• Products
  • Studio
  • Engine
  • Research
  • Work with us
Request access

Audio machine learning and creative sensing

  • What an audio model learns
  • How models see audio
  • Stem separation
  • Audio generation
  • Neural instruments and timbre transfer
  • Camera and gesture control
  • Adding a model to your app

Build your own tools

  • Signals and systems
  • Anatomy of a Maru module
  • Build an effect
  • Build an instrument
  • Build a MIDI tool
  • Realtime and numerical safety
  • Testing sound and behaviour
  • Presets, state and automation
  • Shipping to browser and plugins
  • Building with an AI agent

Instrument design and publishing

  • What makes an instrument inspiring
  • Controls, macros and ranges
  • Presets as product design
  • Interface and first use
  • Finding an audience
  • Identity and story
  • Demos, pricing and launch
  • Rights and maintenance
  1. Learn
  3. Build and share
  5. How models see audio

Audio machine learning and creative sensing

How models see audio

Waveforms, spectrograms, latents and tokens

Coming soon

This guide isn't published yet. It will be a hands-on guide with sounds to play and something to make. Browse the rest of Build and share in the meantime.

PreviousWhat an audio model learnsNext Stem separation

Products

  • All products
  • Plugins
  • Sounds
  • Bundles
  • Studio

Engine

  • Engine overview
  • Build with the Engine
  • Documentation
  • Sampler
  • Hardware
  • Symbolic
  • UI Kit
  • Plugin examples
  • Comparisons

Made for

  • Producers
  • Sound designers
  • Artists
  • Web developers
  • Product teams
  • Hackathon teams
  • Researchers
  • More

Company

  • About
  • Work with us
  • Call for artists
  • Learn
  • Research
  • Jobs
  • Contact
  • Blog

Legal

  • Terms
  • Privacy
Instagram (opens in a new tab)GitHub (opens in a new tab)

Building the future of sound, with you.

© 2026 Maru · private early access