This page separates audio processing from ordinary website delivery so you can understand what stays in the tab and what a web host may still receive.
The live tools use browser APIs to read and decode media you select. Decoded samples are processed in memory, and results are encoded into temporary local URLs for preview and download.
The current SoundTools application has no account system, media upload endpoint, or storage bucket for source and processed audio. Closing or refreshing the tab releases the in-memory working state and temporary preview URLs.
Your browser must request the HTML, CSS, JavaScript, icons, and other site assets required to display SoundTools. The hosting and content-delivery providers may process standard request information such as IP address, browser details, requested path, time, and security signals.
The current design also loads font files through Google Fonts. Those font requests are separate from audio processing and do not include the audio file selected in the workbench.
You do not create an account or provide a name to use the audio tools. The audio workbench does not require an account cookie. Website analytics and advertising are separate from audio processing.
Operational hosting systems may use short-lived security or delivery mechanisms outside the audio workbench. Google Analytics measures site visits and may use analytics cookies. These requests include technical visit information and do not contain your audio. See Google’s privacy policy.
Adsterra supplies one 300 × 250 banner below the audio workbench. We do not show banners on the vocal remover or stem splitter, which require cross-origin isolation. Ads load when the placement comes into view.
Adsterra and its advertising partners may receive an IP address, browser and device information and the referring site, and may use cookies or similar identifiers to deliver and measure ads. The banner runs in a separate sandboxed frame without access to the workbench or its selected audio. We do not send audio files, decoded samples or generated results to the ad provider.
Use “Hide ads” below the banner to stop it and save that choice in this browser. “Show ads” enables it again. Existing opt-out choices are preserved, and ads are disabled when your browser sends Global Privacy Control or Do Not Track. Advertising choices do not change Google Analytics. See Adsterra’s privacy policy.
Noise removal runs through RNNoise WebAssembly. Transcription, music separation, and authorized voice cloning download Whisper, Demucs, or Chatterbox model files, then perform inference on the device through WebGPU. These asset requests do not contain your selected media.
Chatterbox voice-cloning references remain in the current worker session. Generated Chatterbox audio retains the model's provenance watermark; SoundTools does not remove or disable it.
WebGPU tools state their compatibility and model size before processing begins. SoundTools does not fall back to a remote transcription or separation service when local inference is unavailable.