Why We Created Lossless Productions

Julian and I met in Berlin roughly fifteen years ago, in an audio engineering and music production program that felt, at the time, like the most serious place in the world. The city had that effect on you. You were surrounded by people who treated sound like a second language they were determined to speak without an accent. We spent those years deep in it: acoustics, signal flow, the physics behind compression, why certain rooms changed how you heard low end. And then we graduated, kept making music together on and off, kept arguing about the same things producers always argue about.

One of those arguments never really ended.

It started with a simple frustration. We both noticed, independently and then together, that the tools most producers use are designed to hide how they work. A transient shaper with a single knob labeled “Punch.” A compressor where the attack and release are abstracted into three vague settings. A tape emulation where you pick a “vibe” and trust the developer to have gotten the model right. These tools are not necessarily bad. Some of them sound excellent. But when something goes wrong, when the drums disappear in the full mix, when the kick loses body in the second half of the chain, you have no entry point for diagnosis. You just start reaching for different presets and hoping one of them fixes it.

We found this profoundly unsatisfying. Not as a philosophical position but as a practical one. If you do not understand what a tool is doing, you cannot reproduce the result deliberately, explain it to another engineer, or build on it the next session. You are just auditioning accidents.

The idea that kept getting postponed

For years we talked about building something different. A plugin company with a clear line of sight: tools that show their work. Every parameter documented down to the algorithm. Every processing decision explained. The kind of product that treats the user as someone who can handle knowing what a Jiles-Atherton hysteresis model is, or what differential envelope detection actually does to the relationship between attack and sustain.

The problem was cost. Building serious audio software from scratch used to require either a DSP engineer on payroll, a development house with five-figure contracts, or years of learning to code while simultaneously running a studio. None of those were realistic. So the idea sat, revisited every few months over a long chain of messages and the occasional session, never quite disappearing but never quite moving either.

Then something shifted.

Three months with Claude Code

About a year ago we decided to try again, this time with AI-assisted development as the co-author. The honest version of what happened is not a clean success story. It was three months of deep work that involved a lot of wrong turns: DSP choices we had to walk back, algorithms we spent a week on and then replaced, moments where the oversampling sounded correct in isolation and then introduced latency problems we did not catch until late. There were long stretches where we were learning as we built, reading papers on ADAA clipping implementations at midnight and then testing the results the next morning.

But we shipped. And more importantly, we understood everything we shipped. Every module in Lossless Drums (the transient shaper, the tape saturation, the sub bass generator, the compressor, the clipper) was built with that same standard Julian and I had argued about for fifteen years. Nothing works without an explanation. Nothing hides behind a brand name.

I mention the process not to romanticize it but because I think it matters to the argument Lossless is making. We are two people with deep domain knowledge in audio but without formal software engineering backgrounds. The fact that we could co-author something like this, with Claude writing large amounts of the implementation while we pushed back hard on the DSP decisions, rewrote the algorithms we thought were wrong, and spent serious time on the signal chain architecture, is new. That was not available to us before. It changes who can build serious tools.

What we actually built

Lossless Drums is a dedicated drum bus processor. One rack instead of the typical stack of six plugins, each doing something you only half understand.

The rack has seven modules in order: BODY (transient shaper with optional 3-band multiband mode), TAPE (magnetic saturation with three speed characters), SUB (sub bass generator with harmonic and synth modes), AIR (high frequency exciter in dynamic and vintage flavors), VINTAGE (feedback topology bus compressor), CLIP (quintic ADAA clipper plus lookahead limiter), and LOSSLESS (master section with oversampling, gain match, and metering). The first six are drag-reorderable. LOSSLESS always closes the chain.

Every module is designed around the principle that the relationship between input and output should be legible. BODY uses differential envelope detection, a fast and a slow follower running in parallel with their difference isolating the attack from the sustain, so when you move Punch you are moving attack energy and nothing else. SUB gives you two explicit mechanisms: a resonant filter that self-excites in a controlled way, and a zero-crossing sub octave generator that follows the input envelope. You choose the method, you set the frequency, you control the release. No mysterious “add sub” button.

CLIP combines a quintic polynomial clipper with ADAA anti-aliasing and a 1.5 ms lookahead limiter. The LOSSLESS section includes RMS-based gain matching and a latency-compensated dry path so your A/B comparisons are not corrupted by loudness differences. These decisions compound. When every module plays honestly, you can actually trust what you hear.

Oversampling runs at ECO (2x), STANDARD (4x IIR), HIGH (4x FIR), or ULTRA (8x). The choice is yours depending on whether you are tracking live or running a final export. Version 0.9.5 ships for macOS as VST3, AU, and Standalone on Apple Silicon and Intel. Windows and CLAP are on the roadmap.

Why transparency is not just a design preference

I want to say something about the culture that shaped this, because I do not think the technical decisions exist independently of it.

Presets are not the enemy of creativity. But there is a version of the preset economy that quietly erodes something important. When your whole production environment is optimized for speed and surprise (drag in a loop, load a plugin, hit play, decide if you like it) you can spend years making music without accumulating much knowledge. Every session is a new search. The moves that worked last time live in a project file, not in your hands. You are dependent on the catalog staying available, on the preset matching the context, on your memory of which plugin that kick came from.

This is not a moral argument. It is an architectural one. What kind of producer do you want to be in five years? One who has logged thousands of hours with a deep understanding of how transients, saturation, and low-end interact, or one who has auditioned thousands of presets and remembers a few that worked?

Lossless is a bet on the first path. Not because it is more virtuous but because we believe it produces better results over time, and because it is more interesting. Engineering your sound is harder than selecting it. It is also more durable.

Who this is for

Not everyone, and we are fine with that.

If your goal is to work fast and the results are good enough, there are tools built specifically for that workflow. They are often excellent. This is not a polemic against them.

Lossless is for producers who want to know why something worked. Who are frustrated by the gap between what their tools are supposed to do and what they can actually explain about the output. Who want a compressor where changing the ratio produces a predictable, audible, understandable result. Not a magical one.

If you have ever rebuilt a session from scratch just to understand the chain better, or spent an afternoon reading about Fletcher-Munson curves when you could have been laying down a track, you are probably who we built this for.

The signal

Julian and I started this in a Berlin classroom, arguing about acoustics and room modes and what it means to actually understand the physics of what you are doing rather than just reacting to it. Fifteen years later the argument is the same. The tools are different.

If you want to be part of that, one clear email at a time, no affiliate noise, no preset spam: The Signal is where the thinking happens.

I Want to Join The Signal