VST3 / AU 100% Local All Platforms

Mix with AI Intelligence

MixEngine AI lives on every track in your DAW. It analyzes your full mix in real-time, detects your hardware, and delivers intelligent mixing suggestions — all powered by local AI. Your audio never leaves your machine.

Download Free View Plans

Works with every major DAW

Ableton Live Logic Pro Reaper FL Studio Pro Tools Cubase Studio One Bitwig

Why MixEngine AI

Built different from every other mixing plugin

100% Local AI

Ollama runs on your machine. Your audio never touches the cloud. Total privacy, zero latency dependency.

Sees Your Entire Mix

Plugin on every track reports to a local server. The AI knows what every instrument is doing — not just one track.

Smart Hardware Detection

Automatically detects your GPU and RAM, then selects the best AI model for your system. No configuration needed.

Learns From You

Rate AI suggestions with stars. Your feedback trains the system to match your mixing style over time.

Uses YOUR Plugins

Host your own VST3 plugins inside the effect chain. MixEngine AI suggests — your plugins do the processing.

24 Genre Profiles

Each genre has specific target LUFS, dynamic range, EQ curves, and compression rules. The AI adapts to your genre.

How It Works

Three components working together

1

Plugin on Each Track

Place MixEngine AI as a VST3/AU plugin on every track. Each instance analyzes audio and reports to the local server.

2

Local Server Orchestrates

A Python/FastAPI server aggregates all track data, builds mix context, and communicates with Ollama for AI inference.

3

AI Generates Recommendations

The LLM analyzes the full mix and returns structured suggestions: gain, EQ, compression, panning, sends, and bus assignments.

DAW
Plugin on every track
WebSocket localhost:8432
Local Server
Python / FastAPI
HTTP localhost:11434
Ollama
Local LLM
Real-Time Analysis

Real-Time Analysis

6 analyzers running continuously on every track

FFT Spectrum

4096-point, 32 bands, 100ms refresh

Level Metering

Peak, RMS, crest factor, 4x oversampling

LUFS Loudness

Momentary, short-term, integrated (BS.1770)

Spectral Analysis

Centroid, rolloff, flux, brightness

Stereo Image

Correlation, width, phase coherence

Transient Detection

Attack, sustain, dynamic range

AI Architecture

AI Mixing Assistant

The AI understands your entire mix and suggests per-track adjustments

What the AI suggests per track

Gain staging (dB)
Parametric EQ (freq, gain, Q per band)
Compression (threshold, ratio, attack, release)
Stereo panning (-100 to +100)
Send routing (bus, level)
Sidechain assignments
Mid/Side processing
Mix bus assignment

Multi-phase workflow

1

Phase 1: Gain staging across all tracks

2

Phase 2: Frequency cleanup (EQ)

3

Phase 3: Dynamics control (compression)

4

Phase 4: Spatial positioning (pan, sends)

5

Phase 5: Final adjustments and bus routing

Smart Hardware Detection

MixEngine AI automatically selects the best model for your system

Hardware Model Quality
GPU VRAM >= 10GB gemma3:12b Maximum
GPU VRAM >= 6GB llama3.1:8b High
No GPU, RAM >= 16GB llama3.1:8b High
RAM >= 8GB llama3.2:3b Good
RAM < 8GB llama3.2:1b Basic
Auto Mix

Auto Mix

Press the Auto Mix button and MixEngine AI handles the entire workflow automatically. Analyze all tracks, generate suggestions, apply them, and let you review each decision.

1

Analyze all tracks (collect data from every instance)

2

Generate mixing hints (AI reasoning)

3

Apply suggestions (gain, EQ, comp, pan, sends)

4

Review & rate each suggestion (accept/reject with stars)

Feedback loop (ratings improve future suggestions)

RTNeural

ML Inference Engine

RTNeural: real-time neural network inference in the audio thread

Auto-EQ Model

A trained neural network that takes spectral analysis as input and outputs per-band EQ corrections. 60-dimension input vector, less than 0.1ms processing time.

Auto-Compressor Model

Predicts optimal compression parameters (threshold, ratio, attack, release) based on transient analysis and genre profile.

< 0.1ms
Latency
60-dimension input
Input vector
Audio thread safe
Threading
Processing Tools PRO

Four AI-powered engines, all running locally

Noise Reduction

RNNoise neural denoiser. Real-time processing, zero added latency. Removes background noise while preserving voice and instrument clarity.

Source Separation

Demucs v4 stem extraction: vocals, drums, bass, other. Offline processing with progress bar. Export individual stems.

Audio to MIDI

Basic Pitch (by Spotify) for polyphonic audio-to-MIDI conversion. Drag and drop MIDI file export directly into your DAW.

Auto Mastering

Matchering engine for automated mastering. Match your track's loudness and frequency profile to any reference track.

24 Genre Profiles

Each genre has calibrated targets for LUFS, dynamic range, EQ curves, and compression behavior

Genre Target LUFS Dynamic Range
Auto-Detect Adaptive Adaptive
Pop -9 to -11 6-8 dB
Rock -10 to -12 8-10 dB
Metal -8 to -10 6-8 dB
Hip-Hop -8 to -11 6-9 dB
R&B -10 to -12 8-10 dB
Electronic / EDM PRO -7 to -10 5-8 dB
House PRO -8 to -10 6-8 dB
Techno PRO -7 to -9 5-7 dB
Drum & Bass PRO -7 to -9 6-8 dB
Dubstep PRO -7 to -9 6-9 dB
Jazz PRO -14 to -18 12-18 dB
Classical PRO -18 to -24 15-25 dB
Country PRO -10 to -13 8-12 dB
Folk / Acoustic PRO -12 to -16 10-14 dB
Blues PRO -12 to -15 10-14 dB
Reggae PRO -10 to -12 8-10 dB
Latin PRO -9 to -12 7-10 dB
Ambient PRO -16 to -22 12-20 dB
Lo-Fi PRO -12 to -15 8-12 dB
Podcast / Voice PRO -16 to -19 8-12 dB
Film Score PRO -18 to -24 15-25 dB
Game Audio PRO -14 to -18 10-16 dB
Worship / Gospel PRO -10 to -14 8-12 dB

Plugin Interface

7 tabs for complete mixing control

Auto Mix

One-click AI mixing with review workflow

Analysis

All 6 analyzers with visual displays

Strip

Channel strip view with gain, EQ, comp, pan

Rack

VST3 effect chain with drag-to-reorder

Sends

Send bus routing and level control

Buses

Mix bus assignment and group processing

Settings

AI model selection, genre, preferences

Simple, transparent pricing

Pay once, own forever. No subscriptions.

Free

€0 / forever

Everything you need to start mixing with AI.

  • All 6 real-time analyzers
  • AI mixing recommendations (text + explanations)
  • Channel strip view (read-only)
  • Max 2 plugins in effect chain
  • Max 2 send buses
  • 6 genres (Auto, Pop, Rock, Metal, Hip-Hop, R&B)
  • Basic Auto Mix
  • 3 custom presets
  • Default AI model (auto-selected)
  • Rate limit: 1 analysis / 10 seconds
  • 1 device
Download Free
BEST VALUE

Pro

€79,00 / one-time

Perpetual license for v1.x — all minor updates included.

  • Everything in Free, plus:
  • Unlimited plugins in effect chain
  • Unlimited send buses
  • All 24 genres with specific rules
  • Sidechain compressor & gate
  • Mid/Side processing
  • Reference track comparison
  • Mixing techniques library + auto-apply
  • Master chain (clipper + true peak limiter)
  • Automation recording & playback
  • AI model selection (choose any model)
  • Unlimited presets + cloud sync
  • Rate limit: 1 / 3s (burst up to 3)
  • Processing tools (Denoise, Stems, MIDI, Mastering)
  • 3 devices
  • Lifetime minor version updates
  • Major version upgrade: €39
Buy Pro — €79,00

System Requirements

MixEngine AI runs on Windows, macOS, and Linux

Minimum

CPU Intel i5 / AMD Ryzen 5
RAM 8 GB
Disk 5 GB free
OS Windows 10+ / macOS 11+ / Ubuntu 22.04+
GPU Not required

Recommended

CPU Intel i7 / AMD Ryzen 7
RAM 16 - 32 GB
Disk SSD with 10 GB free
OS Latest version of your OS
GPU NVIDIA (CUDA) / AMD (ROCm) / Apple Metal

Compatible DAWs

DAW Minimum Version Format
Ableton Live 11+ VST3 / AU
Logic Pro X+ AU
Reaper Any VST3
FL Studio 20+ VST3
Pro Tools 2021+ VST3
Cubase / Nuendo 12+ VST3
Studio One 6+ VST3 / AU
Bitwig Studio 4+ VST3

One-Click Installation

The installer handles everything automatically

1

Installs the VST3/AU plugin to your system plugin folder

2

Installs the local Python/FastAPI server

3

Installs Ollama (if not already present)

4

Downloads the default AI model for your hardware

5

Installs RTNeural models (Auto-EQ, Auto-Compressor)

6

Registers the local server as a system service

7

Verifies everything works with a quick self-test

Frequently Asked Questions

What is MixEngine AI?
MixEngine AI is a VST3/AU plugin that you place on every track in your DAW. Each instance analyzes the audio on its track and sends data to a local server. The server aggregates all track data to build a complete picture of your mix, then uses a local LLM (via Ollama) to generate intelligent mixing suggestions — gain, EQ, compression, panning, sends, bus assignments, and more. All AI runs locally. Your audio never leaves your machine.
What is Ollama?
Ollama is an open-source tool that lets you run large language models (LLMs) locally on your computer. MixEngine AI uses Ollama to power its mixing intelligence. The installer sets up Ollama automatically — you don't need to configure anything. It runs as a background service on localhost:11434.
Do I need an internet connection?
Not for mixing. All AI inference runs locally via Ollama and RTNeural. You only need internet for: initial installation (downloading AI models ~4-8GB), license activation (once every 30 days via heartbeat), cloud preset sync (Pro only), and checking for updates.
How much disk space does it need?
The plugin itself is about 50MB. Ollama and a default model require 4-8GB depending on which model is selected. The local server (Python/FastAPI) is about 200MB including dependencies. Total: approximately 5-9GB.
Does it work without a GPU?
Yes. Ollama can run on CPU only — it will just be slower. With 8GB+ RAM, llama3.2:3b runs well on CPU. With a GPU (NVIDIA CUDA, AMD ROCm, or Apple Metal), you get significantly faster inference and can run larger models like gemma3:12b.
Can I choose the AI model?
Free users get an auto-selected model based on their hardware. Pro users can choose any compatible Ollama model from the plugin settings. The system detects your GPU VRAM and RAM to recommend the optimal model, but you can override this.
Does the plugin add latency?
The analysis runs in a separate thread and does not add latency to your audio signal path. RTNeural models (Auto-EQ, Auto-Compressor) run in the audio thread with less than 0.1ms processing time. The effect chain adds latency only from the hosted VST3 plugins (with automatic delay compensation).
How many tracks can I analyze?
There is no hard limit. Each plugin instance is lightweight (~5MB RAM, <1% CPU for analysis). The local server aggregates data from all instances. Practical limit depends on your system — most users run 20-60 tracks without issues.
How does the license work?
Free: 1 device, no purchase needed. Pro (€79, one-time): perpetual license for v1.x with all minor updates and bug fixes. Up to 3 devices. When v2.x releases, upgrade for €39. License verification happens via heartbeat every 30 days — works offline between checks.
Is my audio data safe?
Absolutely. All audio processing happens locally. Ollama runs on localhost, RTNeural runs in the plugin process. We never receive, store, or process your audio. The only data sent to our servers is license verification (license key + device fingerprint) and preset sync metadata (Pro only).
What DAWs are supported?
Any DAW supporting VST3 or AU: Ableton Live 11+, Logic Pro X+, Reaper (any version), FL Studio 20+, Pro Tools 2021+ (via VST3), Cubase/Nuendo 12+, Studio One 6+, Bitwig Studio 4+. Linux support is VST3 only (Reaper, Bitwig, Ardour).
What's the difference between Free and Pro?
Free gives you all 6 analyzers, AI mixing recommendations (text with explanations), read-only channel strip, 2 plugins in effect chain, 2 send buses, 6 genres, basic Auto Mix, 3 presets, and 1 device. Pro adds unlimited plugins and sends, all 24 genres, sidechain compressor/gate, mid/side processing, master chain, automation, reference track, mixing techniques library, processing tools (denoise, stems, MIDI, mastering), AI model selection, unlimited presets, cloud sync, and 3 devices.

Still have questions?

[email protected]

Start mixing with AI today

Download the free installer for your platform. Upgrade to Pro anytime from your dashboard.

Windows
.exe — ~50 MB
macOS
.pkg — Coming soon
Linux
.deb — Coming soon

The AI model (~1-8 GB depending on your hardware) is downloaded automatically on first launch. No configuration needed.

Requires: DAW with VST3/AU support, 8GB RAM minimum, 5GB free disk space

Stay Updated

Subscribe to our newsletter for updates, new releases, and exclusive offers.

Successfully subscribed!