nicholas-cc
Re:Hertz-dev, the first open-source base model for conversational audio
on Monday November 04, 2024 @04:10AM (#42039828)
We're working on a HuggingFace release that will help with finetuning. We'd like to do a paper, after a larger release - we're a team of 4.
Re:Hertz-dev, the first open-source base model for conversational audio
on Monday November 04, 2024 @02:30AM (#42039359)
Moshi is a good model to build chat applications on, this is designed to be more of a proper base model with all the quirkiness, naturalness, and researcher-friendliness of base modeling.
Re:Hertz-dev, the first open-source base model for conversational audio
on Monday November 04, 2024 @02:21AM (#42039309)
Yes
Re:Hertz-dev, the first open-source base model for conversational audio
on Monday November 04, 2024 @12:21AM (#42038734)
Hertz-dev is a base model, meaning it's just trained to predict the next token of audio. If your prompt is an old male voice with a British accent, the model will most likely continue speaking in an old male voice with a British accent. Being a base model, hertz-dev is easily finetunable for specific tasks - it would be a simple change to add manual configurations for the gender/age/accent.
Re:Hertz-dev, the first open-source base model for conversational audio
on Sunday November 03, 2024 @11:51PM (#42038567)
I'm one of the devs. Our model is fully voice-to-voice, no text was involved in the making of hertz-dev for exactly this reason.
Possibilities for low-fidelity mind uploading with EEG
Re:Big Trouble in Little Interconnects
on Tuesday January 03, 2023 @09:11PM (#34241059)
Looks like the author's company failed, but he's back working on similar technology? See https://www.linkedin.com/in/kevin-shambrook-a4884a4/.
The Evil Bit
An Introduction to Parsing Theory
Re:Vapor – Decentralized Web over Bitcoinized HTTP
on Wednesday October 07, 2020 @10:29AM (#24708594)
It seems like if you choose to record transactions to the blockchain you end up posting every transaction to the blockchain which is quite expensive; why not only store the hash of the HTTP request and use a Merkle Tree to store multiple requests in a single Bitcoin transaction? The only reason I see why not is that then the actual request data could be lost or not revealed, but since this protocol saves all requests to an internal database as well, I don't see why that would be a problem.
Standing in the shadow of giants – the frontier in open source
Ask HN: Can there exist a function whose computability is uncomputable?
We can mathematically prove that certain mathematical functions and constants are computable or incomputable. For example, we know that the Busy Beaver function is uncomputable, but that arithmetic is computable.Are there any functions for which their computability has been proven uncomputable, and if so, what are some examples and is there any term for such a function?
Re:Loglo: Experimental spreadsheet and stack language for vector graphics
on Wednesday June 17, 2020 @03:53PM (#23555945)
Have you seen Bret Victor’s substroke? Loglo really reminds me of this: http://worrydream.com/substroke/
Ask HN: Why Shouldn’t I Use Lisp?
I’ve been following people like Jonathan Blow, as well as Paul Graham. I’ve learned Lisp and kind of like it, but I’ve heard Jonathan Blow say that A) GC is bad, and B) not to use Lisp and other similar languages. Could you explain these reasons not to use Lisp/higher-level languages with more features?