Sentinels.

Below is a small Web Audio sketch of the kind of soundscape the device would emit. It is procedurally synthesised — the device should never loop a recording. Set your volume low first, tap Listen, and the chorus will keep going as you read. The corner indicator above pauses from anywhere.

The engine above is procedural, not a recording, so the honest question is whether it lands near a real dawn chorus. We rendered sixty seconds of the engine to an offline audio context — at mid-high density, to match the reference's chorus intensity — and ran the same analysis pipeline over it as over a CC0 reference recording (squashy555, Burton-on-Trent, May 2021).

Eight of nine temporal and spectral metrics match within ~15%. The exception is density coefficient of variation: the reference chorus is remarkably steady across 30-second windows, while the engine is ~2.5× more bursty — visible as cluster-and-gap structure in the spectrogram below. A tuning issue, not a structural one.

When the brain registers ambient acoustic safety, the amygdala settles, the parasympathetic nervous system takes over from the sympathetic, and the rumination circuit in the posterior cingulate cortex quiets. Heart-rate variability rises; cortisol drops. The mood lift outlasts the sound.

This is the broad shape of the nature-restoration literature. The narrower, more interesting question — and the one this project is built around — is whether a small device emitting the right kind of signal at the right volume can engage the same response in environments where real birds are absent.

Six documents form the spine of the work: the hardware design, the firmware that drives it, the research it's grounded in, the experimental methodology that would either support or falsify the hypothesis, the bioacoustic model behind the audio engine, and the analysis confirming that engine matches a real chorus. This is a personal-research project, not a product — and that's reflected in what it commits to.

The behavioural and EEG effects of birdsong on stress markers are real and replicated. The specific evolutionary-neuroethological story — the “200-million-year-old predator-detection circuit” — is plausible narrative, not established neuroscience.

Wellness-coded interventions are also notoriously susceptible to placebo, novelty, and reporting-bias effects. The methodology takes this seriously.