Browser-local audio workspace

Batch audio conversion that never leaves your device

Drop a folder of audio in, choose an output, and convert the lot on your own machine. Sonic Batch decodes, processes and encodes inside the browser tab — there is no upload endpoint in this application, so your recordings physically cannot go anywhere.

40
purpose-built tools
15
formats read
9
formats written
0
bytes uploaded

Why a local tool is the right tool here

Audio files are large and often private. An interview, a client mix, a voice memo of a medical appointment, a recording of a meeting — uploading those to a stranger’s server to change a container is a poor trade, and it is slow: a 200 MB file has to go up and come back down before anything happens.

Doing the work locally inverts both problems. Nothing is transmitted, so nothing can be retained, logged or leaked; and the conversion starts the moment you press the button, because the bytes are already where the encoder is.

The engine is FFmpeg compiled to WebAssembly, self-hosted at a pinned version rather than pulled from a CDN, loaded on demand and cached. Lightweight work — WAV handling, loudness measurement, spectrum analysis — uses the browser’s own decoders and our own code instead, skipping the engine download entirely.

What is deliberately absent: no accounts, no AI features, no “enhance” button that guesses at data which is not there, and no unlimited claims. The limits are stated and enforced.

All tools

Batch is the point, not a bonus

  • Reorderable queue with per-file and apply-to-all output settings.
  • A scheduler that reads your CPU count and available memory and picks a safe number of parallel jobs — one at a time on iOS and Safari, where memory ceilings are lower.
  • Real per-file and aggregate progress, driven by decoded position rather than a timer.
  • Cancel, retry only the failures, and pause before the next file starts.
  • Download each result as it lands, or take a deterministic ZIP with a full manifest.
  • Sequential download mode so a long run never accumulates results in memory.

Every file reports what actually happened to it: which processing path ran, what was resampled, whether a downmix was applied, what loudness was measured, and what the true-peak ceiling did to the requested gain. The ZIP manifest records all of it, so a batch is auditable after the fact.

Read how the engine and limits work or check the format and browser matrix.