If there was a program that collected the barks, woofs, and whines of dogs along with a human description of what it means, could you process that data to create a sort of translator for dog speak that tells you the general idea of what they want? (e.g. hungry, glad to see you)
It would probably need an indicator for dog breeds since they will sound different. And obviously dogs don’t speak the same sort of language we do, but could general emotions be translated?
Think a little bit like those apps that (human) bird watchers catalogue bird sounds that are assigned to specific bird species. Many people have a good idea of what a dog wants/needs when they hear the noises it makes, so I can imagine something similar for dog owners. They record sounds of their dogs and what they believe it means, just as bird watchers do for mating calls.
Would this work at all? Or are the sounds of dogs too varying, even accounting for dog breeds? And how does its effectiveness change for cats and other noisy pets?
As someone who loves dogs and has had a couple of wonderful dogs, I don’t think this would work because dogs aren’t using sounds as a language. My experience is that they communicate as much or more with body language as sound.
My dog bark has a bark to get my attention but she shows me the reason. For example, bark while standing by and looking at food bowl means she wants food. If she wants to go out for a walk (always) she will bark and bring me her leash. If she wants me to follow her, she will bark and gesture or look in the direction.
Reminds me of a joke that someone invented a dog translator and all dogs say is “hey” but maybe they say it in different ways to mean different things.
“Hey, SHUT UP SHUT UP SHUT UP SHUT UP!”
I ran into someone that used that as their phone notification sound. I told them that I loved Ocarina of Time, but would have killed Navi given the chance.
Where are my testicles, Summer?
Frankly if you cant understand your dog yourself, an app probably wont help. They don’t talk but they are excellent communicators if you understand how to listen to them. This is especially true for dogs and cats but all animals have body language. Birds are special because they have very specific calls per species that you can learn to recognize, but they are so consistent that apps work too.
People saying dogs don’t communicate as much with sound… Are you a 1 or multiple dog household? Because while dogs communicate lightly with me through sound, my two huskies talk to each other a LOT, and if you watch the behavior and actions surrounding each vocal communication you can tell that something is being transferred, even if it is a really simple message.
Would be curious to know more specifically what they are telling each other.
Ya, some breeds vocalize way more than others. I had a Lab that was virtually silent, and now I have a Shepherd mix that vocalizes to the extent that it’s almost like words.
Most animals don’t use verbal communication. As an analogy, most humans don’t use sign language. Imagine if someone wanted to create a human translator that recorded what humans do with their hands and translate them into meaning. There would be a lot of noise there because most humans don’t talk with their hands, so most of it would be ascribing meaning that isn’t there. Sometimes the person is just itching their back, or just doing the dishes, etc, not trying to say something. Similarly, most other animals vocalise for other reasons, not because they want to make a specific sound to communicate something.
The other pets part is more likely to be useful.
It’s not that cats and dogs aren’t vocal, they are. Cats in particular supposedly are vocal for us, but don’t use sound with each other hardly at all.
Dogs are vocal with each other, and us.
The problem is that the level of complexity involved in their sounds is not just low, but only tiny part of their overall communication. Body language is way more important with cats or dogs. They communicate with their entire body, of which vocalizations are maybe 5% of the total. A bark, as example, can mean a handful of things by itself, but to know what they’re saying, you have to see the entire dog.
Trying to translate dogs and cats off of vocalizations only would be like trying to translate English using only adverbs.
That being said, I would put maybe a teeny bit of hope for dog vocalizations to be reliably translated whereas cats I wouldn’t believe it possible at all
Not in any objective sense.
But in a pragmatic or functional sense, yes.
Animals can communicate with and understand us, but not beyond anymore more than a toddler level.
You aren’t going to be able to have a discussion with your dog about existential angst or the emotional depth of The Pitt. And frankly, you won’t be able to do that with most other humans.
Difficult of this to work because animals don’t use sounds as words. They do communicate with sounds, but the only thing you will be able to extract is what you (hopefully) already are capable of understanding: basic emotions, pleading, pain, etc.
Some animals, on an individual level, have specific calls for specific things. But because it’s a single occurrence you won’t be able to gather enough hypothetical data for your project. And honestly if this is your pet, you are probably smart enough to figure out this type of calls.
Probably not for dogs or cats, but maybe for some species of birds and dolphins. Most animal noises don’t have the same kind of pattern that you find in all language. But those patterns do exist with certain bird songs and in dolphin vocalizations.
To indulge the thought I think you could get a fair way using LLMs and vision models (for body language) by looking at the vector space differences between human languages (and idiomatic [e.g. two fingers in different cultures] and intrinsic [e.g. smiles] gestures), a large set of annotated animal vocalizations and gestures, plugging it into a really hot cup of tea and burning a small forest.
More seriously, I think Cetaceans would be a more tractable problem, and perhaps from there other animals. Wouldn’t surprise me if it was already underway, while LLMs aren’t good for a lot of things techbros want them to be good at, they are good at modelling languages.
There was an (android) app to translate cat meows a few years ago … granted it was a bit frivolous, though they claimed underlying research.
The translation it produced of my cats meows into what they might want was plain low quality, the meows were easier to understand than the word soup produced. (Tying to get the cat to meow on command is a whole different factor)
I’m very confident that when my cats meow they are always saying a variant of “hooomaaaan!”
They’re like little eternally drunk roommates who walk around naked and are way too touchy for a household with children.
you should check out the new Nova episodes on dogs and whether the dog buttons we train them on can be considered communicative.
