Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

15.ai

15.ai

Post reply on HN
Print view View original

51–60 of 192 posts

Re: 15.ai

#51

Post by BuyMyBitcoins »

jamal-kumar wrote: post #5
armchairhacker wrote: post #2

This is really cool. It's a text-to-speech and the gist seems to be that they synthesize it from only a little audio. The results are clearly synthetic and need work. However what's cool is that there are a ton of characters (from popular shows and video games) and there are useful statistics like inferred emotion (which is also in the output). Honestly it's a big problem how a lot of AIs are like "black boxes" where…

I just used it to make spongebob squarepants say bad things.

This thing synthesizes dolphin squeaks? Wow!

Re: 15.ai

#52

Post by jonas21 »

btown wrote: post #15
eljimmy wrote: post #13

This is unrelated but what's with the fascination with HN users and My Little Pony? I've noticed this on a lot of posts in the past few months.

https://en.wikipedia.org/wiki/My_Little_Pony:_Friendship_Is_... explains in detail - between 2010 and ~2015 there was a massive overlap between millennial geek culture and unironic fandom of the rebooted My Little Pony show, especially among millennial men. One dedicated fan hub averaged almost 400k page views per day over its first 3.5 years of existence. And throughout it all, programming projects abounded, such as…

One of my favorite examples from that era:

https://pjreddie.com/static/Redmon%20Resume.pdf

And in case you were wondering what this little pony did next...

https://scholar.google.com/citations?user=TDk_NfkAAAAJ&hl=en

Re: 15.ai

#53

Post by julianeon »

deeplearner1 wrote: post #40

If you want more information about 15.ai, I highly suggest reading their Wikipedia article! https://en.wikipedia.org/wiki/15.ai The whole history behind the project is fascinating: 4chan had a huge role in its development, and the project's work was stolen by an NFT company that a famous voice actor endorsed not too long ago.

Ah, I was wondering why they were so concerned about attribution.

The truth is that, today, if I was going to use a tool to generate voices (say for YouTube), I wouldn't necessarily pick a small SaaS tool. I'd use Amazon Polly or some other GCP-style platform voice creation tool. There are already a few products in the space, and their costs are so low as to be almost negligible (example: Polly, 5 million characters free). For a commercial project, I could probably stay on a free tier for a whole year.

With Dall*E, it seems like the only option, and it's such a superior option that a website could abuse it for commercial profits. But for voice synthesis, it's already dirt cheap and commercially available without limitations.

Re: 15.ai

#54

Post by sillysaurusx »

jameshart wrote: post #28
canjobear wrote: post #21

Earlier quoted context omitted.

Only if you're OpenAI

Fine tuning of GPT3 models is available via their public API. Costs credits, and you need to get their permission to use it in an actual application, but it’s not locked in a lab.

So “Only if you’re OpenAI” :)

If the weights were public, the community would figure out a way to fine tune it.

Re: 15.ai

#57

Post by userbinator »

armchairhacker wrote: post #2

This is really cool. It's a text-to-speech and the gist seems to be that they synthesize it from only a little audio. The results are clearly synthetic and need work. However what's cool is that there are a ton of characters (from popular shows and video games) and there are useful statistics like inferred emotion (which is also in the output). Honestly it's a big problem how a lot of AIs are like "black boxes" where…

I agree this is an amazing demonstration of what AI can do, but I think that the current method of "learn and repeat" that depends on having tons of computing resources available is still too inefficient in many ways. Personally I'm more interested in what parameterisable formant-based synths can do, since they are extremely efficient and can produce a theoretically infinite variation of voices, although the output quality is still not great. Example: https://news.ycombinator.com/item?id=31604299

Re: 15.ai

#58

Post by sterlind »

Deritio wrote: post #8
armchairhacker wrote: post #2

This is really cool. It's a text-to-speech and the gist seems to be that they synthesize it from only a little audio. The results are clearly synthetic and need work. However what's cool is that there are a ton of characters (from popular shows and video games) and there are useful statistics like inferred emotion (which is also in the output). Honestly it's a big problem how a lot of AIs are like "black boxes" where…

Dall-E 2 has customization. You can remove or add things etc. And for GPT you can also specify more details. Only a question of time until you can work with the ai on your art/thing. There are ai models which keep track of context and others which generate a plan of actions. AI is not a blackbox

OpenAI itself is a black box. Until I can reproduce their models or download them myself, and have unfettered access to them, it's just gatekept magic behind an API. So much for democratizing machine learning.

Re: 15.ai

#60

Post by jameshart »

sillysaurusx wrote: post #54
jameshart wrote: post #28

Earlier quoted context omitted.

Fine tuning of GPT3 models is available via their public API. Costs credits, and you need to get their permission to use it in an actual application, but it’s not locked in a lab.

So “Only if you’re OpenAI” :) If the weights were public, the community would figure out a way to fine tune it.

It’s not a matter of ‘figuring out’. The model supports fine tuning. It’s a core feature of the openai API. Running ‘fine tuned’ versions of GPT-3 that are created by customers is literally their SaaS model. They have examples in the documentation. Here: https://help.openai.com/en/articles/5528730-fine-tuning-a-cl...
Post reply on HN