NK PRODUCTIONS

ALA-101 Hybridization

Vintage Warmth, Modern Precision.

The Ultimate Hybrid Analog Coloring Tool

Separated neural modeling of vacuum tube & transformer stages — fused with real-time inference and modern digital precision.

VST3 AU AAX macOS Universal Windows x64 Up to 192kHz
ALA-101 Hybridization Plugin Interface
Features

Hybrid Separated Modeling

Every analog stage — individually captured, independently controlled, digitally fused. Not a black-box emulation, but a fully transparent hybrid system. Model reproduction accuracy of 99.8% recorded (*at nominal level, in specific frequency ranges).

Anti-Aliasing

ADAA — Antiderivative Anti-Aliasing

The Clipper and Standard mode both employ ADAA (Antiderivative Anti-Aliasing) to eliminate aliasing artifacts at the source — not by oversampling alone, but by mathematically computing the antiderivative of the waveshaping function. This means clean, artifact-free saturation even at lower oversampling rates. Combined with a precision-designed compensation FIR filter, the result is transparent, phase-accurate nonlinear processing that preserves the original signal integrity.

Developer

Built by an Engineer, for Engineers

Naruki

Audio Engineer / Developer

Recording and mixing engineer who captured every tube, transformer, and clipping characteristic from his own analog line amplifier. Every model in ALA-101 is designed from hardware he uses daily in professional sessions.

Story

From Hardware to Plugin

I'm an audio engineer by trade; programming was only a hobby. About four years ago, I began studying signal capture and neural network training on my own, and eventually hit the limits of modeling. I'd been using my own modified tube line amplifier, but in recent sessions, even the slightest L/R imbalance would introduce unintended stereo image shifts. When capacitors or tubes weren't fully stable, the sound would feel somehow dull. Recalling settings was tedious. Analog gear was becoming increasingly difficult to use actively. So I thought — why not turn my own hardware into a plugin? I kept writing code, experimenting with architectures and training methods. With persistence and patience, I was able to break through the walls I'd hit before — with my own designs and training approaches. I'm proud to finally release this product.

Visual Feedback

Every Detail, at a Glance

Real-time visual feedback on every knob and meter. Signal level, gain reduction, and saturation — always visible, never guessing.

Arc Meter
Dual arc meter with cyan-yellow-red gradient. Instantly see how much processing is applied. 50% is nominal.
Level-Reactive Knob
Knob arc color reacts to real-time signal level. Instantly see proper gain staging at a glance.
Gain Reduction
Clip Drive knob shows gain reduction in real time. Arc color changes as the signal hits the clipper. Instantly see clipper response and GR amount.
How to Use

Gain Staging is Essential

This plugin reproduces the behavior of real analog gear — proper gain staging is essential to achieve the designed performance. The processing reacts dynamically to input level, just like hardware. You may deviate from the reference, but use the Compensate Gain button to automatically set your input to the reference level, and Auto Level will automatically match the output loudness after processing. These two features ensure a proper starting point and fair A/B comparison. When pushing the signal hard into the Tube or Trans parameters, the organic analog-like compression behavior strongly affects the output, and the perceived level tends to decrease. Since a lower output level means lower perceived loudness, use Auto Level after adjusting parameters to match the listening level. Meter monitoring is also available, and the color-reactive knobs and arcs provide the fastest visual feedback, making manual level adjustment easy as well. Colors shift from blue (safe) through yellow (caution) to red (warning). For knobs that represent the amount of change, 50% is the nominal reference point. This plugin is designed to let you perceive subtle differences, so make full use of the independent Bypass on each section to isolate and compare what each stage contributes to the sound.

Interface

Precision Control, Zero Guesswork

Every section independently bypassable, linkable to master bypass, with real-time visual analysis built in.

Input Section

Input

Input Trim with level-reactive knob. One-click Compensate Gain to reference level. Unity Gain locks input/output balance. Bypass Linked toggles all linked sections for instant A/B comparison.

Output

Output

Output Trim with level-reactive knob. Auto Level matches loudness after processing. Clip > Out routes the clipper post-output for mastering. Delta monitors the difference signal.

RMS Meter

RMS Meter

Dual arc meter displaying input and output RMS levels simultaneously. Color gradient from cyan (safe) through yellow (caution) to red (warning). Use for gain staging.

Peak Meter

Peak Meter

Sample-accurate peak meter for input and output. Color-coded bars with the same cyan-yellow-red gradient. Click to reset peak hold.

Tube Drive

Tube Drive

Dynamic warmth, even-order harmonics, and creamy saturation from the captured tube input stage. L/R or M/S expansion mode. Independent bypass and link.

Transformer Tolerance

Transformer Tolerance

Core hysteresis, low-end heft, and magnetic presence from the captured transformer output stage. Pattern A/B for different hardware characters. Independent bypass and link.

Analog Expansion

Expansion & Fluctuation

Analog Expansion controls L/R channel variation from matched (0%) to full hardware character (100%). Fluctuation adds the subtle, organic movement found in analog equipment.

Hysteresis & Bloom

Hysteresis & Bloom

Enhances low-end response from transformer hysteresis — the organic "breathing" of analog iron. Bloom adds a living, reactive sense of depth and movement. Shape the low-end clarity — tighten the focus or let it breathe.

Low Shelf

Low Shelf

Hardware-measured low shelf EQ (fc=125.8Hz, Q=0.603). Pre/Post Trans position option. Independent bypass and link.

High Shelf

High Shelf

Hardware-measured high shelf EQ (fc=2212.5Hz, Q=0.352). Pre/Post Trans position option. Independent bypass and link.

HPF Resonance

HPF Resonance

Morph transformer HPF Q from flat (0.707) to captured vintage peak. Selectable 5.0Hz (Transformer) / 5.47Hz (Tube). Impossible with analog hardware.

LPF Resonance

LPF Resonance

Morph tube LPF Q from flat (0.707) to captured vintage peak. Scalable cutoff: 18.97 / 37.95 / 75.90kHz by sample rate. Impossible with analog hardware.

Clip Drive

Clip Drive

Recreates clipping characteristics measured from the actual hardware. Real-time GR arc. Pre/Post HPF position selector. Independent bypass and link.

Clipper Options

Clipper Options

Character morphs the knee from soft saturation to hard clipping. Symmetry controls even-order harmonic balance (H2/H4).

EQ Response Overlay

EQ Response

Real-time composite frequency response of Low Shelf, High Shelf, LPF, and HPF. See the combined EQ curve at a glance.

Clipper Character Overlay

Clipper Analysis

Real-time transfer curve, gain reduction, and harmonic analysis (H2–H5). See exactly how Character and Symmetry shape the harmonics.

Model Modes

Six Tiers of Neural Inference

Choose your balance between CPU efficiency and sonic fidelity. From waveshaper to ultra-high-precision real-time inference.

Standard
Waveshaper
CPU
Quality
Hybrid
WaveNet
CPU
Quality
Professional
Real-time Inference
CPU
Quality
Secret Mode
Hybrid+
WaveNet
CPU
Quality
Professional+
Real-time Inference
CPU
Quality
Extreme
Real-time Inference
CPU
Quality
Secret Mode Menu

Enable Secret Mode in the settings menu to unlock Hybrid+, Professional+, and Extreme.

Standard

A conventional modeling approach with minimal CPU load. Uses input-signal-level-based processing, capable of reproducing per-level behavior. However, dynamic behavior reproduction is limited.

Hybrid

/ Hybrid+

Built on the same architecture now widely used in NAM (Neural Amp Modeler) and similar proven capture-based modeling tools. While conventional NAM excels at static tone reproduction, it cannot capture signal-level-dependent dynamic behavior. Hybrid modes extend this with a proprietary conditioning design that enables dynamic response to input level — bringing the architecture beyond static capture into living, reactive modeling that captures time-varying behavior.

Professional

/ Professional+ / Extreme

Hybrid is a truly hybrid design that prioritizes CPU efficiency and effective processing, but Professional and above focus on time-varying dynamic behavior. While Hybrid offers high static reproduction fidelity, it also has architecture-inherent limitations. Professional modes prioritize capturing temporal behavior — how the hardware responds moment to moment. As CPU load increases through Pro, Pro+, and Extreme, transient accuracy and high-frequency reproduction improve progressively. Analysis in specific modes recorded waveform reproduction accuracy of 99.90% (99.63% by ESR evaluation, 99.78% by RMS measurement). Note: these modes do not use conventional processing, so real-time performance may be limited depending on your session. A high-performance computer or a DAW with efficient thread distribution may run Extreme at 4x oversampling without issue, but please be aware this is a fundamentally different processing approach. Running multiple instances in a single session may be difficult in some situations.

Signal Flow

Transparent Processing Chain

Every stage visible. Every stage independently controllable. No hidden processing, no unwanted coloration.

Tube
[ Shelf ]
Trans
Hysteresis
Bloom
Low Shelf
High Shelf
Swappable
Clipper
HPF
LPF
[ Post Clip ]
Neural Network
Movable
Swappable
FAQ

Frequently Asked Questions

What formats and platforms are supported?

macOS: VST3, AU, AAX (Universal Binary — Apple Silicon and Intel). Windows: currently VST3 only (x64). Sample rates up to 192 kHz. Windows AAX support may be added in the future if there is sufficient demand. The AAX DSP architecture is not compatible with modern real-time processing, so AAX DSP is not available.

How does the license work?

One license key activates up to 3 machines. Online activation is required. You can deactivate and reactivate through the plugin's settings menu or by contacting support. Once activated, the plugin works offline for 30 days. If no internet access is detected after 30 days, a 30-day grace period begins. After a total of 2 months offline, reactivation is required.

What are the trial limitations?

The trial version is fully functional but periodically injects short noise bursts into the audio output. No time limit — evaluate as long as you need.

Professional mode is causing CPU spikes. What should I do?

Professional, Professional+, and Extreme modes use per-sample real-time inference, which is fundamentally different from conventional DSP processing. Try lowering the oversampling rate (4x → 2x → 1x), or switch to Hybrid mode for a lighter alternative with similar character. Increasing the audio buffer size may have very limited effect. DAWs with excellent thread distribution (such as Logic Pro or Reaper) have achieved multiple instances running at Extreme 4x.

Can I run multiple instances?

Standard and Hybrid modes can run multiple instances without issue. Professional and above may be limited depending on your system. For heavy sessions, use Standard or Hybrid on most tracks and reserve Professional for key channels.

Why does the output level drop when I increase Tube Drive?

This is the natural compression behavior of the analog tube circuit being faithfully reproduced. Use the Auto Level button to automatically match the output loudness after adjusting parameters.

Do I need to perform gain staging?

Yes. This plugin reacts dynamically to input level, just like real hardware. Use Compensate Gain to set the input to the reference level for optimal performance. The reference level can be configured in settings (+4 dBu = -18 dBFS or -20 dBFS).

Is there a refund policy?

As a digital product, all sales are final. We strongly recommend evaluating the trial version thoroughly before purchasing. See our EULA for details.

Are there plans to support GPU Audio?

While our proprietary hybrid architecture could potentially be adapted to GPU processing, the current GPU Audio SDK alone does not provide the functionality needed to support our processing structure. We are monitoring the development of GPU Audio and will consider support if the SDK matures to a level that makes integration feasible.

Technical Specifications

Built for Professionals

Engine
Real-time Neural Network Inference
Precision
64-bit Float / Double
Anti-Alias
ADAA (Antiderivative Anti-Aliasing)
Oversample
1x / 2x / 4x
Latency
Fixed (hot-swappable across OS rates)
Formats
VST3 / AU / AAX (macOS), VST3 (Windows)
Sample Rates
Up to 192 kHz
Platforms
macOS (Universal Binary), Windows x64
UI
Resizable, Retina / HiDPI