Dogs actually struggle with talking buttons

If you’ve ever played your voice through a phone speaker and wondered why your dog seemed confused, you’re not alone. It turns out that how we speak to our dogs, not just the words we say, can make a big difference in whether they understand us.

A study published in Scientific Reports by researchers from Eötvös Loránd University in Budapest and collaborators in Japan takes a close look at this question. The team, comprised of Fumi Higaki, Tamás Faragó, Ákos Pogány, Ádám Miklósi, and Claudia Fugazza, wanted to know: do dogs recognize words played through different devices the same way they recognize them when spoken directly by humans?

Spoiler: they don’t. And the results have big implications not only for dog training but also for the booming world of “talking button” devices that promise interspecies communication.

Humans are remarkably good at understanding speech even when it’s distorted. We can make sense of a friend’s voice on a scratchy phone line, or follow lyrics in a muffled concert recording. Dogs, too, live in a world filled with human speech. They know their names, many learn action commands, and a rare few. The so-called “Gifted Word Learners” can even pick out dozens of toys by name.

But the thing is that most sound technology is designed for human ears, not canine ones. Humans typically hear in the 20–20,000 Hz range, while dogs can pick up frequencies up to 45,000 Hz, that is, well into the ultrasonic range. If a recording device cuts off certain frequencies, your ears may not notice, but your dog might lose crucial information.

And this isn’t just an academic issue. Over the past few years, “talking button” devices have exploded in popularity. Inspired by viral TikTok clips and books like "How Stella Learned to Talk", thousands of dog owners are trying to teach their pets to “speak” by pressing buttons that play back pre-recorded words like “outside” or “food.” But are dogs really recognizing those words, or are they just learning to associate a button sound with a specific action? That’s where this study comes in.

The experiments: dogs, speakers, and talking buttons

The researchers designed three separate studies to test how dogs respond to words delivered by different sources. The first study involved action commands. 17 typical pet dogs were trained to perform actions like “sit,” “spin,” or “down.” The commands were given in five ways: directly by their owner, through a smartphone-and-speaker setup, through a “talking button,” and with or without the owner silently moving their mouth. In this case, dogs nailed it when their owners spoke directly. They did okay with the speaker. But with the button device, the kind sold for pet communication, their performance plummeted. Study 2 involved the “Gifted Word Learners,” a small but fascinating group of dogs (often Border Collies) who know the names of dozens of toys. These dogs were asked to fetch specific toys when their names were given through the same devices. Again, live human speech worked best. The loudspeaker was okay. The button? A 70% drop in success. Even the most language-savvy dogs struggled. In study 3 dogs were tasked with learning new words from recordings. The question was "Could these super-learners acquire new toy names from a recording instead of a human voice?" And the findings say that they could, but only partly. When owners used a speaker to teach toy names, the dogs learned, but much less effectively than when taught with live speech. Interestingly, when later tested, the dogs actually did better if the owner spoke the word directly, suggesting they could transfer what they’d learned from degraded sound back to real speech.

The issue with talking buttons

So why do these popular devices fail dogs so badly? The answer lies in sound frequencies. The researchers compared the audio quality of the devices by playing “white noise” through them and analyzing the results. Speakers (like a JBL portable) preserved most of the sound spectrum, though they still lost some detail. Buttons (like the FluentPet Classic), however, chopped off huge chunks of the sound range, especially frequencies below 1000 Hz, where the fundamental “body” of human speech lives. For us, that degradation might sound a little tinny but still recognizable. For dogs, it’s like trying to understand a word with half the letters missing. No wonder they don’t recognize it.

If you’re a pet parent hoping to use buttons to “talk” with your dog, the sobering news is: your dog probably isn’t hearing the word you think they are. Instead, they may be learning that pressing a specific button produces a specific sound, which then leads to a specific outcome—much like a doorbell makes you check the door. That’s not the same as recognizing a word. This doesn’t mean the devices are useless. Dogs can still learn associations with the sounds. But if your goal is language-like communication, the current technology falls short. On the flip side, the study shows that dogs can generalize from recordings to real speech, at least the talented word-learners can. That suggests dogs are flexible listeners, but they need clear, rich sound to do their best.

As humans, we often assume that if technology “works for us,” it works for animals. But dogs experience the world differently. A cheap microphone that trims frequencies might not bother your ear, but for your dog, it’s the difference between a meaningful word and meaningless noise. So, if you call your dog through a tinny Bluetooth speaker or expect them to respond to a recorded command, remember: your pup isn’t ignoring you, they might just not recognize the sound as your voice at all.

If you want to read more, the article "Sound quality impacts dogs’ ability to recognize and respond to playback words" is available on Scientific Reports at http://dx.doi.org/10.1038/s41598-025-96824-8