As no one on the planet (who owns a copyright anyway like me ) wants to upload training material to a snooping cloud …seems rvc may be a way to adapt to a standalone non-cloud program …such as Spectralayers in the future.
One of the many many wonderful aspects of Spectralayers is the fact copyright owners (like me) continue to maintain control over their material that resides on a local machine only.
rvc seems to have various implementations depending on which/what nvidia etc card…
I’m beginning to be in the camp that is a-ok with a dev just telling me specifically which exact combination of components to use …and hey…I’ll buy them…for this specialized type of work anyway.
Like in the early pro tools days.
The only downside is that code AND gear component advances are making such big leaps yearly now …
At any rate…as to rvc (or similar).
I may try rvc and/or pinokio myself…but this geek-level code stuff is often just beyond my area of interest.
I’d much rather stuff like this just find its way into programs like Spectralayers.
Voice cloning could be legally problematic in commercial applications.
There are commercial tools available which give you a small selection of voices (where the voice actors agreed that their voices can be cloned). But you cannot add your own voices.
There is a Vocoflex from Dreamtonics which can clone voices from short audio samples but in return you must identify yourself on camera and with your ID/passport to purchase the software and the output gets digitally watermarked so you can be hold liable in case you clone voices you’re not allowed to clone.
So there might be good reasons why it’s not available in tools like SpectraLayers or RX.
It is already problematic when cloud based as those are the operations currently being sued (successfully so far) for sourcing from sr protected works.
When/if detached from the cloud…ie…under control of individual user on local machine, the software manufacturer is at least out of the loop. Which avoids contributory action.
At any rate, the clone code is yet another tech genie out of the bottle, so use of it is guaranteed to increase.
Ha…unions couldn’t kill the Mellotron with those memorable lawsuits…this is same sorta story, different century
The logical legal defense for Generative AI will always be, that it wasn’t the company that offered the tools but the user who created infringing content with them.
This might work in the USA but not in Europe, for example, already decades ago European countries legally banned software that can circumstance copy protection. Next, web service providers or payment providers might cut a commercial company off, if their software allows to impersonate other people, for example by changing faces or cloning voices.
Of course, you might still be able to download and use the software, but there will be no commercial usage possible.