Public beta · 2.5 Beta 12 · macOS 15+ · Apple Silicon & Intel For twenty years, sample browsers have known your files by name. AudioCortex 2 knows them by ear. It listens, organizes, transforms — and its cybernetic assistant, Oscilla, lives inside it and can do all of that for you. A creative instrument that happens to have perfect memory.
Open the disk image and drag AudioCortex into Applications. If macOS says it can’t verify AudioCortex: open System Settings ▸ Privacy & Security, scroll down and click Open Anyway.
$99 during the beta, $149 at 1.0 · free for AudioFinder owners who bought in 2020 or later · seven-day trial · no subscription, ever.
▶ Watch the 60-second tour on YouTube · Read the help (PDF)
The bar above the lanes is laid out by what you’re doing: the set and its time on the left (title, a tempo you drag or type, TAP, the metronome and a 1, 2 or 4-bar count-in), Play, Stop, Record and Loop in the middle, and + Track, Add, Bounce and a ⋯ menu on the right. It now fits beside the Oscilla pane.
Rainawave · made with AudioCortex 2
Rainawave · music made with AudioCortex 2 · only the visuals are made with AI
The sidebar is the order of the work: find material, play, build, watch. A green dot says where the sound is coming from.
Browse, audition and tag hundreds of thousands of sounds; see any one as a waveform or spectrogram and edit it in place. Type ? and describe a sound — "something dark and metallic" — and it finds what sounds like that, whatever the file is called. Select a sound and ask for more like it.
Oscilla lives in every room. Ask it to find a lush guitar, stretch it into a cathedral, lay it on a track and bounce it — and it does the work, tool by tool, in front of you. Use your own Claude, Gemini or OpenAI key, your own model server (LM Studio, Ollama and more), Apple’s on-device model, or Oscilla Local: Iced Audio’s own model, downloaded once, running entirely on your Mac with no key and no account.
Pick a style, key, tempo and length — or describe it — and the Arranger writes sections, parts and chords and plays them on the built-in synths, bar by bar, with the drums on their kit. Mute or solo a lane, swap its sound, shift its octave, change the style while it plays. Record a take and it lands in Tracks as stems, ready to mix.
A 16-step drum machine with kits made from real drums, and a synth with wavetable, analog and FM voices — oscillators, filter, envelope, a hexagonal keyboard. Drop your own samples into a kit folder and they are a kit. You are never starting from an empty library.
Load any two tracks: tempo and key are read, beatgrids drawn, and Sync locks one deck to the other. A crossfader, a level and a speed fader per deck, cues, crates. The waveforms scroll as the music plays — and Oscilla can pull the next track for you.
Layer what you make and what you find. Every lane has a channel strip — gain, pan, sends, inserts with the built-in effects or any AudioUnit, in its own window with its own presets — and a master with a limiter. Bounces add up as you build a song a part at a time. A project keeps the whole take: sounds, lanes, stems and mix in one package.
A patchable canvas: a source file through Veil, into a reverb, out to a file. Every texture engine and effect is a node. Save patches, load them again — or ask Oscilla to wire it for you.
The pitch spiral listens to everything the app plays. Chords bloom as constellations, drones burn as a single spoke, detuning smears off-axis.
The Phylogenetic Explorer classifies your sounds and arranges them as a hex map — kicks beside kicks, pads beside pads, the words on the hexes saying what lives there. Click to dive in, play what you find, record what you play.
A MIDI library that reads every file for tempo, meter, key, feel and parts, and spells out its first chords. Search for bass lines in F# around 160 BPM; open any file as a piano roll; play it on a sound of your choice, or hand it to a song.
Wavetables, analog and FM presets, drum kits — browse them, play them from the keyboard, favourite them, and send one to the song with a click.
This is where AudioCortex 2 stops being a browser and becomes an instrument. Every engine is a tool Oscilla can use, a node you can patch, and a command you can run on a selection.
Time-stretch 2× to 200×. Transients dissolve; a guitar strum becomes a twenty-minute glacier. The engine behind the Monolith and Aquaphoria pieces.
The layered stretch measured from the records we love — a few seconds of guitar becomes a slow, wide cloud that keeps its chords.
Rebuild any sound out of grains of another sound. Paint a melody in whale song.
The impulse response can be a bell, a choir, a mosaic — so your material rings inside other material. The library becomes the room.
Crossing, Freeze, Morph, Veil, Sediment, Wavesets, Ascension, Widen, Resonate. Clean-room spectral processing from the electroacoustic canon, rebuilt in fast native Swift.
Point it at a stem pack or a finished track: it reads the tempo map, the pattern and the chords, and plays reinforcement under it on the built-in instruments.
Many AI music generators are trained on other artists’ songs without permission and make variations of their music. AudioCortex doesn’t do that, and never will. No AI makes the audio: every sound comes from your recordings or from the synthesizers and effects you control.
— Iced Audio · Read our full policy: cybernetic assistance controls and automates, never generates →
Open the disk image and drag AudioCortex into Applications. If macOS says it can’t verify AudioCortex: open System Settings ▸ Privacy & Security, scroll down and click Open Anyway.
AudioCortex 2 is a sound library that listens, an effects rack that dreams, and cybernetic assistance that does the work — in one native Mac app.
Open the disk image and drag AudioCortex into Applications. If macOS says it can’t verify AudioCortex: open System Settings ▸ Privacy & Security, scroll down and click Open Anyway.
2.5 Beta 12 · macOS 15 (Sequoia) or later · Help (PDF) · support@icedaudio.com
Iced