Post by rolltiide » Wed, Nov 13, 2019, 11:44 PM UTC

arkades wrote: post #159

I once got a call while I was lecturing some students. It was repeated three times in three minutes - I assumed it was an emergency and stepped out. I was greeted by someone explaining that my father had caused a car accident, and they were calling on his behalf. That someone would need to send over some money for repairs or they’d call the police. Sure. They added that their cousin, the driver, is a parolee now hold…

I know someone that had a fling abroad and their fling began asking for money for treatment over facebook

The American assumed it was a scam and the person did die

I have often found that truth is stranger than fiction, and people are too conditioned for fiction that they can’t perceive truth

Re: AI Clones Your Voice After Listening for 5 Seconds (2018)

#238

Post by zaroth » Wed, Nov 13, 2019, 11:45 PM UTC

Benjammer wrote: post #219

zaroth wrote: post #216

Earlier quoted context omitted.

I’m convinced that POTS is already a dead man walking. The younger generation has no interest in talking on the phone in the first place, and with the amount of spam calls my phone number is absolutely the worst part of my phone. Texting and apps is a much more pleasant way to interact with someone, and bonus of no hold times and much can be automated. I think business texting is an upcoming startup unicorn which wil…

>business texting You mean Slack?

B2C - not team internal chats.

It will have “jumped the shark” once there’s an SMS button on the business listing that Google Maps displays.

Instead of the wonky/creepy Google demo which did speech to text and then analysis and then text to speech relayed over to the business, every business will just communicate directly with customers over text.

It’s not that this isn’t already done (to some extent). And more so in some countries outside of the US.

But I have no doubt it will become the primary/preferred way to connect with any business, to the point where you will text with an 800- number long before it would occur to you to dial an 800- number to get service.

Like for example, the warranty claim I just made on my Dyson handheld vacuum for a battery replacement. Search for “Dyson warranty claim” and they tell you to dial their 800- number. Now their phone helpline is absolutely the best of the best, but even still most people would [will eventually] prefer to interact via text.

Another example, making a reservation directly with a restaurant (which I prefer to do versus using OpenTable which will take a cut for doing nothing), is a perfect usecase for texting. Also ordering take-out if you already have a favorite order saved, obviously all the notification type things which make sense over SMS instead of email, making an appointment when a dedicated app is too much overhead, etc. etc.

Re: AI Clones Your Voice After Listening for 5 Seconds (2018)

#239

Post by JaRail » Wed, Nov 13, 2019, 11:53 PM UTC

azinman2 wrote: post #135

godelzilla wrote: post #120

Earlier quoted context omitted.

Thanks... Too long of a scroll to find somebody posting the actual science behind the click-bait.

Nit: this is more design/engineering than science. There is no hypothesis being tested about how the world works.

Did you read section 3 of the paper where they evaluate their system?

> We primarily rely on crowdsourced Mean Opinion Score (MOS) evaluations based on subjective listening tests. All our MOS evaluations are aligned to the Absolute Category Rating scale [14], with rating scores from 1 to 5 in 0.5 point increments. We use this framework to evaluate synthesized speech along two dimensions: its naturalness and similarity to real speech from the target speaker.

They're testing if the generated speech sounds natural with a well-defined and reproducible experiment. That's science.

Re: AI Clones Your Voice After Listening for 5 Seconds (2018)

#240

Post by blancNoir » Wed, Nov 13, 2019, 11:53 PM UTC

pmoriarty wrote: post #136

AbbasHaiderAli wrote: post #39

Wow, impressive results! Already a few examples in the comments of what bad actors could do this tech. I wanted to share an example of something good. I lost my dad about 6 years ago after a Stage 4 cancer diagnosis and a 3 month rapid diagnosis. I have some, but not a lot of video content of him from over the years. My mom still misses him terribly so for her 60th birthday I tried to splice together an audio message…

Philip K Dick wrote about people going to commune with artificial personality constructs of their deceased loved ones. Unfortunately, it's been a long time since I read it, so I don't remember which book it was in. Maybe someone who's read him more recently can remember. Update: Apparently, lots of other people wrote about this too, but PKD wrote about this before any of the ones mentioned so far, as he wrote about t…

Check out the movie with Jon Hamm, Marjorie Prime, screenplay by Jordan Harrison. Without spoiling too much, there's a company that can create holographic projections of loved ones which a woman's family gives to her as a gift which is a hologram of her deceased husband, but when he was a young man. The interesting part, narratively, is that while the holograms are near perfect physical recreations, their personalities and memories must be trained by those who knew them, family/friends which raises the question of how we're perceived in fragmentary and contradictory pieces depending on whose doing the training and the amalgamation of a person that's ultimately constructed from these parallax accounts. The writing is actually quite strong and the only scifi aspect is the holograms so I wouldn't say there's much of scifi crutch. I know it's not PKD and there are similar Black Mirror episodes, but I thought the drama itself was robust and displayed the range of Jon Hamm to be someone other than Don Draper.