Music Technology ยท Built by AI West

A new kind of instrument.
Four layers of mastery.

SoundForge combines professional DSP effects with a generative AI conductor via MCP. Claude orchestrates the emotional arc of a performance -- the instrument plays the notes, AI shapes what they mean. Nothing else does this.

The problem

โ€œI wanted to explore generative composition. Nothing bridged performance and AI narrative.โ€

๐ŸŽ›๏ธ

Modular synthesis was too complex. Preset-based tools were too shallow. Nothing let a performer shape tone and have AI respond to the emotional intent.

๐Ÿค–

Existing AI music tools generated audio independently. None integrated with a live performance signal or responded to what you're actually playing.

๐ŸŽต

The gap between human performance and AI composition was a wall. SoundForge made it a conversation.

Inside the instrument

Click any knob. Navigate any phase.

Click a knob on the Effects screen to reveal its slider. Switch to AI Conductor and click any phase dot to navigate the emotional arc. Expand to go full screen.

๐Ÿ”’soundforge.aiwest.co/instrument
S

SoundForge

AI Instrument

JM

Josh M.

AI West

Effects Chain5 active ยท CPU 34% ยท 2.1ms
Obsidian Crush
Distortion
drive72
tone45
mix60
Amber Delay
Time
time55
fb68
mix40
Jade Reverb
Space
size80
decay62
mix55
Ruby Tremolo
Modulation
rate35
depth78
shape50
Sapphire Chorus
Modulation
rate42
depth55
mix65
Onyx Gate
Dynamics
thresh60
attack30
release45

Signal Chain

Obsidianโ†’Amberโ†’Jadeโ†’Rubyโ†’Onyx

4

System layers

2.1ms

Latency

โˆž

Emergence

The architecture

Four layers. One instrument.

๐ŸŽ›๏ธ

Layer 1 -- DSP effects engine

Six professional-grade effects built in Faust DSP -- distortion, delay, reverb, tremolo, chorus, gate. Sub-millisecond latency. Runs standalone or as a VST3/AU plugin.

๐ŸŽน

Layer 2 -- Tonnetz generative engine

A harmonic generation system based on neo-Riemannian Tonnetz theory. AI selects chord voicings that respond to the emotional context set by the conductor.

๐Ÿค–

Layer 3 -- Claude conductor via MCP

Claude receives the performance state and emotional arc, then sends real-time instructions back to the instrument. It doesn't generate audio -- it shapes what the instrument does.

๐Ÿ”„

Layer 4 -- Emergent feedback loops

Each layer feeds back into the others. The DSP state influences harmonic generation, which influences AI instructions, which influence DSP. The system evolves as you play.

Built with

Real infrastructure.
Not duct tape.

Faust DSPJUCE / VST3Claude MCPTonnetz TheoryWebAudio

Want something built like this
for your business?

This is one of seven systems we've built. Every project starts with understanding what's actually costing you time or money -- then we build the thing that solves it.