Use Cases

Each page here carries one complete job from the material you have to the result you want. Most are CLI sessions you can paste into a terminal; the note and voice pages add a Python step, and the app pages are JavaScript built on the WASM package.

Every command on these pages was run against the current engine, and the numbers shown are the numbers it printed. Most of the test material is synthetic and deliberately simple, so expect the pages to point out where it misleads a detector.

Pick the job

New to audio? Start with Build Practice Tracks. It needs one song file, every step is a single command, and you can hear each result straight away. The table runs from easiest to hardest, and each page builds on vocabulary from the ones above it.

Level	You have	You want	Runtime	Page
1	A song you are learning	Slowed, transposed, and separated practice tracks	CLI	Build Practice Tracks
2	A WAV or MP3 of a piano piece	The notes as a MIDI file, checked against the score	CLI + Python	Turn a Piano Recording into MIDI; try it first in Piano Transcription
2	A pile of rough recordings	Clean takes, with a record of what was wrong	CLI	Clean Up a Batch of Recordings
2	Host, guest, and a music bed	A ducked, loudness-compliant episode	CLI	Produce a Spoken-Word Show
3	A folder of stems	A finished, platform-ready master	CLI	Mix and Master a Song in the CLI
3	A mix and a commercial track you admire	Your mix sitting in the same tonal and loudness territory	CLI	Match a Reference Track
3	A master about to be delivered	A pass/fail gate that runs in CI	CLI	Gate a Delivery in CI
3	A vocal take with a few flat or sharp notes	Those notes on pitch, the vibrato kept as sung	CLI + Python	Tune a Vocal Take
4	Local stems and a web app interface	Interactive mixing and a WAV download	JavaScript (WASM)	Build a Browser Mixer
4	An app that hosts inserts or a mastering chain	Controls generated from the engine's own parameter descriptors	JavaScript (WASM)	Build Controls from Parameter Descriptors

Level 1 is one command per step. Level 2 adds a batch script, a preset you edit, or a short Python script that scores a result. Level 3 chains several stages and asks you to judge loudness and tone. Level 4 is application code. If a term on a page is new, Audio Basics and Mixing Basics cover the vocabulary.

Every page opens with a What you need box: the files, their format, and the install line. Where a page's material is synthetic, the box carries a short script that writes equivalent test files, so you can run every step before you point it at your own audio.

What the CLI pages assume

The CLI walkthroughs use the Python CLI (pip install libsonare). Two CLI properties shape them and are covered in full in the CLI Reference:

• Not every command is on both CLIs. The scene mixer (mix), the preset-driven master and mastering-chain, and declip are Python-CLI only; the low-level metering commands (meter, clipping, phase, stereo, dynamic-range) are native-CLI only. See Which CLI Are You Using?.
• Most commands are mono by nature. Mixing and mastering keep a stereo pair end to end; most measurement commands downmix and say so on stderr. This is why these pages read loudness out of a mastering report rather than off a separate lufs call. See Stereo and Mono Handling.

Every command takes --json, and every failure maps to a documented exit code, so each walkthrough can run unattended in a script.

The CLI + Python pages run the same Python package (Python 3.11 or later). The Node tabs on the CLI pages and the JavaScript pages use the WebAssembly npm package, npm install @libraz/libsonare (Node.js 22.12 or later), as in Getting Started. Each Node and Python tab shows how its audio is loaded, so every variable in it is defined.

Reaching for an API instead

A use case that ends in a UI, a plugin, or a realtime path belongs in an API, not a shell script. Start from JavaScript API, Python API, or C++ API; the concepts are the same and the pages here cross-link to the matching guide at each step.