Optional page engagement
Use the footer checkbox to allow measurement of page views, approximate active time, scroll depth and counts of actions like starting a tool or exporting a file. Active time counts while the page is visible, up to one minute after your last interaction. Records include a public page path, broad source, screen size and time. They exclude your files, filenames, file durations, microphone recordings, transcripts and visitor identifiers. You can change your choice in the footer.
Files and recordings
Files you choose and microphone recordings are processed on your device. They are not sent to an audio-processing server. Samples are downloaded from this site only when requested.
If you install AudioClover as an app, files you open with it or share to it from another app are handled the same way. A shared file is caught by the site’s service worker on your device and kept in your browser’s storage (IndexedDB) for up to ten minutes, until the tool you pick reads it. It is not sent to us. When you pass a result to another tool, it is parked there the same way. Batch results are zipped or saved into a folder you choose, on your device.
Microphone tools
The mic test, voice recorder and decibel meter use your microphone only after you press start and your browser asks for permission. The sound is analysed inside the page on your device. The mic test’s 5-second clip and voice recordings are kept in the tab’s memory until you download them or close the page; nothing is uploaded. The decibel meter keeps only the numbers on screen, plus your calibration offset in this browser’s local storage. We do not receive device names, levels or audio.
Extra decoder for older formats
Browsers cannot read some older formats, such as WMA, WMV, AVI, FLV, MPG, AIFF and AMR. For those files only, your browser downloads the open-source ffmpeg.wasm decoder (@ffmpeg/core 0.12.10, about 30 MB) from cdn.jsdelivr.net, a public code CDN, the first time it is needed. Like any web server, jsDelivr receives your IP address and browser details with that request. Your file is not part of it: the decoder runs in your browser, and the file itself still never leaves your device. The decoder files are pinned by checksum, so the browser rejects any other version.
Speech model for the filler word remover and audio to text
The filler word remover and audio to text transcribe speech with one open speech-recognition model, Whisper base.en (OpenAI, MIT licence), downloaded once and shared by both. Nothing is downloaded until you press Find filler words or Transcribe. Then your browser downloads the model, 80 MB (98 MB in browsers that use WebGPU), from huggingface.co, which may redirect to its own CDN, and its 28 MB engine (transformers.js and ONNX Runtime Web) from cdn.jsdelivr.net. Like any web server, Hugging Face and jsDelivr receive your IP address and browser details with those requests. Your recording is not part of them: the model runs in your browser, and the audio and its transcript never leave your device. The engine files are pinned by checksum and the model by version. Your browser keeps all of them in its storage so later runs work offline; clearing site data for audioclover.com removes them. The audiogram maker uses the same model and engine for its optional captions, downloaded only when you press Generate captions.
AI model for the vocal remover
The vocal remover separates songs with an open-source AI model (Spleeter by Deezer, MIT license) that runs in your browser. Only after you press Separate, your browser downloads the model files (about 39 MB) from audioclover.com and the ONNX Runtime engine (about 27 MB, the same files the filler-word remover uses) from cdn.jsdelivr.net. jsDelivr receives your IP address and browser details with that request, as with the decoder above; your song is not part of it and never leaves your device. All files are pinned by checksum and kept in your browser’s storage so later separations need no download. Clearing site data for audioclover.com removes them.
Offline copy
After your first visit, your browser keeps a copy of this site’s pages and scripts (a service worker cache) so the tools open without a connection. It holds no files you process, and nothing in it is sent to us. Clearing site data for audioclover.com removes it.
Email updates
If you choose to subscribe, we send your email address, consent time, and the signup page to our signup service. We use these details for the updates you requested. An unsubscribe link will accompany any email we send. You can also unsubscribe immediately using the link provided after signup.
Hosting and measurement
The hosting provider may process request information to deliver the site and protect its service. We never send file or recording names or contents to analytics.
We use Google Analytics, a service provided by Google and loaded through Cloudflare from this site’s own domain, to count visits and tool use, such as finished conversions and edits and the tool steps described below. Google Analytics sets first-party cookies (such as _ga) so it can tell one visit from another. This data is not used for advertising, and the site sets no advertising cookies. See Google’s privacy policy for how Google handles it. You can opt out with Google’s Google Analytics opt-out browser add-on or by blocking cookies for this site.
To learn which websites and assistants send visitors (for example a search engine or ChatGPT), your browser stores one small first-visit note in local storage: the referring site or campaign tag, the first page you opened, and the date. It is not a cookie and holds no personal details. When you export a file or subscribe, that note and the tool name are counted by our own server; we do not store your IP address with it. Clearing site data removes it.
To find tools that fail or confuse people, each tool also counts its steps (file chosen, processing started, finished or failed, download clicked) with only the file type, a size range such as 1–10 MB, and an error category. File names, contents, and error details are never sent.
Future services
Any hosted API or paid service will explain its data handling before you use it.