0wave Studio

A web synthesizer with an essential DAW environment. Build sounds in the Studio, compose with them in the Playground.

It is meant to be usable without a background in music: the first run offers three real starting points (design a sound, build a loop, or look around), every term of art carries a one-sentence explanation you can switch off, and an empty Playground will build you a four-bar loop to press play on. See docs/GUIDANCE.md.

Design Read

Reading this as: a desktop-first creative audio instrument for musicians and curious sound designers, divided into two focused environments: Studio for building sounds and Playground for composing with them. The visual language is minimal, industrial and tactile, inspired by compact music hardware without copying any specific product.

Taste Skill dials: DESIGN_VARIANCE: 5 · MOTION_INTENSITY: 3 · VISUAL_DENSITY: 7.

Design skill: design-taste-frontend from Leonxlnx/taste-skill, content hash 6d838b246d0e35d0b53f4f23f98ba7a1dd561937e64f7d0c7553b0928e376c3e (see skills-lock.json). The skill applies fully to the landing page (post-MVP); inside Studio and Playground we apply brief inference, anti-default discipline, hierarchy, color/shape consistency, accessibility, complete interaction states, and the pre-flight check. Product requirements win on conflict.

Quickstart

pnpm install
pnpm dev        # http://localhost:3000 → /studio
pnpm typecheck
pnpm lint
pnpm test       # unit + integration (Vitest)
pnpm build      # production build

Supabase is optional. Without credentials the app runs in PUBLIC_LOCAL mode (IndexedDB only). See docs/DEPLOYMENT.md for the three access modes and cloud setup.

Layout

src/
  app/                  Next.js App Router routes (/studio, /playground)
  components/
    shell/              Top bar, Start Audio gate, app shell
    guide/              Onboarding, guided tour, plain-language hints
    controls/           Shared tactile controls (knob, fader, toggle…)
    studio/             Synth, Record, Import, Sample Editor, Library
    playground/         Transport, tracks, timeline, editors, mixer
  lib/
    audio/              Framework-independent audio engine (no React)
    schema/             Versioned project schema + zod validation + migrations
    state/              Zustand stores (serializable project, UI-only state)
    persistence/        IndexedDB repo, autosave, import/export, access modes
    presets/            Curated system presets and the starter loop
    music/              Note/frequency math, quantization, swing
  i18n/                 Centralized UI strings (en, it)
supabase/migrations/    SQL migrations + RLS policies
docs/                   Architecture, decisions, engine, deployment…

Hard rules (enforced by review)

• No fake controls: every UI control drives real engine or store state.
• Every user-facing string lives in src/i18n, in both dictionaries.
• Motion is defined in globals.css and degrades under prefers-reduced-motion.
• No AudioNode in serializable state; no React re-render at audio rate.
• One AudioContext, created/resumed only from a user gesture.
• Pattern Editor and Piano Roll edit the same NoteEvent model.
• Product name is always 0wave Studio (zero, capital S).

Documentation

Doc	Contents
docs/ARCHITECTURE.md	Layer boundaries, module map, SSR safety
docs/DECISIONS.md	ADR-001 to ADR-010
docs/AUDIO-ENGINE.md	Clock, scheduling, signal path, voice management
docs/STUDIO.md	Modes, synth editor, recording, import, sample editing
docs/PLAYGROUND.md	Transport, tracks, timeline, both note editors
docs/GUIDANCE.md	Onboarding, glossary, hints, tour, motion system
docs/PIANO-ROLL.md	Interaction contract for the piano roll
docs/PROJECT-SCHEMA.md	Versioned schema and validation boundaries
docs/DEPLOYMENT.md	Vercel, Supabase, access modes, production checklist
docs/BACKLOG-AI-CONNECTOR.md	Post-MVP text-to-patch specification
docs/HANDOFF.md	State at handoff, what is unverified, next steps, landmines
docs/TASTE-PREFLIGHT.md	Taste Skill checklist, deviations, failures found

Browser support

Browser	Synth + sequencer	Microphone recording	Audio import
Chrome / Edge (current)	Yes	Yes (MediaRecorder, WebM/Opus)	WAV, MP3, OGG, FLAC
Firefox (current)	Yes	Yes (WebM/Opus)	WAV, MP3, OGG, FLAC
Safari 17+	Yes	Yes (MP4/AAC; pause may be unavailable)	WAV, MP3, AAC/MP4

Decoding is delegated to the browser, so the accepted formats are the browser's, not a list we guarantee. Unsupported files report a clear error. Recording requires HTTPS (or localhost). Minimum usable width is 1024px.

Known limitations

These are real gaps, not roadmap phrasing:

• Desktop-first. Below 1024px the layout scrolls horizontally; there is no mobile-specific interface.
• No time-stretch and no pitch-shift with independent duration. Audio clip length does not warp the material.
• Multi-select in the piano roll is shift-click based; there is no rubber-band selection yet.
• Per-track effect chains exist in the engine and schema, but the Playground UI currently exposes only volume, pan, mute and solo; effects are edited on the sound in the Studio.
• Audio clips are rescheduled for a bounded number of transport loop passes (16), which covers normal sessions but is not infinite.
• Cloud sync (Supabase) is implemented behind the repository adapter but has not been executed against a live project; see the report in docs/DEPLOYMENT.md for exactly what remains unverified.
• Undo granularity is snapshot-based: continuous knob drags coalesce into the surrounding edit rather than producing one entry per pixel.

See docs/DECISIONS.md for architectural decisions and docs/ for per-area documentation.