Music is medicine.

Music as therapy is not a figure of speech. It slows the pulse, lifts the mood, carries grief that words cannot reach. Yet heard through cheap earbuds on a noisy train, the same recording arrives with a fraction of its force. Fidelity is not vanity. It decides how much of the feeling survives the journey to you.

That is a claim, and claims should be tested. So here is what the evidence says, as plainly as we can put it, including the parts that do not flatter the Hi-Fi industry.

What music does to the body.

Start with the chills. Most people know the feeling of a piece of music raising the hair on their arms, and for a long time it was treated as poetry rather than physiology. In 2011 a team at McGill led by Valorie Salimpoor and Robert Zatorre put listeners in a PET scanner with music they had chosen themselves and measured dopamine, the neurotransmitter that drives reward and motivation. They found it being released in the striatum at the moment of peak emotion, and in a quite different part of the same circuit during the seconds of anticipation before it. One pathway for wanting, another for getting, and music drives both on a schedule the composer set. The paper is in Nature Neuroscience, and its authors noted that this was the first demonstration of an abstract reward, something with no obvious survival value, producing the same chemistry as food or money.

That matters because it moves the conversation out of taste and into mechanism. Music is not merely something we like. It is something the body responds to with the machinery it uses for the things it needs.

What medicine has measured.

The clinical evidence is larger than most people expect, and the two best summaries of it are not written by enthusiasts.

In 2015 researchers at Queen Mary University of London pooled seventy three randomised controlled trials, close to seven thousand patients, on music before, during or after surgery. Published in The Lancet, the analysis found that music reduced postoperative pain, reduced anxiety, reduced the amount of pain relief patients needed, and raised their satisfaction, with no harm reported in any of the trials. Two details stand out. The effect held even when patients were under general anaesthetic. And the authors’ recommendation was that patients should choose their own music, because the effect follows the listener’s relationship with the record, not a prescribed genre.

In 2025 the Cochrane Collaboration, which exists to be sceptical for a living, updated its review of music based therapy for people with dementia: thirty studies, more than seventeen hundred people. Their conclusion was measured, as Cochrane conclusions are. Music probably improves depression and may improve behavioural problems; it does not appear to change cognition or agitation, and they say so. But the lead author’s summary was that the effect sizes make music a reasonable alternative to drugs, and a more person centred one, in a population where very little else works late in the illness.

So the word medicine is not a flourish. In surgery it reduces the dose of real medicine. In dementia care it does a job that medicine struggles to do.

The question nobody wants to ask.

Here is where it gets uncomfortable for a company like ours. Almost all of that research was done with whatever playback was to hand: hospital headphones, a speaker on a trolley, a care home stereo. The effects were real anyway. So is the quality of the chain beside the point?

The honest answer is that the evidence is thinner here, and mixed, and it is worth reading carefully rather than waving.

In 2016 researchers at Hong Kong University of Science and Technology ran listening tests on instrument sounds at several MP3 bit rates and asked people to rate their emotional character. The results were striking: compression strengthened the sad, mysterious and shy qualities of a sound and weakened the happy, romantic and calm ones. Lossy compression did not simply make music a little worse. It changed what the music was saying.

A team at Queen Mary, Ronan, Reiss and Gunes, went further and measured people’s bodies rather than their opinions, tracking physiological change, facial expression and head movement while listeners heard high and low quality mixes of the same ten songs. The finding is the one the industry should sit with: production quality had a clearly larger emotional effect on critical listeners than on casual ones. A good chain matters more to people who have learned to hear. That cuts both ways. It is a caution against overclaiming, and it is also evidence that hearing is a skill that grows with what you give it.

And a 2023 study by Vigl and Zentner, with more than three hundred participants, found that the intensity of emotion rose consistently with the quality of the performance, from amateur to professional renditions of the same piece. The better the music was delivered, the more people felt. Playback is the last stage of delivery. It would be strange if it were the only one that did not count.

Add the research on the loudness war, where listeners in blind tests consistently rated more heavily compressed masters as lower quality, and a picture forms. The signal arrives at a modern ear having been squeezed at mastering, squeezed again by a codec and finally pushed through the smallest transducer in the building. Each stage takes a little of the dynamic range, the texture and the space that the emotional effect appears to ride on. The research says music works anyway. It does not say the journey is free.

What we take from this.

Three things.

First, music does measurable good, in bodies, in hospitals and in people for whom not much else is left. That is the reason Step Deeper exists, and it is a better reason than the specification sheet.

Second, nearly all of the evidence for that good was gathered on ordinary equipment, which makes it a floor rather than a ceiling. Nobody has yet run the trial that asks how much more music can do when the chain gets out of its way. We would like to see that trial, and we would like to help build the equipment it is run on.

Third, the gap between a listener and a record is not fixed. The people who feel the most from a good mix are the people who have learned to hear it, and what you learn to hear depends on what you have heard. Our job is to shorten that journey, one component at a time, until the thing the surgeons measured and the thing a crowd feels are the same thing, arriving in full, in your own room. I watched that happen on a dance floor before I could explain it. That story is here. This one is the explanation.