Build and share

Use audio models creatively, build your own instruments and tools, and bring them to the people who'll play them.

Audio machine learning and creative sensing

Understand what audio models can contribute to a musical idea, and hear the limits before choosing a technique.

  1. What AI can do in musicListening, editing, generating and playing along: the jobs a model can take on
  2. What an audio model learnsTraining, inference and why a model's output is an estimate
  3. How models see audioWaveforms, spectrograms, latents and tokens
  4. Tagging, search and similarityHow models label sounds and find ones that sound alike
  5. Stem separationHow models pull drums, bass and vocals out of a mix, and where they struggle
  6. Repair and restorationRemoving noise, room and clicks, and what goes with them
  7. Audio generationHow models make new sounds from noise, tokens or prompts
  8. Generating notesMelodies, drum patterns and chords from a model, and why editable output matters
  9. Neural instruments and timbre transferInstruments whose sound comes from a trained model
  10. Voices, cloning and consentSinging synthesis, voice conversion and whose voice it is
  11. Playing with a model liveLatency, listening and taking turns with a generative bandmate
  12. Training on your own soundsData, fine-tuning and personal models of an instrument or amp
  13. Camera and gesture controlTurning movement into musical control
  14. Adding a model to your appCapability checks, downloads, workers, progress and cancel

Build your own tools

Translate a musical intention into a focused tool whose sound and behavior you can test.

  1. Signals and systemsLinearity, memory and why EQ and distortion behave so differently
  2. Anatomy of a Maru moduleDeclaration, ports, parameters, processor and face
  3. Build an effectFrom a musical promise to a working processor
  4. Build an instrumentVoices, notes, envelopes and stuck-note safety
  5. Build a MIDI toolEvent timing, routing and deterministic randomness
  6. Realtime and numerical safetyDenormals, allocation, smoothing and block boundaries
  7. Testing sound and behaviourSilence, clipping, DC, extreme settings and listening
  8. Presets, state and automationMaking saved sessions come back exactly as they were
  9. Shipping to browser and pluginsTargets, formats and what each one has been tested on
  10. Building with an AI agentBriefs, reviewing changes and catching unsupported claims

Instrument design and publishing

Make an instrument understandable and useful to someone else, from its controls to the way it is shared.

  1. What makes an instrument inspiringImmediacy, response, identity and depth
  2. Controls, macros and rangesMaking every knob worth turning
  3. Presets as product designA default that sells it and a set with distinct jobs
  4. Interface and first useLabels, hierarchy, accessibility and the first minute
  5. Finding an audienceWho it's for and how to test whether they care
  6. Identity and storyTurning an artist's sound into something playable
  7. Demos, pricing and launchHonest demos, costs, channels and a first launch experiment
  8. Rights and maintenanceLicences, credits, updates and support