Menu-bar app · macOS · Linux · Windows
Spotify in 432 Hz. Without converting a single file.
Netflix, YouTube, your own player just the same. The Tuner sits in the system audio — not in the file.
Also for Linux · Windows retunes files · SHA-256 Also for macOS · Windows retunes files · SHA-256 System audio is retuned today by macOS and Linux · SHA-256 On a computer: macOS or Linux. Test release 0.1.0 · free · check the SHA-256
The test · no download needed
Sound on, then drag from 440 down to 432.
Two sine tones in your browser. What pulses is the beating — two close tones cancelling each other in turn. That is how instruments have been tuned for centuries: you wait until it stops. The Tuner itself shifts the finished sound without touching the tempo.
What it does
System-wide
The sound passes through the Tuner before it reaches your speakers. On macOS and Linux.
Every app
Spotify, YouTube, Netflix, your own player. The Tuner sees the finished sound, not the file.
No converting
Nothing is converted, nothing is stored. The tempo stays as it is.
Verifiable
No voices, no stars. Measurements only.
There is not one single user yet — so we quote none. What there is: numbers from the build and from the measurement. They stand here together with why they count.
- 11 ms latency with the fast engine That is the line between “works with video” and “doesn’t”: the sound stays on the lips. If you would rather have the better sound than the sync, switch to the good engine — 240 ms, useless with picture.
- 0 allocations in the real-time path Not claimed, checked: the build script takes the finished library apart and lists every call that leads out of the real-time path. There are none. No memory, no lock — and so nothing that can click in the middle of a track.
- 88 tests, 436 assertions, zero failures Built on macOS 26 with Xcode 26.6. The number stands here because it comes about anew with every build — it is not a badge, it is a measurement taken today.
- 432.33 Hz measured — three systems, one result 440 Hz in, target 432: out come 432.33 Hz with the fast engine and 432.09 Hz with the good one. On macOS, on Linux and on Windows. Spread between them: 0.01 cents.
- 0.09 % of one processor core That is 1077 times real time. The fan does not notice anything is running, and neither does the battery — so there is no reason to switch the Tuner off in between.
- 0 requests to other people’s servers This page loads nothing from elsewhere: no font from Google, no analytics, no cookie. The Tuner asks for no account, no email address and sends nothing away. You download a file, and that is that.
- detected
- 443.1 Hz
- shift
- −43.9 cents
- engine
- delay line · 11 ms
Test release 0.1.0 · free · no account
Go and get it. Then three steps.
Go and get it. Then wave it through once.
On a phone it doesn’t work. Here is why.
The app itself: menu bar, takes the system audio and replaces it. The first time you switch it on, macOS asks for “record system audio”.
Icon in the tray — pitches under their names, on and off, choice of engine, start at login. Underneath, a LADSPA plugin in PipeWire’s filter chain. Built for Debian 13.
Measured over D-Bus: click on “Verdi tuning” → 432.331 Hz at the output device.
Its own window, tray icon, installer. Installs without administrator rights — no UAC prompt, no password.
Retuning files works. System audio goes through a virtual audio device where one exists — Windows allows listening in, but not replacing. The app tells you what it can do on your machine.
What happens next — on the Mac
Open the DMG, drag the app into “Applications”, double-click.
macOS turns it away. That is expected: this test release is neither signed nor notarised, and the right-click route that used to work has not helped since August 2024.
System Settings → Privacy & Security → “Open Anyway”. From then on it starts like any other app. The first time you switch it on, macOS asks for “record system audio”.
On Windows the same window is called SmartScreen. The route there: “More info”, then “Run anyway”. The reason is the same — the release is not signed. The installer itself needs no administrator rights afterwards.
The Linux package is not affected by any of this. apt install ./iyambae-tuner_0.1.0-1_amd64.deb — after that the icon is in the tray. Under GNOME it needs the AppIndicator extension once; the package recommends it, and the program tells you so instead of staying invisible.
Before you start it
Built and checked — 88 tests, 436 assertions, zero failures on macOS 26 — but not yet used by a human being in everyday life. That is the honest state of a test release.
Both files carry their SHA-256 sum. Check them instead of believing us: shasum -a 256 on the Mac, sha256sum -c on Linux.
If you ask
Everything else, folded away.
There for whoever asks. Out of the way for whoever wants to download.
What is 432 Hz anyway?
A concert pitch. It fixes where A4 sits, and everything else follows from it. The standard says 440 Hz; 432 Hz lies eight hertz below that, which is −31.77 cents, or just under a third of a semitone.
The Tuner shifts the whole sound by exactly that distance. It does not pick out single notes, and it changes nothing about the music except its pitch.
So where does 440 Hz come from?
From an agreement, not from nature. When Alexander J. Ellis gathered five hundred years of European tuning pitches in 1885, his table ran from 370 to 567 Hz — more than a fifth of spread. The reason was banal: organ pipes were built to the local measure of length, and the “foot” in France was not the “foot” in Saxony. Between the Paris tuning fork of 1822 at 432 Hz and the one of 1855 at 449 Hz lie thirty-three years — and the same city.
440 comes from 1834, proposed by Johann Heinrich Scheibler, a silk manufacturer from Krefeld. His reason was as unspectacular as it gets: 440 was the mean around which Viennese pianos drifted between warmth and cold. The conference that turned it into a world standard in 1939 met at BBC headquarters and preferred 440 to 439 because 439 is a prime number and could not be produced from the BBC’s quartz oscillator by chains of dividers.
Detail · 430 to 450 Hz this is where it all happens today
- 415Historically informed performance — exactly one semitone lower
- 432Milan 1880, the number Verdi argued for
- 435Diapason normal, Paris 1859
- 440Scheibler 1834 · London 1939 · ISO 16
- 443Germany and Austria, common today
- 445Berlin Philharmonic under Karajan
- 449Paris tuning fork 1855
- 470–480Highland bagpipe
“A piece of childishness that would almost lay itself open to ridicule.”
What does the research say?
Small studies, mixed results, no meta-analysis, no independent replication — inconclusive. That is where it stands, and we report it in both directions without varnish.
Aravena et al. (2020) compared 432 Hz, 440 Hz and no music before a tooth extraction; 42 people in three groups. For salivary cortisol the 432 Hz group was significantly lower than both the others: 0.49 against 1.35 and 1.59 µg/dL. The authors conclude that 432 Hz is “effective in decreasing salivary cortisol levels”. On the anxiety scale, however, 432 and 440 did not differ — there, only music beat silence.
Other work answers other questions. Jebabli et al. (2025) measured the anaerobic performance of kickboxers during their warm-up; there 440 Hz came out ahead. That is a clean study — but anyone who wants to know whether music relaxes you while listening learns nothing from it.
What can be said from all this: that listening to music relaxes people is well documented. Whether it is down to the tuning is open. The full texts are below under “What others write about it” — read them yourself instead of believing us.
Is this a medical device?
No. The IYAMBAE Tuner is not a medical device, not a remedy and not a therapeutic instrument. It is a tool that shifts pitch.
It is not intended to detect, treat or alleviate any illness, and we claim that nowhere. If something is wrong with you, see a doctor.
What about 528 Hz and the Solfeggio frequencies?
528 Hz is not a concert pitch but a single note — roughly a C5. A concert pitch fixes A4, and everything else follows from it. The Tuner shifts the whole sound by an interval; it does not set a single note.
It can still be worked out: if you want your C5 at 528 Hz, you set A4 to 444.0 Hz — the slider can do that. At A4 = 432 Hz the same C5 sits at about 513.7 Hz.
About the Solfeggio series itself we say nothing. We know of no source on it that we could pass on here in good conscience, and so we claim neither that it does something nor that it does not.
Which systems are supported?
macOS 14.4 and up — the complete app in the menu bar. Linux with PipeWire 0.3.60 and up — recently with a tray icon as well: pitches under their names, on and off, choice of engine, start at login. Underneath, a LADSPA plugin works in the filter chain. Demonstrated over D-Bus, the click measured: “Verdi tuning” → 432.331 Hz at the output device. Built for Debian 13; for Ubuntu and Zorin it has to be rebuilt.
Under GNOME the tray icon needs the AppIndicator extension once. The package recommends it, and the program says so at startup — instead of staying invisible and leaving the impression that nothing had been installed.
Windows: files yes, system audio not yet. There is its own window, a tray icon and an installer that needs no administrator rights. What it can retune there is files. Not the running system audio — the virtual output device for that is missing.
The reason for this order is technical: Windows lets system audio be listened in on, but not replaced. Anyone who wants to genuinely replace it has to become the output device. On macOS the capture interface does that; on Linux, PipeWire’s filter chain. On Windows it takes a virtual device — a driver of our own is the next step.
Why not iOS or Android?
Because that is where the sandbox draws its line, not where the effort runs out. On iOS there is no counterpart in any form. On Android you may only capture someone else’s audio if the playing app permits it — and Spotify, Netflix and YouTube do not.
What would be possible is something else: a player of our own that takes the music in hand itself. That is being thought about.
Why an app and not simply a website?
Because a website has to guess and the app can measure. A page in the browser assumes the recording lies at 440 Hz, calculates the fixed factor 432⁄440 and applies it to everything. With a recording that lies at 443 — the value common in Germany today — you end up at 435 instead of 432.
The Tuner determines the actual tuning of the running sound and works out the distance to the target from that. What lies at 443 is shifted 443 → 432, not 440 → 432. And it reaches the system audio, not just one tab.
- 1.85 %That is how far the time-stretch factor is from unity. For an octave it would be 50 %.
- ±0 %The tempo stays. Simply playing back more slowly would turn 120 BPM into 117.8.
- 0.09 %of one core with the fast engine — 1077 times real time, measured on Windows.
- 0allocations in the real-time callback. Not claimed: the build script takes the DLL apart and lists every call that leads out. There are none.
How much latency does it cost?
Shifting pitch without touching tempo costs time — and how much depends on the method. You choose.
- 11 msDelay line. Fast enough for video and for playing along. The sound stays on the lips.
- 240 msHigh Quality (Signalsmith). Sounds better, but is useless with video — a quarter of a second of offset is visible.
The switch crossfades over 25 ms so that the change does not click.
What has this got to do with the radio?
If the Tuner is running on the same computer, IYAMBAE Radio can notice it and switch off its own guessing retune. Then the app that measures takes over.
You are asked twice in the process: the page asks once, and the Tuner asks you. Without your consent no website learns that you are using it — not even that it is installed. A list of permitted websites in the program code would have been more convenient. It would only have answered the wrong question.
What does it cost?
This test release costs nothing. It asks for no account, no email address and no payment details — you download a file, and that is that.
Nothing has been decided about a price for a later, finished version. If that changes, it will say so here.
What happens to my data?
This page sets no cookies, measures nothing and loads nothing from other people’s servers. The fonts sit on the same server as the page — a link to a font collection in the USA would send your IP address there before you had clicked on anything.
The Tuner needs no account and sends nothing away. It processes the sound on the computer itself and records it nowhere. The only interface to the outside is the service for the radio on 127.0.0.1 — it does not leave your computer and answers only if you allow it.
The details are in the privacy policy.
Vier Fragen
Was sollen wir als Nächstes bauen?
Wir raten ungern. Vier Fragen, keine Anmeldung, keine Adresse — nichts davon lässt sich dir zuordnen. Eine Minute.
Wie hörst du heute in 432 Hz?
Falls überhaupt — „gar nicht“ ist eine gültige Antwort.
Was fehlt dir am meisten?
Das, woran du am häufigsten hängenbleibst.
Hast du für 432-Hz-Software schon einmal Geld ausgegeben?
Gefragt ist, was war — nicht, was wäre. Erinnerungen sind belastbarer als Absichten.
Was würdest du bauen, wenn du könntest?
Freiwillig. Der Satz, der uns am meisten hilft, steht meistens hier.
Danke.
Das fließt in die Entscheidung ein, was als Nächstes gebaut wird.
Other people’s sources
What others write about it.
Other people’s pages, passed on without assessment. We do not adopt their content as our own; their operators are responsible for it. Accessed on 21 August 2026.
-
de.wikipedia.org
Kammerton
Encyclopaedia article on the history of tuning pitch.
-
en.wikipedia.org
A440 (pitch standard)
English-language article on the 440 Hz standard.
-
pmc.ncbi.nlm.nih.gov · full text free
Aravena et al., J Appl Oral Sci 28 (2020), e20190601
Randomised clinical trial, 42 people, tooth extraction.
-
pmc.ncbi.nlm.nih.gov · full text free
Jebabli et al., PeerJ 13 (2025), e19084
Double-blind crossover study, kickboxing, warm-up.
-
pmc.ncbi.nlm.nih.gov · full text free
Jebabli et al., Sci Rep 15 (2025), 36120
Double-blind crossover study, athletes, warm-up.