Loggings > North American Shortwave Pirate
Art Bell Stretchyman Parody 9290 USB 1452 UTC 3 DEC 2024
Charlie_Dont_Surf:
--- Quote from: ChrisSmolinski on December 03, 2024, 1452 UTC --- Art Bell parody (I assume assembled with clips) complaining about his Stretchyman transmitter.
--- End quote ---
Perhaps assembled from clips but man, that might mean a hell of a lot of time to put this together and an encyclopedic memory of his old shows. I don't even know if the slang word "randos" was in use when Art was on the air.
I don't know much about the current state of affairs in AI, but what about AI trained from his old recordings and then fed a script as a "deep fake"? If they can do it for politicians' voices, why not Art's voice?
RobRich:
There are several "AI" voice cloning apps, services, and open source tools. YMMV regarding actual results.
I have not (yet?) used it, but for an open source option, CoquiTTS with XTTS-v2 appears to have a web-based public demo install at huggingface.
https://huggingface.co/spaces/coqui/xtts
https://github.com/coqui-ai
Otherwise it can be done locally, though you would likely need a decent GPU. Same for Bark and lots of other open source voice AI software. Search Github if interested.
As for proprietary options, ElevenLabs is supposedly pretty good, but its free option is limited to like 10 minutes a month.
Teotwaki:
How they did it is one thing but getting it on the air on two different frequencies is epic.
Charlie_Dont_Surf:
--- Quote from: Teotwaki on December 03, 2024, 2309 UTC ---How they did it is one thing but getting it on the air on two different frequencies is epic.
--- End quote ---
There are commercially available general coverage SSB tranceivers for much less money than in years past. They are generally QRP but that's not a big hindrance if you have the $$$ for some amplification (which also is pretty cheap, especially if you are handy at building kits).
Then there is plenty of info on general-coverage modifications for some of the more popular ham rigs. One that comes to mind is the Yeasu FT-817 (and maybe the FT-818?); remove a few co-located SMD resistors and you are in business, if I recall correctly.
So the frequency agility doesn't knock my socks off. Some of the "regulars" on 43 meters (e.g., WDOG and Thunder Chicken Radio) have been known to show up on 31 meters (or so) occasionally.
Had "Art Bell" been on two frequencies simultaneously, that would have been more impressive to me. ;)
Charlie_Dont_Surf:
--- Quote from: RobRich on December 03, 2024, 2126 UTC ---There are several "AI" voice cloning apps, services, and open source tools. YMMV regarding actual results.
I have not (yet?) used it, but for an open source option, CoquiTTS with XTTS-v2 appears to have a web-based public demo install at huggingface.
https://huggingface.co/spaces/coqui/xtts
https://github.com/coqui-ai
Otherwise it can be done locally, though you would likely need a decent GPU. Same for Bark and lots of other open source voice AI software. Search Github if interested.
As for proprietary options, ElevenLabs is supposedly pretty good, but its free option is limited to like 10 minutes a month.
--- End quote ---
And there you have it. We're on the doorstep of AI-generated piracy.
Navigation
[0] Message Index
[#] Next page
[*] Previous page
Go to full version