Because the Talalay and Dunlop processes involve vulcanization at 115+ C to turn the material into a foamed rubber, which denatures the proteins that the immune system recognizes and overreacts to. Denatured protein - think egg white once it's heated and turns white, instead of clear - has its structure radically altered. The molecules get pulled apart, tangled with others, and can in no way be recognized by the antibodies that trigger the immune response.
Similarly, Talalay latex mattress material is usually only about 30% natural and 70% synthetic, and the synthetic does not cause immune response.
If you powder the natural material and directly expose it to IgE, the dominant protein of interest for allergies, you can get a reaction (https://pubmed.ncbi.nlm.nih.gov/10436396/), but in practice with sheets and the outer cloth covering on the mattress basically no proteins ever come into contact with the body. And even in that study only Hev B I was detectable, which is only one of many latex proteins that trigger the immune response, and only 3 of the 21 tested human sera actually had a reaction to the direct mixing with the powdered latex. As far as I understand it, there has never been a confirmed case of an allergic reaction to a latex mattress.
For some reason, most of these (and other narration AIs) sound like someone reading off a teleprompter, rather than natural speaking voices. I'm not sure what exactly it is, but I'm left feeling like the speaker isn't really sure of what the next words are, and the stresses between the words are all over the place. It's like the emphasis over a sentence doesn't really match how humans sound.
Yup, and that's going to be the case until AI's can really model human psychology.
Speech encodes a gigantic amount of emotion via prosody and rhythm -- how the speaker is feeling, how they feel about each noun and verb, what they're trying to communicate with it.
If you try to reproduce all the normal speech prosody, it'll be all over the place and SoUnD bIzArRe and won't make any sense, and be incredibly distracting, because there's no coherent psychology behind it.
So "reading off a teleprompter" is really the best we can do for now -- not necessary affectless, but with a kind of "constant affect", that varies with grammatical structures and other language patterns, but no real human psychology.
It's a gigantic difference from text, which encodes vastly less information.
(And this is one of the reasons I don't see AI replacing actors for a looong time, not even voice actors. You can map a voice onto someone else's voice preserving their prosody, but you still need a skilled human being producing the prosody in the first place.)
What if you have it read the script, then say, “hey, at this point, what is the character feeling? What are they trying to accomplish? What is there relationship to each person in the scene?”
And then you get that and prompt the model to add inflection and pacing and whatever to the text to reflect that. You feed that into the speech model.
It seems like it could definitely do the first part (“based on this text, this character might be feeling X”); the second part (“mark up the dialogue”) seems easier; the third part about speech seems doable already based on another comment.
So we are pretty close already? Whatever actors are doing can be approximated through prompting, including the director iterating with the “actors”.
> What if you have it read the script, then say, “hey, at this point, what is the character feeling?...
Sure, but now how do you make sure all the answers to those questions are consistent? Across clauses, sentences, paragraphs? To do that, you need to have an entire understanding of human psychology.
And I haven't seen any evidence that LLM's possess that kind of knowledge at all, except at the most rudimentary level of narrative.
Just think of how even professional directors struggle to communicate to an actor the emotional and psychological feeling they're looking for. We don't even have words or labels for most of the things, and we say "you know how you feel in a situation when <a> and <b> but <c>? You know that thing? No, not that, but when <d>. Yeah, that." Most of these things operate on an intuitive, pre-verbal level of thinking in our brain. I don't think LLM's are anywhere close to being able to capture that stuff yet.
You have to shape the voices in the tools, if you just spit them out they're junk but if you take the time to shape the voice a bit it gets better quickly, this is a cheap 11labs voice with 30 seconds spent on some basic shaping: https://s.h4x.club/bLuNlJWx
Still a bit teleprompter-ish but there are tools to go in and adjust pace and style throughout and you probably hear a lot of stuff with people not using those creative features. 11labs might very well be one of the best bits of software I've used, it's a great deal of fun to play with and if you're willing to spend the time the results are superb - I don't even have a use case, I just like making them because they're fun to listen to, ha!
You---can, really--slow, speed up or change, how, things sound, by, -- using queues like this, to control how the voice,,, - tells the story {{3sec}} - once you find a voice you like, you can go in and {{1sec}}
PlayHT's voices are nowhere near as good as ElevenLabs. These self-reported studies are marketing.
In any case, voice is such a thin vertical that I half expect the Chinese to release an open source TTS model that out-performs everything on the market. Tencent probably has one of these cooking right now.
Oof. I've heard recent AI generated narrators, and they were OK (much better than a few years ago, much worse than professional humans), but something about the digital postprocessing in this article's youtube video reminded me of fingernails on a chalkboard.
This is an excellent point: there's some sort of oddity where it's hard to say its definitively AI, but I can definitively say it's a...low quality human?
I really like "off a teleprompter", it accurately characterizes the subtle dissonances where it sounds like someone who is reading something they haven't read before. 0:14 "infectious (flat) beatsss (drawn out), which is near diametrically opposed to the paired snappy 0:12 "soulful (high / low) vocals (high)."
I find myself disagreeing with many of the examples. E.g. according to the article:
Bad: It is quite difficult to find untainted samples.
Better: It is difficult to find untainted samples.
Bad: We used various methods to isolate four samples.
Better: We isolated four samples.
Something being quite difficult reads significantly differently than just being difficult. You haven't made the sentence better, you've changed the meaning.
And the fact that you used various methods instead of a single method is information missing from the second sentence.
> Something being quite difficult reads significantly differently than just being difficult. You haven't made the sentence better, you've changed the meaning.
The problem there is that the meaning quite carries can vary significantly depending on the reader or the context where the word is read (so it can read differently depending on how previous sentences have primed the reader, which means the same person might read it differently with that context than if they start at that sentence. Quite differently, in fact!
This is because in spoken form the word changes a lot with tone of voice. Technically "quite difficult" means "slightly difficult" but in many places actually means "damn near impossible".
I'd day that while removing the word isn't wrong, replacing it with a more specific comparison would be better.
And as I understand it, Brits are particularly fond of using "quite" in a sarcastic fashion, so "quite difficult" in England might mean "not difficult at all, you sodding idiot" or something along those lines.
Brit here, this isn't something I'm familiar with. "Quite" usually means "somewhat" as in "I found the test quite hard". In upper class speech it can mean "very" as in "that was quite the challenge" or "agreed" when said on its own as a response to a statement.
Interesting. Maybe it's a regional thing, or a generational thing. Or maybe I'm just flat out mis-remembering. Or maybe some of my British friends told me that, but they were just taking the piss. :-)
It's something I've come across references to more than a few times over the years though.
EDIT:
OK, FWIW, I can't find any solid reference at a quick glance to the form I was thinking of, but Google's "AI Search" GenAI thing does reflect what I was getting at, so I don't think it's completely something I made up. Unless me and the Google AI both hallucinated the same thing.
Here's what Google has to say:
In British English, when someone says "quite" with a
slightly sarcastic tone, it usually means they are
implying something is "not at all" or "very much the
opposite" of what they are describing, essentially
downplaying a positive quality to express mild
disapproval or skepticism.
Example:
"Oh, that new restaurant was quite good." (Meaning: it
was actually pretty bad)
"He's quite the brilliant mind." (Meaning: he's not very
intelligent at all)
I probably did overstate the degree of emphasis of this though.
> when someone says "quite" with a slightly sarcastic tone
The sarcastic tone is the secret sauce which makes the difference with a lot of words, including qualifiers like "quite". Try applying a sarcastic tone to "definitely" in the Earth is "definitely" flat and you'll see how people react.
Teacher: "There's plenty of languages where two negatives will negate each other and create a positive, but no languages where two positives will make a negative."
Student, sarcastic tone: "Yeah, right."
(I have no idea where I got this from, read it online ages ago)
A member of the British upper crust can correct me if I'm off the mark, but the definition there is of a really existing usage, and then the example doesn't match it at all. Did you make the example up yourself, by any chance?
There is, in ordinary people's language, "yeah, it was quite good", when talking about a movie or something, which could easily mean, it was moderately ok, not amazing in any way. It'll depend entirely on tone, you could say it in a chirpy tone and you'd mean that it was actually pretty good. This is the most common usage, and familiar to our brothers and sisters and non-binary-siblings across the pond, I suppose.
And then there's your mathematics teacher saying, "Oh, this lemma really is quite trivial", meaning it's very, very trivial, or a "quite difficult proof", meaning you've to drag yourself across hot coals for hours before it hits you.
Then there is also the meaning you describe above! E.g., a bunch of aristocrats are having dinner, and the candelabra suddenly breaks loose, flies through the air, and smashes into a thousand pieces with a crash. Luckily, no one is hurt.
Everyone looks around, shocked, there's a few shrieks of course, and then one of them says: "Oh, what a smashing evening!" and the other says, in a bored drawl, "Quite". It's like an additional layer of being removed from and above the mere idea that the original thing could have been worthy of a positive comment (in this case, the dinner).
Regardless, your point is valid. Adding a valueless word like “quite” does not improve clarity or meaning and can only have a negative impact. Not worth the risk.
> And as I understand it, Brits are particularly fond of using "quite" in a sarcastic fashion,
Brit here. Many of us are fond of using _any_ word/phrase with sarcasm, irony, or both.
> so "quite difficult" in England might mean "not difficult at all, you sodding idiot"
Depending on tone and other context "quite" can mean anything from a little to a huge amount. It can also mean exactly, as in "Well, quite.".
This is why you need to be careful in professional and academic contexts, or anywhere in writing for that matter, and use domain specific terminology as much as possible.
This is another thing that is captured in tone more than anything though the Brits do have a well deserved reputation for sarcasm. Difficult to convey in print what meaning you want the recipient to get.
So, completely difficult. But completely difficult doesn't sound quite right, probably as less syllables are preferred over many unless there's a quite good reason to prefer the latter.
You've omitted the definition on the lower part of the page:
quite
adverb, predeterminer
"a little or a lot but not completely:"
I'm quite tired but I can certainly walk a little further.
There was quite a lot of traffic today but yesterday was even busier.
It was quite a difficult job.
He's quite attractive but not what I'd call gorgeous.
It would be quite a nuisance to write to everyone.
The same dictionary also includes a grammar article clarifying that quite [usually] means "a little, moderately, not very", when the adjective or adverb it modifies is gradable (e.g "good" or indeed "difficult") and it being an intensifier in [generally rarer] situations where the adjective or adverb isn't (e.g "it is quite wrong to say that 'quite' invariably means 'exactly')
Great examples, but you should probably remove the "completely" header - as the following examples don't fall under it. I'll delete this comment in 15 minutes ( • ‿ • )
Not only is “completely” the definition they're quoting from the dictionary, it is also exactly what is exemplified by the examples, so I'm not sure what you mean by “don't fall under it”.
Ooooh, now I get it- I completely misunderstood that! Indeed, I can substitute every quiet with completely and it's meaning never changes from how it would've been interpreted!
(And I totally forgot to delete the comment too)
I just didn't realize that and only considered how I'd interpret the meaning of completely on is own. And that meaning doesn't translate to every example, hence my previous confusion
And this is why I now have to read 30 page design docs that could have been 3 pages and said the same thing.
Please try to understand why people have such strong dislike of floral writing, especially in technical texts. If you read a lot of papers or designs, it makes your life miserable.
Yes, it's the usual advice of how artists/authors/scientists make something: 1) Make the thing, 2) Try removing each part, 3) If the work fails without that part, put it back.
For example, adverbs are good when readers might have the wrong image without them. E.g., "Alice [quickly] walked." Most of the time, writing is better without words like "very" or "quite."
When it comes to technical writing the only thing I can really discuss is documentation, and the key thing I'm personally looking for there is structure.
It could be about basically anything, just please, pretty please, for the love of god, make it structured. And I don't mean sections with catchy headings, I mean as structured and reference-like as possible.
I want to minimize the amount of time I spend reading prose and searching around, as well as the chance of missing things. I want to hit CTRL+F and be put where I need to be stat and have that be enough. Structure alone can convey a lot of the idea behind how something works - please trust me to able to utilize it to make basic leaps in logic.
A bad example for this is AWS documentation. It's a mish-mash of prose and structured reference. A good example is the AWS CLI documentation (although if they lead with example usages first, that'd be even better).
Writing good technical text is an art. There is a certain amount of fluff that helps, and it’s almost unnoticeable when it’s there. Without it, it’s too terse. Quite often, my complaint of technical documentation is “it did exactly what the docs said it would do, except in a situation that I didn’t expect it to do that”.
Yeah most of his examples looked terrible to me. It's actually part of why reading papers is so damn difficult even when the paper says something simple. They're obsessed with this stilted formal tone that no one actually likes and leaves out subtle but important context clues.
I've enjoyed reading many of the blog posts by Antithesis, really cool work.
I don't really see a fit for the automated testing product in our stack at the moment, but I would love to use a time traveling hypervisor that I can hop into whenever I'd like.
Currently, it seems your pricing is pretty focused on the automated testing service. Do you have pricing or plans that offer just the deterministic dev environment?
(antithesis employee here) We don't currently just offer the deterministic dev environment, but we do offer extended 30 day demos for prospects interested in trying out the tech and seeing how it works. If you're interested contact us directly! contact@antithesis.com
As someone who is about to launch my SAAS, what are some of the things you look for, as someone with purchase authority?
What are the positives or negatives you see when dealing with vendors (i.e. what makes you lean towards or against dealing with a vendor)? I'm trying to figure out how I can make the process as easy as possible for exactly people in your position.
I'm not the person you've asked, but I'm somebody who has been purchasing SaaS/software for businesses large and small for years. My take:
1. If SSO and other basic modern security features are locked into "Enterprise" pricing tiers then the service is at the bottom of the list (see: https://sso.tax). I'd love to say instant disqualification but too many SaaS companies have it in their head that only wealthy enterprises use SSO, despite SSO platforms being widely available and some quite cheap to acquire and start using.
2. If I need to request a quote to start any kind of service to see what the product is about then I'm not likely to pursue it. Don't make me jump through hoops when I'm just trying to see if a product can fit my needs.
3. If license terms are too complex or easy to violate that's a hard pass. Infrastructure monitoring tools are a great example. The licensing is often per "device" or per monitored metric, and some vendors are very loose with their definition of "device". (Don't use LogicMonitor with k8s unless you like throwing money in the garbage can). Hard lessons learned.
4. If the only details I can find regarding how you secure your product are claims of SOC2 and ISO27001 certification then that's a very likely pass. Those controls are great to have, necessary even, but anyone who has had to work to meet those compliance objectives knows that they're much more about organization controls than they are product security. Give me an idea about how you protect data and whatnot on a security page somewhere, not an attestation that dev and prod are separate and you have logs.
On the side of the positives, outside of not hitting the negative marks, I value ease to work with, responsive and competent support, strong pre and post-sales solutions architecture and support/training (if the product is complex enough to warrant that), and supports SSO. I bring up SSO again because it's a hard requirement for SaaS purchases everywhere I go -- no SSO, no go. Social login is not a substitute and is highly undesired.
SAML is fine, no need for OIDC. Someone else mentioned SCIM, which is a highly desired addition, but not a hard requirement for me so long as a public API exists that I can use to automate the user lifecycle.
This is one of the best thought out UX I've ever seen. It's extremely well laid out and simple to navigate through, all the design choices are very meaningful, UI elements (like the unit conversion) are available inline when you need them...
I'd love to read about how he does it. How talented do you have to be to not only learn all this stuff but also model them in 3D and simulate them in real time and interactively?
Guessing from personal experience, I would not be surprised if 70 % of the domain knowledge is learned in the process of writing and modeling.
Whenever I encounter a tricky subject I'm having a hard time learning, I start writing an article explaining it to someone else. It really forces me to confront the gaps of my knowledge because I can see so clearly that "Wait, I can't explain what happens between these two steps here. What am I missing?"
Hah. That question comes up on every single one of his explainers. The answer is none; he hand-crafts the JavaScript and WebGL shaders. From the design language, I’d guess he has built up a library of templates and snippets to draw from by now, but all in all, each of these pages is a bespoke work of art.
A long time would be my guess but writing interactive articles is much more engaging and addicting. We've started adding interactive stuff to our docs and it is really engaging so the spent hours just fly by!
Also check out https://pudding.cool if you’re unfamiliar and enjoy extremely high effort visualizations alongside editorial and educational text content.
