Learning Path

This page routes you through the libsonare docs by what you want to build. You do not need to understand DSP, MIR, WebAssembly, or mastering terminology before starting.

Acronyms you will see early

DSP (digital signal processing) means measuring or changing audio as numbers. MIR (Music Information Retrieval) means reading musical information such as BPM, key, and chords from audio. WASM / WebAssembly is the browser runtime format. API means the functions and classes your app calls. CLI means a terminal command. At this stage, you only need enough vocabulary to choose the next page.

The sidebar groups the guides into ten subject areas: Analysis, Instruments & MIDI, Mixing, Mastering, Editing, Repair, Arrangement & Projects, Realtime, Room Acoustics, and Playback. The headings below use the same names, so the area you find a goal under is also the sidebar group its guides sit in. The runtime references and the evidence pages follow the areas.

If you have not run anything yet, start with Getting Started: one install line and one short program per runtime.

Reading the docs front to back

1. Introduction for vocabulary and the overall audio-analysis pipeline.
2. Audio Basics for samples, sample rate, decibels (dBFS, RMS, LUFS), frequency and harmonics, and stereo, if any of those are new to you.
3. MIR Overview for how spectrograms, chroma, onsets, and the analysis results build on one another.
4. This page, to find the subject area your goal belongs to.
5. Getting Started, Installation, and Examples to run one small program.
6. Feature Map to find the right API family.
7. One walkthrough from Use Cases when what you want is a finished deliverable rather than an integration.
8. One guide from the subject area that matches your goal, below. Mixing, editing, and mastering each have a basics page (Mixing Basics, Editing Basics, What Is Mastering?) to read first.
9. One runtime reference from API By Runtime.
10. The evidence pages only when you need implementation details, algorithm references, validation scope, or performance context.

Start With Your Goal

Find your goal under the area it belongs to. Read first is the page written for that goal; Then read is where the detail lives once the first page has made sense. Rows that start from a use case are complete walkthroughs with commands you can run.

Analysis

Every result in this area (tempo, key, chords, sections) is read off the same first step. The audio is cut into short overlapping windows and each window is turned into a spectrum, so the whole recording becomes a picture of which frequencies are present at each moment. If you have never seen that picture, this is it:

STFT · SPECTRALIDLE

STFT — seeing time and frequency at once

FREQ · REL dBTIME →

Scroll to load

A tone gliding from 400 Hz to 1 kHz, with four clicks. Each column is one short-time spectrum; brighter means more energy. Change nFft: at 256 the clicks are thin lines and the tone is a band about 170 Hz wide; at 4096 the tone is a line about 20 Hz wide and each click smears across about 280 ms.

nFft (window)

I want to build...	Read first	Then read
A browser app that shows BPM, key, chords, or sections	Getting Started	Music Analysis, WebAssembly Guide
A Python script or notebook for audio analysis	Getting Started	Python API, Music Analysis
A terminal workflow for quick checks or batch analysis	Getting Started	CLI Reference, CLI Commands
Song structure, self-similarity, or note-segmentation analysis	Music Analysis	JavaScript API, Python API
Bars and time signatures scored over a beat series you already have	Music Analysis	Meter and Grouping
A migration from librosa	librosa Compatibility	Music Analysis
A librosa figure (frame and sample conversion, decibels, peak picking) checked against the engine from a shell	CLI Utilities	librosa Compatibility
A source-separation view that keeps the harmonic and percussive paths available for editing	Source Separation	Music Analysis, Editing DSP

Instruments & MIDI

Nothing in this area plays back a recording: the engine ships no sample data, and every instrument is computed while the note sounds. Sound Sources explains that decision and what follows from it. Read it before any page that assumes you have already chosen a synthesis method.

I want to build...	Read first	Then read
A choice between a physical model, FM, subtractive synthesis, and a SoundFont you bring	Sound Sources	Built-in Synthesizer, Physical Models
A synth or instrument app that renders MIDI to audio	Built-in Synthesizer	MIDI Input, Realtime and Streaming
A MIDI clip editor or sequencer with MIDI 1.0 and MIDI 2.0 events	MIDI Editing	MIDI 2.0 & UMP, Project MIDI
The acoustic voices (piano, bowed string, reed, brass, flute) heard side by side or voiced from code	Physical Models	Built-in Synthesizer
A lookup of which engine and voicing a General MIDI program falls back to when no SoundFont covers it	GM Tone Map	GM and GS
SoundFont (SF2) playback through a built-in player	SoundFont 2 Player	GM and GS, MIDI Input
A C++ app that needs the instruments only, without analysis or mastering	Link targets	Built-in Synthesizer, SoundFont 2 Player

Mixing

I want to build...	Read first	Then read
A browser or native mixer	Mixing Engine	Mixing Scene JSON
A browser mixing app, read from a working demo's source	Build a Browser Mixer	Mixing Engine, Mixing Demo Project JSON
A starting mix proposed from the tracks themselves, with a reason for every move	Mixing Assistant	Mixing Scene JSON, Mixing Engine
Reverb, modulation, or delay on a strip or a bus	Effects Inserts	Mixing Scene JSON
Insert controls drawn from the engine's own parameter descriptors instead of a hand-kept table	Build Controls from Parameter Descriptors	Effects Inserts, Realtime Engine
A talk episode with the music ducked under the voices, delivered at a loudness target	Produce a Spoken-Word Show	Mixing Engine, CLI Reference

Mastering

I want to build...	Read first	Then read
A mastering UI or automatic mastering workflow	Mastering Assistant	Mastering Processors
A finished master from a folder of stems, in the terminal	Mix and Master in the CLI	CLI Reference, Mastering Processors
A mix brought into the tonal and loudness range of a reference track	Match a Reference Track	Mastering Assistant
A pass/fail check on a master before delivery, run in CI	Gate a Delivery in CI	CLI Reference
A UI wired to the mastering chain's own controls, or a render report you have to explain	Mastering Implementation	Mastering Processors, Build Controls from Parameter Descriptors

Editing

I want to build...	Read first	Then read
Pitch, time, voice, or source-separation editing	Editing DSP	JavaScript API
Audio transcription, clip replacement, and tempo-aware MIDI placement	Audio to MIDI	MIDI Editing, Project MIDI
A piano recording turned into a MIDI file and checked against the score	Turn a Piano Recording into MIDI	Audio to MIDI
Note-level edits that render back into an audio clip	Note Editing in Audio	Editing DSP
Region-based spectral edits (attenuate, mute, gain, or heal a time-frequency rectangle)	Spectral Editing	Editing DSP
Slowed, transposed, and separated practice tracks of a song	Build Practice Tracks	Editing DSP, Source Separation
A recorded take tuned to the melody in a MIDI file, from the shell	CLI Utilities	Project MIDI

Repair

Measure the defect before choosing a repair stage. Broadband noise, mains hum, clipped peaks, impulsive clicks, crackle, and diffuse reverb need different evidence and different reconstruction limits.

I want to build...	Read first	Then read
A repair workflow that chooses stages from measured defects	Audio Repair Workflow	Repairing Noise and Hum, Repairing Clipping, Clicks, and Crackle, Repairing Diffuse Reverb
A batch of rough recordings cleaned up, with a record of what was wrong in each	Clean Up a Batch of Recordings	Audio Repair Workflow

Arrangement & Projects

I want to build...	Read first	Then read
A DAW, arrangement, or MIDI-sequencing tool	Project Editing	Project MIDI, Bouncing Projects
A sung take corrected note by note (pitch, formant, level) with undo, and the result placed in a project	Vocal Editing	Tune a Vocal Take
A project saved to a file and opened again	Project Save & Load	Project Editing
Microphone capture, loop takes, and comping into a clip	Recording & Takes	Project Editing
A project rendered to audio, with or without the built-in synth	Bouncing Projects	Built-in Synthesizer, SoundFont 2 Player

Realtime

I want to build...	Read first	Then read
A live visualizer, rhythm game helper, or AudioWorklet tool	Realtime and Streaming	WebAssembly Guide
Transport, track lanes, live MIDI, and automation in a playback engine	Realtime Engine	Realtime and Streaming
A realtime microphone voice changer	Realtime Voice Changer	WebAssembly Guide

Room Acoustics

I want to build...	Read first	Then read
Room sound, estimates, or generated room character	Room Acoustics	JavaScript API, Python API
Inverting mel/MFCC features for previews or debugging	Inverse Features	librosa Compatibility

Playback

I want to build...	Read first	Then read
Movie audio rendered to binaural headphones or to calibrated stereo, 5.1, or 7.1 speakers	Playback Renderer	Channel Formats
A check of how each module treats mono, stereo, 5.1, and 7.1 input	Channel Formats	Playback Renderer

How to read the task guides

Each task guide starts with the decision criteria, then shows the smallest useful code path, then lists caveats and related pages. When the same workflow runs in Browser / WASM, Python, and CLI, the page shows all three in a ::: code-group. Inherently WASM / C++ APIs, such as the realtime engine or the AudioWorklet bridge, point to the batch alternative instead of implying that Python and the CLI have the same live callback API.

Reference

Implementation checklist

Before implementing from a task page, check these points.

Check	Where to look	How to decide
Whether your input is a file, encoded bytes, decoded samples, or live blocks	The page's "when to use" / "which API" section	Browser / WASM usually takes Float32Array plus sampleRate; use Audio.fromMemory* or browser codecs for encoded bytes. Python / CLI often read files directly.
Which runtime calls the workflow	The Browser / Python / CLI examples in ::: code-group blocks	When all three are shown, compare API names, argument names, and result shapes before porting code.
Whether the terminology is clear enough for UI copy or implementation comments	Inline ::: tip / ::: info / ::: warning / ::: details blocks and the Glossary	Terms that affect implementation decisions stay close to the code; longer background belongs in callouts or glossary pages.
Whether a runtime gap exists	Binding Parity and the runtime references	If a task page does not show an example for a runtime, the API may be unavailable there or the runtime may be the wrong fit.

The four layers

Layer	What it means	Pages
Concepts	What samples, decibels, BPM, key, STFT, chroma, LUFS, true peak, and related terms mean	Introduction, Audio Basics, Glossary
Tasks	What you want to build, in the ten subject areas above	Feature Map, the area guides from Music Analysis to Playback Renderer
Runtime	Where the code runs: browser, Python, Node, CLI, C++	Getting Started, runtime references
Evidence	How the implementation is structured and validated	DSP Implementation Notes, Algorithm References, Implementation Validation

Most users need only the first three. Open the evidence pages when you need to expose DSP controls in a UI, explain a render report, check whether a processor is appropriate for realtime use, or cite the basis for an algorithm.