Your Cells Can Hear: What the Kyoto Sound Study Actually Found

You have probably seen it by now.

A carousel lands in your feed. Ten slides, dark background, clean type. It tells you that scientists at Kyoto University played sound directly into living cells and watched their genes change. Forty-two genes after two hours. A hundred and forty-five after twenty-four. Two master regulators of fat cell formation collapsing by 82% and 74%.

Nearly five thousand likes. Eight hundred comments. Someone you respect has shared it.

And you feel two things at once.

The first is a small lift in the chest — finally, someone measured it. The second, arriving a beat later, is the familiar tightening. Is this real, or is this another one of those posts?

That second feeling is worth trusting. Not because the post is wrong — as it happens, most of its numbers are correct — but because you have been burned before. We all have. The field is full of confident claims that dissolve the moment you look for the paper behind them.

So let's look for the paper.

Here is what I want you to have by the end of this: a clear account of what the Kyoto study actually measured, what it found, what it did not find, and exactly what you can say about it to a client, on your website, or to the sceptical GP at a dinner party — without ever having to walk it back.

The study is real, and it is good

The paper is Acoustic modulation of mechanosensitive genes and adipocyte differentiation, by Masahiro Kumeta, Makoto Otani, Masahiro Toyoda and Shige H. Yoshimura. It was published in Communications Biology — part of the Nature portfolio — on 16 April 2025.

It is peer-reviewed. It is open access, so you can read every word of it yourself without paying anything. At the time of writing it had been accessed over sixty thousand times.

This matters, because a great deal of what circulates as "sound science" traces back to a blog post citing a blog post citing a conference abstract from 1997. This does not. There is a real paper, from a real lab at a real university, and it says real things.

Kumeta has been working this question for years. His earlier study on the same problem appeared in PLOS One in 2018.

What they actually built

Here is the part most people skip, and it is the part that makes the whole study trustworthy.

The problem with testing whether sound affects cells is that sound is messy. If you put a loudspeaker next to a dish of cells, you are not testing sound alone. You are testing sound, plus the vibration of the dish, plus the vibration of the bench the dish is sitting on, plus whatever heat the speaker produces. When something changes in the cells, you cannot say which of those did it.

That was the weakness in the earlier 2018 work, and Kumeta's team says so plainly.

So they built something better. A vibrational transducer coupled directly to the culture vessel, delivering acoustic pressure into the medium without transmitting heat, mounted on a temperature-stabilised stage.

Then they chose their sounds: a 440 Hz tone, a 14 kHz tone, and white noise. They tested sine, triangle and square waves. They tested pressures of 10, 25, 50, 100 and 250 pascals, and settled on 100 Pa for the main experiments.

One small correction here, because it has been repeated widely. Several posts have said the team chose 440 Hz because it is the note orchestras tune to. The paper says nothing of the kind. Its stated reason is that 440 Hz and 14 kHz "were selected as representatives of low and high audible frequencies." No musical rationale at all. It is a low tone and a high tone, chosen to span the range we can hear.

That is not a criticism of the study. It is a correction of the story that grew around it.

Finding one: cells respond to audible sound

They exposed the cells, then sequenced everything that happened next.

After two hours, 42 genes had changed activity — 33 upregulated, 9 downregulated. After twenty-four hours, 145 genes had changed.

No drug. No added biochemical signal. No heat. Acoustic pressure in the audible range, and the cell's genetic activity shifted.

That is the headline, and it holds up.

One honest caveat, because we are being careful. The paper also notes that the 440 Hz vibration caused convective mixing of the medium — the fluid around the cells moved. So the cells experienced pressure and flow. The study does not fully separate the two. The 14 kHz tone produced much less flow and still produced responses, which helps, but the cleanest possible version of this experiment has not been run yet.

That is normal. That is how science moves.

Finding two: the mechanism

This is the part I find most interesting, and it is the part that almost never makes it into the social posts.

The study does not say sound affects cells by some unspecified resonance. It traces a pathway.

Cells anchor themselves to their surroundings through structures called focal adhesions. Think of them as grip points — the places where a cell holds on. Those grip points are mechanically sensitive. They register physical force and convert it into biochemical signalling. The field calls this mechanotransduction, and it is well-established cell biology, not a fringe idea.

What Kumeta's team showed is that the sound response runs through this system. Specifically, it depends on focal adhesion kinase activation, which drives Cox-2, which drives prostaglandin E2 synthesis, which produces the downstream gene changes.

So the answer to "how can a cell hear when it has no ear?" is this: it feels. Pressure arrives, the grip points register it, and a chemical cascade follows.

That is a mechanism. It can be tested, blocked, and argued with. It is a far stronger foundation than "everything is vibration," and it is the version worth learning, because it is the version that survives a sceptical question.

Finding three: the fat cells, honestly

Here is where the online versions of this story go astray — not by lying, but by choosing which true number to show you.

The most sound-responsive cells in the study were pre-adipocytes: cells that have not yet committed to becoming fat tissue. Under three days of acoustic stimulation during that decision window, two master regulators of fat cell formation dropped sharply. Cebpa fell to 0.18 of its normal expression. Pparg fell to 0.26.

That is 82% down and 74% down. Both figures are accurate. I checked them against the paper.

Now the number the posts leave out.

The actual effect on fat accumulation in those cells was a 13 to 15% suppression.

Sit with that gap for a moment, because it is the most useful thing in this entire article.

Two regulators dropped by more than three quarters. The thing those regulators regulate changed by about one seventh. Large upstream change, modest downstream outcome. This is extremely common in biology and it is precisely why gene-expression numbers make such seductive marketing and such poor evidence on their own.

Anyone who reads "82% down" and nothing else walks away believing sound had a dramatic effect on fat. It had a measurable, real, genuinely interesting effect of thirteen to fifteen per cent — in a dish, in mouse cells.

Both things are true. Only one of them is the whole picture.

Finding four: the result practitioners should care about most

This is the finding I would put on a wall.

The team compared different exposure patterns. Seventy-two hours of continuous sound. Two hours a day for three days. Single unbroken twenty-four hour blocks.

Two hours a day across three days — six hours in total — produced the same suppression as seventy-two hours of continuous exposure.

Six hours matched seventy-two.

And a single unbroken twenty-four hour block? No statistically significant effect at all.

Read that again. More sound did not mean more effect. Repeated sound, with silence in between, matched twelve times the total exposure. One long unbroken dose did nothing measurable.

If you have been practising or facilitating for any length of time, something in you already knew this. The regular short session outperforms the occasional marathon. The weekly sound bath does something the one-off intensive does not. We have said this to each other for years on the basis of what we observe in the room.

Here is a cell-culture study, with no interest in our field whatsoever, arriving at the same shape of answer through a completely different door.

That convergence is worth more than any single dramatic percentage. And notice that it costs nothing in credibility to say, because it is exactly what the paper found.

Finding five: the one that complicates the story

The team also tested what happens on second exposure.

Two hours of sound. Twenty-four hours of silence. Two hours of sound again.

On the second exposure, the cells responded at 26% of their original strength.

They adapted. They became less responsive, not more.

I want to flag this one carefully, because it cuts against a claim you will see attached to this study — that repeated practice makes a state easier to reach, that the body learns to respond more readily over time. That may well be true of nervous-system regulation in whole human beings. It is not what happened here. In this dish, on this timescale, repetition produced desensitisation.

An honest reading holds both: repeated short exposures beat one long exposure, and an individual cell's response weakened on immediate re-exposure. Those are not contradictory findings — they were measured on different things — but you cannot claim the first as evidence for a "sensitising" story while the second sits in the same paper.

If someone builds a claim about your daily practice deepening over two weeks and cites this study for it, the study does not support them.

What the study does not show

Now the part that requires some discipline.

This was cell-culture work. Cells in dishes. Primarily C2C12 mouse myoblasts, alongside nine other cell lines including 3T3-L1, HeLa, MCF7 and NIH3T3.

There were no animals in this study. There were no humans in this study.

Which means it does not show that sound healing works on the human body. It does not show that sound baths reduce body fat. It does not show that gong, voice, bowls, or any traditional practice is validated by this research. It does not show that the human body is a "standing-wave instrument." And it does not show that chanting works because it generates sound inside the tissue.

Some of those statements may turn out to be defensible one day on other evidence. None of them are defensible on this paper, and attaching them to it is how a credible study becomes a discredited claim.

The distance between "cultured mouse cells altered gene expression under 100 pascals of direct acoustic pressure" and "sound heals the body" is enormous. Every step across it needs its own evidence.

Why the honest version is stronger

Here is what I want to leave you with.

There is a habit in our field of reaching for the biggest available number, because the work feels so significant in the room that modest evidence seems like a betrayal of it. I understand that impulse completely. You have watched someone's breathing change under a gong. You have felt a room drop into stillness. Thirteen per cent feels like an insult to that.

But consider what happens next.

You share the 82% figure. A client's physiotherapist looks up the paper. She finds the 13 to 15%, and she finds that it was mouse cells in a dish. She does not conclude that you were careless. She concludes that sound healing is oversold, and she tells her patient that.

One overreach, and the credibility of everything else you said goes with it.

Now run the other version. You say: there's a 2025 Kyoto University study showing cells respond genetically to audible sound through a specific mechanism, and interestingly, short repeated exposures worked as well as continuous ones. It's cell-culture work — it doesn't tell us about the human body yet, but it's a real finding and the field is growing.

The physiotherapist looks it up. Everything checks out. She now thinks you are someone who reads papers.

That is worth more than any percentage.

We are not short of evidence for what sound does to the human nervous system — there is a real and growing literature on that, and it stands on its own. What we are short of is practitioners who represent it accurately. Being one of those is a professional advantage, not a limitation.

The takeaway

The study is real and it is solid. Peer-reviewed, open access, Nature portfolio, April 2025.

Cells respond to audible sound through a traceable mechanism — focal adhesions, Cox-2, prostaglandin E2. Not resonance in the mystical sense. Mechanical sensing in the biological sense.

The gene numbers are large and the outcome number is modest. Quote both, or neither.

Rhythm beat duration. Six hours spread across three days matched seventy-two continuous hours. One unbroken block did nothing. This is the finding closest to what we observe in practice.

Cells adapted on re-exposure. An honest reading includes that.

And no humans, no animals, no bodies. Dishes only. Everything past that boundary needs its own evidence.

The most credible thing you can do with a study like this is describe it accurately. It is more interesting that way, and it never has to be defended.

The Australian Sound Healers Association (ASHA) exists to hold that standard — grounded education, professional recognition, and a community of practitioners who take the work and the evidence seriously. You can join us here: https://www.australiansoundhealersassociation.com.au/membership

Leith James

If this landed for you, share it with one practitioner who needs to read it before they repost that carousel.