← Back to demo

Privacy policy

Last updated: May 1, 2026

The Slerp audio demo at slerp.audio runs entirely in your browser. Audio files you load are not uploaded to our servers. Settings, presets, and similar data may be stored in your browser’s localStorage on your device only.

On-device audio analysis (Studio reactivity). When you tap the STUDIO button on the deck, the app performs a one-time analysis of the track you have loaded — stem separation (drums / bass / vocals / other), beat detection, and energy / vocal envelopes — so the visuals can react to those signals. Every step of that analysis runs locally in your browser, on your device’s graphics processor, using the browser’s WebGPU API. The audio you load and the analysis results derived from it are not sent to our servers, to the content delivery network, or to any third party. There is no server-side inference path; if your device cannot run on-device inference, the Studio feature gates itself off and the deck falls back to live frequency-based reactivity (no analysis performed at all).

To make on-device analysis possible, your browser does download a fixed analysis model bundle (roughly 50 MB) from the same content delivery network that serves the rest of the site, the first time you tap STUDIO. Before running the model the app verifies the downloaded bytes against a known content fingerprint, so a tampered or substituted file is rejected. The model bundle is cached by your browser like any other static asset; it never sees your audio. Analysis results for a given track are cached in your browser’s on-device storage (IndexedDB) so dropping the same file twice is instant — clearing site data in your browser deletes them. None of this cache is mirrored to any server.

If you contact us by email, your messages are handled by your email provider and ours under their respective terms. We do not use your communications to train third-party models.

Site delivery and operations. Pages are served over HTTPS through a third-party content delivery network (CDN). We have not enabled per-request access logging for this delivery — no logs are written to storage we control — so we do not create, store, or review access logs (including client IP addresses) for routine page loads. Your browser still connects to the CDN to load the site; we simply do not keep line-item access logs of those connections in our own systems.

Analytics — Plausible. We use Plausible Analytics to count daily visitors and pageviews. Plausible is cookieless and does not store any personal identifier on your device or on our servers. To deduplicate visits within a single day, Plausible computes a one-way hash of your IP address, the website domain, and a daily-rotating salt that is discarded every 24 hours; the hash cannot be reversed back to your IP and cannot be linked across days. Plausible processes this data on EU servers and does not share it with advertisers. The tracker is bundled directly into our application (no third-party script URL is loaded), and pageview pings POST to /api/event on our own domain, which we proxy to Plausible — so no third-party hostnames appear when you visit slerp.audio. In addition to pageviews, we send a small number of aggregated performance custom events: Core Web Vitals (LCP, INP, CLS, and related signals) are reported in coarse buckets rather than raw millisecond values, and we send one summary per session of animation frame rate, also bucketed into ranges (not a continuous trace), together with low-cardinality technical context such as which visualizer path ran. These events help us spot regressions without profiling individuals.

Lawful basis (EU/UK). Where the GDPR or UK GDPR applies, we rely on legitimate interests (Art. 6(1)(f)) for the limited processing described above: serving the site and understanding aggregate traffic and performance patterns with privacy-preserving analytics. No special-category data is processed and no automated decision-making is performed. To exercise a data-subject access, deletion, or objection request — or to ask any privacy question — email hello@slerp.audio.

If we add additional analytics or change how long we retain or process this data, we will update this page.

Essays: Build notes — prompt cache & usage.

Questions: hello@slerp.audio