Pleasantly surprised to see this on the HN homepage! Our agency (Longbeard) rebuilt this website a few years ago. It was fun working with the great people there, including Br. Guy who is a fantastic ambassador for the VO.
As you can imagine the Castel Gandolfo telescopes are mainly historical at this point due to light pollution, so their VATT facility in Arizona (Mount Graham) is now where most of their actual astonomical work is conducted.
Interestingly, the VATT in AZ is directly adjacent to the LBT facility on Mount Graham, which has a near-infrared instrument formerly named - believe it or not - LUCIFER (Large Binocular Telescope Near-infrared Spectroscopic Utility with Camera and Integral Field Unit for Extragalactic Research).
This was changed to LUCI in 2012 as the name predictably caused some problems and confusion.
Back when I was in grad school I observed at LBT (or maybe more accurately I was on a team that was observing at LBT). The Vatican's observatory down the road was affectionately referred to as the "Pope Scope."
Although I know the church has a lot of money, I can't imagine they dedicate a particularly large budget to the observatory compared to, say, a huge research university or a specialized scientific research project.
Openly musing, I'm just curious how they decide which projects to pursue, and how they contribute to the community? I know nothing about astronomy, but I imagine like many areas of research, they have conferences on various topics, so perhaps the VO scientists participate in conferences relevant to their research. Which again, of course, brings up the question "how do they decide what to research"?
> Although I know the church has a lot of money, I can't imagine they dedicate a particularly large budget to the observatory compared to, say, a huge research university or a specialized scientific research project.
The VO has about the same share of the State of Vatican City's budget as NASA does of the US budget (but the Vatican Advanced Technology Telescope has a separate foundation, largely funded by private donations, that funded its construction and funds its ongoing maintenance, without going through the general Vatican City budget.)
The church chose a mildly ridiculous, astronomically difficult and contentious definition of Easter: the first Sunday after the first full moon on or after the spring equinox.
So it combines 3 complex systems: sun, moon and calendar.
Religious extremism and pedantry increased the priority of accuracy and agreement. For some reason, Christians argued (fought wars) over the definition of Easter, and then regarded anyone picking the wrong day as a malevolent heretic.
I imagine Easter could be one of the issues that led Swift to write the Lilliputian egg dispute in Gulliver's Travels.
I think vaticanobservatory.org is run by the Vatican Observatory Foundation (a private foundation supporting some aspects of the VO's work) whereas vaticanobservatory.va is run by the Vatican Observatory in the strict sense (the government department of the State of Vatican City.)
> I think vaticanobservatory.org is run by the Vatican Observatory Foundation (a private foundation supporting some aspects of the VO's work) whereas vaticanobservatory.va is run by the Vatican Observatory in the strict sense (the government department of the State of Vatican City.)
This exactly. When our agency was working with them, there was some initial hope that we could consolidate the websites on the .va property, but we quickly realized that was not going to happen, primarily due to the separation required with the foundation wing.
Also, it is in general very difficult for an outside vendor to get clearance to build on a .va domain, and you need someone internally to apply a lot of leverage to get the wheels spinning with Vatican IT. Our agency was able to do this with building out the migrants-refugees.va website, thanks to the help of (now Cardinal) Michael Czerny who ran the M&R Section at the time and was given a lot of direct executive power from Pope Francis, but boy, that was still not easy.
Funny timing as I just revisited the brilliant "Step Right Up" (and the entire Small Change album) a few days ago, and had no idea about this case. Glad Tomcat won. Can't help but see some parallels between this and current Scarlett Johansson vs. OpenAI debacle.
My very favourite line, among many, from "Step Right Up":
__The large print giveth and the small print taketh away.__
It surprises me a little because the Dorito commercial sounds like a fairly generic blues backing track and a raspy voiced singer. From the article, there's an indication that Doritos was looking to replicate the sound and feel of the song, and they were aware of the legal concern. But to me it kind of sounds like Tom Waits is doing an impersonation, the Doritos song is doing an impersonation, but it's not totally obvious from the song that the Doritos impersonation necessarily goes via Tom Waits rather than back to the original source material.
I used to imitate Tom Waits when I'd play Tom Waits songs on the piano. And Leonard Cohen when I played his songs. But after 30 years of smoking I finally achieved my long term goal of just sounding like a cross between the two of them without trying. You can't really own smoker's growl. The best you can do is rent it for a few years.
I didn't read the case, but in a lot of these type of cases, there are memos or emails that make the case more than if you were just listening to the commercial. Like in the Scarlett Johansson case where they reached out to her to do the voice. The may have had internal memos that called the commercial the "Tom Waits Commercial" or the like. They may have had internal legal memos advising that they could be sued, etc.
The commercial was using a tom waits song sung by an artist who impersonated Tom waits and the folks involved discussed the potential legal implications and did it anyways.
I did read the article and listened to the Doritos song as I mentioned in my comment. I agrree the statements about how they came up with the song are damning, as I mentioned. But they aren't doing a Tom Waits song.
The logical parallels might be there but in legal substance they are different cases. A deliberate replication of an artist's distinct singing voice, while singing lyrics associated with that artist, is very different than a passing similarity to someone's non-unique speaking voice. Note that the Johansson issue is playing out only in the court of public opinion.
If you think they weren't trying to replicate sj's voice, I can only assume you've got open ai stock clouding your judgement... It looks like they caaaaarefully avoided writing down 'sound like sj' while choosing+directing a sound-alike actor for the role.
>> If you think they weren't trying to replicate sj's voice
Coevolution. The AI and the now-famous actor are the end products of a selection process aimed at a similar audience. The AI creators were likely aiming for the most attractive female voice, according to a bunch of young men. That parallels the process used by those casting action movies, which are also aimed at young men. It should be to nobody's great surprise that the end products of both share many features.
What is very odd is how similar celeb faces actually are. There really is a mathematical standard for beauty when it comes to female faces, which is why so many celebs are have eerily identical proportions. There is probably a model too for voices.
Interesting that the OpenAI astroturfing brigade is coming out of the woodwork.
You do know of course that they asked to clone her voice. And when she said no did it anyway, then asked again immediately before release, and then Sam tweeted that it was her?
And then when her lawyers ask "So who did supply the voice for this?" they immediately take it down, and immediately hand over a trove of "proof" to WaPo, most of which is so flimsy it'd fall over in a small breeze.
Her agent, who has requested anonymity...
Tells WaPo that his client, who also "needs to remain anonymous" due to "fears for her safety" tha she told him that OAI "didn't mention SJ".
Tells WaPo that she gave him a statement that they can print...
What are these fears for the actress' safety that require her, and her agent, to remain anonymous?
Why was OAI ready with a trove of legal-reviewed documents to hand to WaPo the moment this blew up?
> Note that the Johansson issue is playing out only in the court of public opinion
It takes a while to get a lawsuit going. According to the article, Waits learned about the ad on October 3, 1988, filed a suit in November 1988, and the case was tried before a jury in April and May 1990.
There's another case with bette Midler and I think it was a Superbowl ad. They approached Midler, she said no, they hired a sounds like, she sued and won.
With sj, they approached her 6 months ago, she declined, and again just prior to launch to which she didn't respond. And Sam tweets one word 'her'.
"Passing similarity"? All her friends and coworkers say it sounds like her - she describes being contacted by friends who thought she'd done it. A lot of the public, including HNers, identified the voice as hers, independently - not in some A/B test, but they heard it and said to themselves "Johanasson did the voice for this."
You're the only person in the room who thinks OpenAI didn't use samples of her voice for their voice model - or that the voice has "passing similarity."
Altman approached her asking her to do it. She refused; they went ahead and did it anyway, probably using interviews since the audio would be very clean - and two days before they went live with it, Altman tried to negotiate a second time and was rebuffed. Released it anyway.
Why would Altman feel the need to panic-negotiate a second time, days before they went live, if they hadn't used her voice for it?
If OpenAI designed it to sound like her, that's Midler tort. They actually used her voice to train it, which means Midler tort and more. The question will be how much more - I hope they get sued into the ground.
> You're the only person in the room who thinks OpenAI didn't use samples of her voice for their voice model
Really? No one seems to be claiming that anywhere I've seen.
We know that they really did use a voice actress, and that it really is her actual voice, and that she was hired for it before ScarJo was approached. There's no "panic negotiation" there -- the most panic needed would be if someone internally identified that there could be a PR issue if people mistakenly thought it was close enough to seem like intentional impersonation.
Of course they wanted ScarJo, once they thought of it! It'd be great publicity. But that doesn't then mean that anyone who sounds somewhat like ScarJo (but certainly not identical) becomes retroactively unusable or makes it vocal impersonation.
> But that doesn't then mean that anyone who sounds somewhat like ScarJo (but certainly not identical) becomes retroactively unusable or makes it vocal impersonation.
yes. yes it does. That's what TFA is saying. Waits' case is exactly about this. They hired a singer to make a recording to sound like Waits. OpenAI hired a voice actress with a voice like Her to make their AI sound like Her. And then their CEO tweeted "her". It's that simple.
Ford hired a singer to impersonate Waits in a parody of a Waits song in his exact style.
OpenAI hired a voice actress with a pleasant, low, breathy voice who shares a hint of a regional accent with ScarJo to be one of a bunch of voices for their models. Eight months after releasing it, the CEO tweeted "her." It's not illegal to hire someone who sounds a bit like a celebrity for something, and certainly not one who literally sounds more like a different celebrity (Rashida Jones).
This isn't a violation of rights of publicity for a number of reasons, but the most dispositive of them is that she doesn't actually sound that much like ScarJo. There's no vocal fry! That's the most distinctive part of her voice!
Pt 1. https://www.youtube.com/watch?v=2BNHHy5etQc
Pt 2. https://www.youtube.com/watch?v=3FYYB9kqkE4