Decibri has four interface surfaces. Python, Node.js, and the CLI run on the same Rust audio backend. The browser build is a separate JavaScript implementation in the same npm package, carrying capture with channel selection, resampling, energy-based voice activity detection, and playback.
Native Python package via PyO3 / abi3, pre-built wheels for Linux, macOS Apple Silicon, and Windows. Exposes Microphone and Speaker plus matching AsyncMicrophone and AsyncSpeaker for asyncio.
Rust native addon via napi-rs, cpal audio backend, pre-built binaries for Windows / macOS / Linux. Exposes Microphone (Readable stream) and Speaker (Writable stream).
Same npm package; conditional exports serve an AudioWorklet implementation when bundled with webpack, vite, or similar. Requires HTTPS and a user gesture for microphone permission.
Standalone statically-linked binary: decibri-cli. Capture WAV, play WAV, enumerate devices, report version. Installable via npm, cargo, or direct download.