What you can build
Everything already available to build on: 60 named effects, a library of instruments and MIDI tools, interface controls, sampler banks, generative audio, musical data, embedding, plug-in builds and selling.
You rarely start from nothing on Maru. The effect, instrument or control you have in mind may already exist as working, editable source. Most new tools are new combinations of tested pieces, given your own sound, interface and name. Ask your agent to recompose them, or open the source and do it yourself.
This page is the map. Each capability works whether you build in Studio, with your AI assistant or in code, and each links to a guide that explains how.
Already available
| Capability | What you get | Why it helps |
|---|---|---|
| 60 named effects | Drive, modulation, filter, dynamics, delay and reverb algorithms you can chain in MultiFX | The effect you want probably exists. Recompose instead of writing DSP. Effects and instruments |
| A library of modules | Synthesizers, physical-modeling instruments, samplers, effects, MIDI tools and sequencers in the module library | Start from a working product and change it |
| Sound building blocks | Filters, delays, reverbs, distortion, dynamics, modulation, spectral tools, physical models, oscillators and voices | New products are mostly new arrangements of tested parts. DSP library |
| Interface controls | Knobs, sliders, keyboards, pads, meters, scopes, visualizers and themes bound to parameters | A distinctive interface without writing control logic. Module UI |
| Sampler instruments | Melodic instruments and drum kits from your recordings, mapped in Studio | A playable instrument with no DSP or interface code. Sampler banks |
| Generative audio | Sounds, loops, kits and variations from a prompt; stems and MIDI from recordings | New material for your instrument, with the models run for you. Generate audio |
| Musical data | The Symbolic theory library, MIDI files and transforms | Chord tools, arpeggiators and analysis without audio code. Symbolic library |
| Embedding and racks | One module inside a React or web app, or several connected in a rack | Put your tool inside your own product. Host a module in React |
| Plug-in builds | Browser modules and native plug-in builds from one source | Write once and build the formats your module supports. Native plugins |
| Selling | Checkout, license activation, trials, refunds and testers | Charge for your work without building a store. Sell and share |
| Agent tools | A skill, a CLI and two MCP servers over one set of operations | Your assistant changes a real project, not a guess. Connect your AI assistant |
What's newly possible
Some of the most interesting starting points used to need a research lab. Each of these is a module you can play now and change:
| Possibility | Start from |
|---|---|
| Generate whole kits from one idea, morph between kits, split tracks into stems | Moodbox, Tetra Box, StemSep |
| Play a learned sound space live, restyle live audio, jam with an accompanist | Terrain, LiveTransfer, Accompanist |
| Turn singing or playing into notes and expressive pitch | Pitchy, Vocoder, Groove MIDI |
| Use a webcam as a controller | Hand Chord, HandMod, CamMod |
| Write music as code with real music theory | Code Sequencer, Pretty Chord, ImproSax |
| Simulate rooms and play neural amp captures | Room Acoustics, Capture Player, Amp Rig |
| Build physical-model and expressive instruments | Handpan, Mouth Synth, PolySlide |
From no code to custom DSP
Go as far as you want. Every step is a complete result on its own.
| Step | Result | In Studio | With your assistant | In code |
|---|---|---|---|---|
| 1. Sampler instruments and kits | A playable bank with presets | Sampler bank project with keyboard and pads | Generate sounds, then start a bank project | Optional |
| 2. Recompose | A new effect or instrument from library pieces | Ask the Studio agent | Ask your coding agent to edit a starter | Combine modules and building blocks |
| 3. Your own interface | An existing module with a new look and feel | Ask the Studio agent | Ask your coding agent | Face source with the UI kit |
| 4. Generate sounds | Takes, kits and variations for a project | Generate into a bank's sources | Ask the hosted connector | The generation client |
| 5. A Generate button | An instrument face that asks for new sounds | Not in Studio yet | Ask your coding agent | The generation face capability |
| 6. Custom DSP | A new processing algorithm | Ask the Studio agent | Ask your coding agent | A restricted-TypeScript Processor |
| 7. Embed and connect | A module in your app, site or rack | Not in Studio yet | Ask your coding agent | React or imperative host, rack host |
| 8. Musical data | Theory, MIDI analysis and transforms | Symbolic Cycle | Ask your coding agent | Symbolic library |
Steps 1 to 4 need no code at all. Before step 6, check steps 2 and 3: the sound you want may already be one combination away.
Where things stand
Maru is in beta. Studio needs a beta account. The local workbench and plug-in builds are available to developers in the selected-developer beta; see the Quickstart. Hosted generation is in preview. Each guide states what its path needs, next to the step that needs it.
To understand why it is built this way, read How it works. If you run a DAW, an online service or an existing plug-in line, read Integrate Maru into your product.