Loading page
A limited public demo is planned — here's what to expect.
AudioLens has a public hosted demo. Every feature — all nine analysis panels, both models, every bundled dataset — is available there, no install required.
AudioLens has several built-in sample datasets, fetched from their official sources by scripts (nothing is bundled in the repo), and three of them — RAVDESS, L2-ARCTIC, and the Speech Accent Archive (SAA) — are licensed non-commercial only (see Licenses for the exact terms). AudioLens is a non-commercial, open-source project, and the hosted demo is a non-commercial public demo with no paid tier — so using these datasets there is compatible with their license terms, provided the demo shows each dataset's license and attribution and keeps share-alike terms for derived material. The demo doesn't offer the datasets as downloads; users who want the data get it from the official sources with the scripts on CLI & scripts. Attribution/citation requirements still apply and are listed on the Licenses page. This only becomes a real question for someone who forks AudioLens into a commercial product; that fork would need to drop these three datasets or obtain a separate license for them.
Hosted-demo sessions and anything uploaded to them expire after 24 hours.
Files fetched into Backend/data on a self-hosted install do not.
No rate limiting exists in AudioLens's code today. Usage caps (requests per session, clip length, concurrent jobs, etc.), if any are ever introduced, would be a policy decision made separately, not a feature that's already built. Don't rely on this page — or any other page in these docs — as evidence that quotas exist; as of this writing, they don't.
Prefer to run AudioLens on your own machine? Self-hosting gets the full stack running locally with Docker Compose in about ten minutes — see also the Quickstart.