Anyone can convert a file with FFmpeg; remembering the right sample rate, bitrate and loudness target for console audio versus a phone is the part that takes a search every time. Naming the target instead is what makes a folder of raw takes into a shippable set in one call — and the queue status tool exists because that call takes a while.
An FFmpeg front end. Sample rate, bitrate, channels, codec, volume and loudness normalisation are parameters rather than command-line flags, and a preset bundles the right combination for a known target.
- process_audio_file — one file, with volume, format and effect operations described as an object
- batch_process_audio — a whole directory matched by glob, processed through a concurrent queue
- apply_preset — a named target rather than a set of numbers you have to remember
- list_presets, filterable by game, voice, music or effects
- get_queue_status to see how a long batch is progressing
- Presets for mobile, desktop and console game audio, voice and dialogue, music mastering, sound effects, and one tuned for AI voice output
- Volume work in decibels with loudness normalisation to a target LUFS
- Fades, trims and loops alongside the format conversion
- A standalone command-line mode, so the same processing runs outside an assistant
FFmpeg installed and on your PATH — the server is a wrapper and does nothing without it. Node.js 18 or newer; npx runs it without a permanent install. Paths passed to the tools are absolute, and overwriting an existing output file is off unless you ask for it.