Ember Drones.

Ember Drones generates calm, non-distracting sound for focused work. Nothing you hear is a recording. The drone, the noise, the water, the embers, the chimes: all of it is synthesized live in your browser the moment you press play, by a generative engine designed so that no two minutes are ever quite the same.

Why it helps: the honest version

A lot of focus-sound products lean on claims the science doesn't support. Ember Drones is built the other way around: it only ships mechanisms with real evidence behind them, framed at their real strength.

Masking: the strong claim

A steady sound covering the irregular noise around you, speech especially, measurably reduces how distracting that noise is. This is ordinary acoustics, the same principle commercial office sound-masking systems are engineered on, and it's the backbone of this app. Ember's noise layers are shaped to the speech band and held perfectly steady, because steadiness is what masking depends on.

Low salience: the design claim

What breaks focus isn't sound; it's events. Lyrics, melodies, beats, and sudden onsets all pull attention, and research consistently finds intelligible speech and lyrical music to be the worst offenders. So everything here is designed to give your attention nothing to grab: no melody, no rhythm, no words, and movement slow enough to stay beneath notice.

Noise and attention: honestly framed

Research on broadband noise finds small benefits for attention that vary a great deal from person to person. The benefits are most consistent for listeners who struggle with attention, and absent (or slightly negative) for others. That's exactly why the noise layers here are optional and adjustable rather than promised as a cure for anything.

How the engine works

At the center is a drone of about eight voices over a single root, tuned in pure whole-number ratios (just intonation) so the intervals lock rather than beat against each other. Changing the mood moves one chord tone and nothing else: harmony without melody.

A motion setting decides whether the harmony moves at all. Still, the default, holds one chord forever. Drift lets the chord color wander every few minutes. Cycle sways between two or three chords over a bass that never moves, each change a long crossfade between perfectly tuned voicings, at a pace you choose, in the tradition of slow tape-loop ambient. Weather lets the harmony wander by itself: it drifts away from home, is always pulled back, and rests on one chord for minutes at a time, so no two hours are ever alike. Even at its most mobile, nothing here ever plays a melody.

Each voice swells and recedes on its own slow cycle, and the cycle lengths deliberately share no common factor, so their combination takes a very long time to ever line up the same way twice. On top of that, slow random drifts replace fixed cycles, voices occasionally fade to silence and return, and the whole texture ages gently across a long session. The same identity, never the same twice.

The textures (surf, ember crackle, wind chimes, soft struck blooms) are built from the same raw materials: oscillators, filtered noise, and time. No samples anywhere.

None of these techniques are our invention. They come from six decades of the generative ambient tradition: tape-loop phasing, systems that play themselves, music designed to color a room rather than command it. Those ideas meet the acoustics of modern speech masking here. The engine runs entirely on your device, and its code is deliberately open to view.

Quiet by design

Ember Drones is meant to sit just above the edge of hearing: present enough to soften the room, quiet enough that you forget it's there. If you notice yourself listening to it, turn it down.

No edges, nothing to miss

The sound is designed to have no edge. There is no thread to lose: you can talk over it, take a call, step out, and come back without having missed anything, because there is nothing to miss. A session rises gradually out of silence rather than starting, and a timed session subsides over its final minute rather than stopping. The aim is for the sound to blur into the room's own noise instead of fighting it, until what you hear stops dividing into the sound and the noise at all.

Common questions

Does background sound actually help you focus?

Partly, and it depends which claim you mean. Steady sound reliably reduces how distracting other noise is, which is well established acoustics rather than neuroscience. The separate claim that broadband noise directly improves attention is weaker: the effect is small, it varies a great deal between people, and it is most consistent among listeners who already struggle with attention. Ember Drones is built on the first claim and deliberately cautious about the second.

What is sound masking?

Sound masking is covering irregular background noise, especially speech, with a steady and spectrally shaped sound so the intruding noise becomes less intelligible and less distracting. It is the same principle behind the systems installed in open plan offices. Masking works by reducing the signal-to-noise ratio of the speech reaching your ears, which is why a quiet, steady, well shaped sound outperforms a loud one.

Is brown noise or pink noise better for focus?

Neither is universally better; they differ in tone rather than in effectiveness. Brown noise is darker and heavier, weighted toward low frequencies, and many people find it calmer. Pink noise is more balanced and carries more energy in the speech band, roughly 800 Hz to 5 kHz, which is the range that matters most for covering conversation. Ember Drones offers both, plus a tone control that tilts between them, because the best setting is the one you stop noticing.

Do binaural beats help you focus?

The evidence that binaural beats improve focus is weak and mixed, so Ember Drones does not use them. The same applies to Solfeggio frequencies and to 432 Hz tuning. All three are common in focus and wellness audio, and none of them are supported well enough to build on.

Is the sound recorded or generated?

It is generated. Every sound is synthesized live in your browser using the Web Audio API. There are no audio files, no samples, and nothing streamed from a server. The engine builds each drone voice, noise layer, and texture out of oscillators and filtered noise while you listen, which is why it never loops and no two sessions sound alike.

Is the sound AI-generated?

No. There is no machine-learning model anywhere in Ember Drones. The engine is a hand-written generative system: rules and ranges chosen by ear, one decision at a time. Randomness decides when things happen, but a person decided everything that can happen. Nothing is generated on a server, and the code runs openly in your browser, where anyone can read it.

Is Ember Drones free, and do I need an account?

It is free, there is no account or sign up, and nothing you do is tracked or sent anywhere. Your settings and saved mixes stay in your own browser. You can install it as an app, and once installed it runs with no connection at all.

How loud should focus sound be?

Quiet. It should sit just above the edge of hearing: enough to soften the room and cover nearby speech, little enough that you forget it is playing. If you catch yourself listening to it, turn it down. Loud background audio is one of the few things research consistently finds unhelpful for demanding work.

How is this different from a focus playlist?

Music creates events. Melodies, beats, and especially lyrics all pull attention, and intelligible speech is the most disruptive background sound for reading and writing. Ember Drones is deliberately non-melodic, has no beat and no words, and changes slowly enough that nothing arrives as an event. It is built to be ignorable in a way that music is not.