Help & manual

Everything the studio does, why, and where every number came from. Read it in order the first time, or jump to a chapter below.

Contents

1 · What these tones are

Four different things get called “binaural beats” in casual use. They are physically different, and one of them survives a loudspeaker while another does not.

What the body of research actually supports

A periodic sound at rate R produces a measurable neural response phase-locked to R while it is playing — the frequency-following response and the auditory steady-state response. That effect grows with level up to about 80 dB SPL, and it is established. What is not established is the second half of the popular claim: that listening to a rate for a while durably changes your brain state. The 2023 systematic review of 14 EEG studies found 5 in line with the entrainment hypothesis, 8 contradicting it, and 1 mixed, concluding there was no clear trend either way. The evidence page has the citations and a claim-by-claim ledger.

2 · The studio, part by part

The rule that explains everything about the editor

3 · The library: packs and sessions

The built-in library is original to this site: every session is a tone schedule written here, with no audio or text from any recording, programme or product. The packs are themed rather than clinical.

Upstream Gnaural's preset archive is deliberately not bundled — it carries no licence. Use Open .gnaural for files you already have.

4 · Reference: every number in the quick picks

These tables are generated from the same data file the studio's quick-pick buttons use, so the page and the buttons can never disagree. The status column is the honest one: what a number is.

EEG band names (descriptive conventions)

Schumann resonances — measured physics

The resonances are real electromagnetic phenomena in the Earth–ionosphere cavity. Playing an audio tone at 7.83 Hz represents the number; it does not couple your living room to the ionosphere.

Resonance breathing — paced practice

Around six breaths a minute the cardiovascular feedback loop is usually said to resonate; that is why the rate is called the resonance frequency. The HRV literature supports short-term changes in heart-rate variability; it does not support outcome claims. The Pacer tool in the studio is the visual version of these rates.

Focus-level layouts — our own schedules from published measurements

Notation: carrier[beat] — a beat of beat Hz around a carrier of carrier Hz. These layouts are original schedules written for this site from publicly documented frequency sets; see the next chapter.

Solfeggio and other historical lists

These are a 1990s folk system (Puleo & Horowitz), not a scientific one, and the popular “528 Hz repairs DNA” line has no peer-reviewed support. They are pleasant steady tones with interesting provenance, and that is exactly how this site treats them.

Noise beds

5 · Focus levels and the Gateway tapes

The best-known body of work in this field is the Monroe Institute's Gateway programme and its Hemi-Sync audio. It is worth knowing exactly what it is, because it is the source of most of the vocabulary that circulates — and because it is where a lot of confident but unsupported claims come from.

How this studio relates to it

Practical note: the programme is a headphone practice. Every focus-level preset here is a binaural layout, so it needs headphones for the beat to exist at all. If you want the speaker version of the same idea, use the isochronic quick picks or the Speakers pack.

6 · Schumann resonances, solfeggio and the rest

Schumann resonances

In 1952 Winfried Otto Schumann worked out that the cavity between the Earth's surface and the ionosphere acts as a resonator for extremely-low-frequency electromagnetic waves, and the resonances were measured from the 1960s onwards — the fundamental sits near 7.83 Hz, with higher modes around 14.3, 20.8, 27.3 and 33.8 Hz, all drifting slightly with ionospheric conditions. Every lightning strike is broadband excitation.

They are real. What is not real physics is the step that usually gets made next — that a tone at 7.83 Hz from your speakers connects to, tunes into, or harmonises with that cavity. Audio is pressure waves in air at 20 Hz–20 kHz; the cavity resonance is an electromagnetic mode at 7.83 Hz. They share a number and nothing else. The studio's Schumann entries are labelled physics for the number and make no claim about the tone.

Solfeggio, 432 Hz and the historical lists

The solfeggio set (174, 285, 396, 417, 528, 639, 741, 852, 963) comes from a 1990s book by Joseph Puleo and Leonard Horowitz, built on a numerological reading of the syllables of a Latin hymn. It is a modern invention with a retrofitted history, not a medieval tuning. The “528 Hz” entry in particular carries a widely repeated DNA-repair claim with no peer-reviewed support; the studio labels the whole set tradition and says so on the page.

The A = 432 Hz debate is a different case: there is real history there (orchestral pitch has wandered all over Europe for centuries, and 440 Hz is a 20th-century standardisation), but the audible difference between 432 and 440 is simply the interval between two pitches. Nobody has demonstrated an effect beyond that.

Resonance breathing

Slow paced breathing near six breaths a minute (~0.1 Hz) sits at the frequency where blood-pressure oscillations reinforce the breathing rhythm through the baroreflex — the “resonance” in the name is that feedback loop, not a mystical one. Studies of paced breathing show changes in short-term heart-rate variability; claims about longer-term health outcomes are where the evidence thins out. The studio's Pacer tool and the 0.1 Hz quick picks are for doing it, not for proving anything.

Rife machines, “angel frequencies” and what is not here

Royal Rife's 1930s frequency lists, the “angel” frequency sets, chakra-tone mappings and similar material circulate widely. Rife-type devices have been the subject of US Food & Drug Administration warnings and cancer-quackery rulings, and there is no controlled evidence for the frequency lists. This studio deliberately does not ship those as preset packs: a tone generator has no business dressing a list like that up as a feature. If you want to play a specific number, the manual layer builder takes any rate in Hz — with no claim attached.

7 · Headphones, speakers and rooms

The Speakers pack exists because a loudspeaker version of this idea is a genuinely different design problem, and the differences are worth knowing:

8 · Building your own session

1. Start from something close. Load a library session, then bend it. Everything is editable.
2. Add a layer. Use a quick pick (one click, with the right type chosen for you) or Build a layer by hand for any rate and carrier you like. Quick-pick layers are added at the length set in the tools panel.
3. Edit the entries. Click a voice in the Voices panel; its entries appear as a table of numbers. Duration, carrier, rate, and the two channel volumes are all editable in place. Add or delete entries with the buttons beside each row.
4. Respect the wrap rule. Values ramp into the next entry and the last wraps to the first — so a hold means repeating a value, and a fade-out means the following entry is zero.
5. Check the graph. The stage graph is the fastest way to see a mistake: a bed that should be flat and is not, a rate that slides when it should hold.
6. Listen at a low level before you commit to a long session, and use the timer if you are going to fall asleep to it.
7. Keep it. Save .gnaural writes the file, Export WAV renders audio, and Copy share link puts the entire session into the URL — paste it into a message and the recipient gets the same session, still fully editable.

Recipes worth trying

• A speaker session for a room: brown noise bed at 0.04, plus a 10 Hz isochronic pulse on 300 Hz at 0.12, plus a slow fade-in over two minutes.
• A desk-focus hour: load a speaker-focus session, then add a 40 Hz pulse on 320 Hz for the first ten minutes only (short duration entry, then delete-then-hold the rest).
• A gentle landing: an alpha-range pair (10 Hz around 200 Hz) for 10 minutes, sliding to a theta-range 6 Hz for 15, with the last three entries fading out.
• Paced breathing: start the pacer at 6 a minute, and add the 0.1 Hz pulse quick pick so the room keeps the pace for you while you breathe.

9 · The .gnaural file format

The studio reads and writes Gnaural's own XML format (gnauralfile_version 1.20101006), so files round-trip with the desktop program. The shape is simple:

<schedule>
  <title>My session</title>
  <loops>1</loops>                    <!-- 0 = loop forever -->
  <voice>
    <description>pair</description>
    <type>0</type>                  <!-- 0 pair, 1 pink, 3 pulses, 4 alt pulses, 7 white, 8 brown -->
    <entry parent="0" duration="60" volume_left="0.12" volume_right="0.12"
           beatfreq="7.83" basefreq="200" state="1"/>
  </voice>
</schedule>

• duration is seconds; volume_* is a linear 0–1 gain; basefreq is the carrier in Hz and beatfreq is the true rate in Hz (the engine halves it internally, as upstream does).
• Types 7 and 8 (white and brown noise) are this project's extension. Upstream reserves 5 and 6 for sample-based water and rain, which this port does not implement. A file using 7 or 8 plays here and is reported as an unknown type by other Gnaural builds — the site never writes them into a file unless you built a bed with them.
• Values are written with up to six decimals; nothing is rounded in a way that changes a rate.

10 · Taking it elsewhere

• The desktop Gnaural (the program this engine follows) opens anything you save here, and its presets open here.
• Any audio tool can take the WAV export — it is rendered in your browser at 44.1 kHz, exactly the schedule you see.
• DAWs and controllers: a rendered WAV drops into any arrangement or sampler. The live scheduling can only happen in this page: it is generated sample-by-sample by the same engine used for export.
• Offline: keep the page open, or save the whole site folder — there is no server dependency in the studio itself beyond fetching the page once.

11 · Questions and troubleshooting

I pressed Play and nothing happened

Browsers refuse to start audio until you interact with the page — press Play again (the first press creates the audio context). If the level meters move but you hear nothing, check the master volume and the system output device.

A session sounds like it does nothing

Many sessions are deliberately quiet beds with a slow pulse — listen on headphones before deciding. Also check the mode pill: fallback means the page fell back from the audio worklet to the main-thread engine, which works but can stutter on a busy machine.

My edit made the level fade away

You edited a “hold” into a fade: an entry whose following entry is lower ramps down over its own duration. Repeat the value on the next entry to hold it (see the wrap rule above).

I loaded a preset and the beat vanished

You are on speakers with a binaural pair. Use the Speakers pack, an isochronic quick pick, or headphones.

Importing a pack did nothing

Packs must be .zip archives containing .gnaural files; the reader rejects damaged or oversized archives with a message rather than half-importing them. Anything it does import is remembered in your browser and can be removed with Remove imported.

The WAV export is huge

44.1 kHz stereo float-to-16-bit is about 10 MB a minute. Export a shorter length for a taste test; the session itself is a few kilobytes of XML.

Does anything leave my machine?

No. There is no server-side processing, no account, no analytics, no fonts or scripts from anywhere else. The share links encode the session in the URL fragment, which browsers never send to a server.

12 · Safety and honest limits

See also: the evidence page and reference list · credits, licence and provenance · pack downloads.