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

Eleven v3

elevenlabs.io

Post reply on HN
Print view View original

161–168 of 168 posts

Re: Eleven v3

#161

Post by kaycey2022 »

nsonha wrote: post #128
ricketycricket wrote: post #28

From the example: "Oh no, I'm really sorry to hear you're having trouble with your new device. That sounds frustrating." Being patronized by a machine when you just want help is going to feel absolutely terrible. Not looking forward to this future.

Are you specifically looking for reasons to be offended? Even if a human said this, it would have been completely fine.

I dont know man. It makes me inclined to shut off that conversation. Because it sounds like something a nitpicky, “nose all over your business”, tut-tutting Karen would say. It doesn’t convey competence, rather someone trying to manage you using a playbook.

Re: Eleven v3

#162

Post by TeMPOraL »

mrandish wrote: post #158
TeMPOraL wrote: post #112

Earlier quoted context omitted.

What's more likely to be a problem, is the request to be concise. For some reason, this still seems to not be widely known among even technical users: token generation is where the computation/"thinking" in LLMs happen! By forcing it to keep its answers short, you're starving the model for compute, making each token do more work. There's a small, fixed amount of "thinking" LLM can do per token, so the more you squeez…

If that's correct then it's a significant problem with LLMs that needs to be addressed. Would it work to have the agent keep the talky, verbose answer to itself and only return to a finally summary to the user?

That's what the "reasoning" models do, effectively. Some LLM services hide or summarize that part for you, other return it verbatim, and ofc. you get the full thing if you're using a local reasoning model.

Re: Eleven v3

#163

Post by staticman2 »

saberience wrote: post #149
staticman2 wrote: post #146

Earlier quoted context omitted.

I promise this comment will circle back to Elevenlabs: When my cat died after a few months of cancer treatment, the staff of the animal hospital sent me a condolence card with comments by staff members. On the one hand, this was a very touching, very human thing to do. On the other hand, this was presumably a work assignment that had to be passed around and completed for staff members to meet their employer's goals,…

I don't really care whether this chat goes to Elevenlabs or not. This may shock you, but people who are doing reading for audiobooks, enjoy doing it! I'm not sure you've ever listened to professionally recorded audiobooks, but there are actors who are absolutely amazing at this, and clearly doing it with passion and love. E.g. Andy Serkis doing the Lord of the Rings books on Audible. This clearly isn't a person chain…

>Artists enjoy doing their art, whether it's writing, reading out loud, playing music.

I guess this is probably generally true. It's really not always true, though. Neil Gaiman told an anecdote on his blog about knowing some writers who hate writing and are miserable.

The fictional TV show The Larry Sanders Show does a good job of finding comedy in the misery of showbusiness: the main character is a neurotic talk show host who is desperate for top rating, jealous of his rivals, and gets no joy from the process of making a hit tv show. I'm not saying most stars are like that, but there's probably some truth there.

Re: Eleven v3

#164

Post by crazy3lf »

riebschlager wrote: post #31

I didn't see anything about this in the documentation or prompting guide, but... is it supposed to be able to sing? Since I am a fundamentally unserious person, I copied in the Friends theme song lyrics into the demo and what came out was a singing voice with guitar. In another test, I added [verse] and [chorus] labels and it's singing acappella. [1] and [2] were prompted with just the lyrics. [3] was with the verse/…

This seems like it has potential to be a de-earworming AI.

Re: Eleven v3

#165

Post by GrayShade »

lharries wrote: post #53
gozzoo wrote: post #35

Earlier quoted context omitted.

Exactly the same thing with Bulgarian voices.

It should work well for the logged in voices here: https://elevenlabs.io/app/voice-library?language=ro We are in the process of updated the homepage voices for the new languages

It does, indeed. Thanks for looking into it.

Re: Eleven v3

#166

Post by frozenseven »

saberience wrote: post #133
sumedh wrote: post #130

Earlier quoted context omitted.

What difference does it make?

Are you seriously even asking that question? It’s like having a robot that can give you a hand-job and someone saying, “well it’s a robot…” and you saying “what difference does it make?” You tell me? What difference does it make talking with an old friend versus an ai simulation of an old friend? What difference does it make seeing the artist who actually painted something talking about why they painted it, versus ge…

>It’s like having a robot that can give you a hand-job and someone saying, “well it’s a robot…” and you saying “what difference does it make?”

What difference does it make? Also, where can I buy one?

Re: Eleven v3

#167

Post by horhay »

p1necone wrote: post #85

All of the examples sound like people doing scripted radio ad reads rather than natural speech. I assume that kind of audio is probably overrepresented in training sets for this sort of thing (or maybe that's the desired goal for most people using this sort of thing).

Training "high" points in voice inflection has been the priority, we've seen this in the 4o voice outputs and to some degree the Google NotebookLM podcast outputs. I would assume it's because they're trying to make it "act", but now it's a problem of swinging too hard on one end of the spectrum.

Re: Eleven v3

#168

Post by sebastiennight »

staticman2 wrote: post #146
saberience wrote: post #145

Earlier quoted context omitted.

Are you suggesting that you can compare a formulaic bank email to your mom reading you a bedtime story? I'm not sure you can connect those two things. Of course, when I go and check my balance at an ATM machine, I don't mind that an actual person isn't reading me the balance. But this isn't an area where we appreciate or want another human being involved. If you're a "normal", "well adjusted" human being, you appreci…

I promise this comment will circle back to Elevenlabs: When my cat died after a few months of cancer treatment, the staff of the animal hospital sent me a condolence card with comments by staff members. On the one hand, this was a very touching, very human thing to do. On the other hand, this was presumably a work assignment that had to be passed around and completed for staff members to meet their employer's goals,…

I have spent three weeks of my life recording my latest book as an audiobook. (125,000 words)

It was the most difficult experience of my life, ranking way above the pain of writing the book itself, and on par with month 1 of becoming a father. (I'm not joking.)

It was also an experience I'm incredibly proud of, and do not regret for one second.

AI audiobooks are the soulless experience. I see a use case of using AI for translating the audiobook, but generating it like that in the first place is a bit sad.

Post reply on HN