The Physics of
Silence & Sound.

01

The Sound Your Blood Makes

Place a stethoscope over the artery of an arm with a cuff inflated above systolic pressure, and you hear nothing. The artery is squeezed shut. Now let the cuff deflate slowly. At a precise pressure, a faint, rhythmic tapping appears — blood forcing its way back through a partially closed vessel, turbulence striking the arterial walls.

That tapping is the Korotkoff sound. Each occurrence is brief — roughly 20 to 60 milliseconds — with a fundamental frequency between 16 and 50 Hz. Almost all of its acoustic energy sits between 20 and 300 Hz: the band where human hearing is least sensitive. The most important sound in cardiovascular medicine is, by design, nearly inaudible.

02

Five Phases, Two Numbers

In 1905, the Russian physician Nikolai Korotkoff described five phases in these sounds as the cuff deflates. They map directly onto the two numbers every blood-pressure reading is made of:

K1 — the first tapping: systolic pressure, the moment blood re-enters the arm. K2 — a turbulent murmur. K3 — the sound turns crisp and louder. K4 — it muffles. K5 — the sound disappears altogether: diastolic pressure, the artery finally resting open.

The fifth phase is the decisive one. International practice records diastolic pressure at the disappearance of the sound — and disappearance, unlike appearance, has no crisp edge. It fades.

Select a phase to hear its character — synthesised demonstration, not a clinical recording

03

Why the Clinic Never Left It

Electronic monitors have existed since the 1970s. Yet every major validation standard — AAMI, ESH, and today's ISO 81060-2 — still defines accuracy by comparison against a human listener with a stethoscope and a mercury column.

That is not nostalgia. It is because the listener is the reference: the ear hearing the Korotkoff sound is what "true" blood pressure means in these protocols. The clinic never left the method, because nothing else had earned its place.

AAMI · ESH · ISO 81060-2 reference method

04

The Ear That Couldn’t Be Copied

If a trained ear is the reference, why not simply record what it hears and let software decide? Three obstacles stood in the way for a century.

Loudness is perceptual. A microphone measures pressure; a listener judges loudness. Between 20 and 300 Hz, perceived loudness falls far below measured energy — and changes with frequency. A machine that amplifies the signal without modelling the ear will chase energy the ear never hears, and miss sounds the ear never misses.

The room is not silent. Clothing rustles, cuffs creak, a hand shifts. The listening must be weighted toward what matters and away from what does not — the same problem broadcast engineers solved with perceptual loudness standards.

The tail submerges. The final Korotkoff sounds do not stop; they sink into the pulse itself. There is no threshold below which the sound is "gone" — only a judgement, made the way a trained listener makes it.

05

How We Taught a Machine to Listen

Beat 1 does not hand the problem to software and hope. Its signal chain is analog-first: the sound is conditioned, filtered and shaped into its envelope entirely in hardware, before the processor sees anything. The processor’s only job is the final determination — by transparent, published rules.

Low power · no cloud · no hidden model — the listening is honest by construction

The decisive module is the one at the end: tail-phase detection. Where earlier attempts waited for an absolute quiet that never cleanly arrives, Beat 1 evaluates the distribution of residual sound energy across the tail of the measurement — separating a true disappearance of the Korotkoff sound from a sound that merely went quiet. It is the machine's equivalent of a trained ear deciding there: that is where the artery opened.

06

How It Will Be Proven

An honest claim about the gold standard must be tested the way the gold standard is tested. Beat 1 is undergoing a validation programme per ISO 81060-2, measured side-by-side against mercury sphygmomanometry with trained observers.

The acceptance criteria are the clinical ones: a mean difference within ±3 mmHg and a standard deviation within ±5 mmHg, across low, normal and high pressure ranges. The programme is in progress. We will publish the outcome when it concludes.

Validation programme in progress · data pending

07

Three Ways to Measure

Every blood-pressure technology belongs to one of three families. They are not competitors on a spec sheet — they are different answers to a different question: estimate, listen, or both.

Categories, not brands. Oscillometric monitors remain a reliable estimation technology; Beat 1 exists because estimation and hearing are not the same thing.

08

Questions, Answered Plainly