About noshigeSound
Recreating vintage gear at the waveform level.
noshigeSound builds pro-grade audio plugins to do one thing: archive, in the digital domain, the singular sound DNA of vintage analog gear — its impossibly smooth, unmistakably strong character — accurately enough to null against the real hardware’s waveform, so it never goes out of production. Made by an AI engineer with a deep love for classic analog gear.
01
A complete digital archive.
Train a massive AI on every response of decaying vintage gear — at the waveform level — to create and preserve a complete digital archive that outlives the circuits.
02
The magic of analog hardware, within everyone’s reach.
Bring an experience close to analog gear, easily, to those who can’t use it for whatever reason — and to those who don’t know it yet.
Why
Why we make these.
Most AI modelling plugins capture one snapshot of a piece of hardware — a fixed curve, an impulse response, a model trained at one setting. They sound right at a glance, then start to drift the second you turn a knob.
We took a different path. AI parametric modelling treats the gear as a system: every knob, every nonlinearity, every interaction. What you get is a digital twin that generates a waveform as close to the real unit’s as it gets — playable in your DAW, behaving the way the gear does in the rack.
noshige76 was the first plugin built end-to-end on this approach — an emulation of the classic "76" limiter that has been on more records than anyone can count. noshige2A brought the same treatment to its optical counterpart, and Twin-GTR then applied it to guitar amps and pedals. The Classic line is the originals; TwinGear is where we’re going next.
Founder
How this began.
In 2020, as a college student, I produced my college festival’s official theme song with my own band and recorded it myself in Tokyo studios. Watching the mix take shape taught me something I never let go of: however good the song, the mix decides how hard it moves you. I fell for mixing — and for the vintage analog gear behind that sound, nearly all of it far beyond a student’s reach.
I was a law student, but a friend in the sciences introduced me to machine learning, and one idea took hold: what if AI could learn great tone? I taught myself programming in 2021, and by 2022 I had models learning the behaviour of a real compressor and a real guitar amp from gear data — iterating relentlessly to get real-time performance and sound quality in the same model. From 2023 I worked as a professional AI engineer building speech-synthesis systems, sharpening the exact craft these plugins are built on.
In 2026, AI-assisted development finally made a one-person plugin company viable, and I decided to release properly. My motivation hasn’t changed since that first studio session: put the texture of analog within reach of people who — like my past student self — can’t rack up the hardware, and archive the DNA of great gear before it stops working.
Hear the 2020 song that started it all
— Sota Tano, founder
TwinGear
Where this is going.
TwinGear is the technology brand behind Twin-GTR — and, soon, TwinGear Cloud. With Cloud you capture the response of your own amp, preamp, or pedal at your bench, upload the recordings, and get back a custom .napx module — modelled the same way our flagship gear-twins were, encrypted end-to-end and bound to your account. It loads into Twin-GTR (and every future custom-gear-capable plugin in the line) like the bundled twins do.
The promise is durable: hardware can break, vintage gear gets harder to source every year, and your favourite signal chain shouldn’t depend on parts going out of production. A TwinGear twin sticks around as long as your account does.
TwinGear Cloud releases in 2026 Q3. Pricing model still being decided — we’ll publish before opening the doors.