Evo

Evo is a section of this site dedicated to audio workflow experiments that are in active development. The reference pages elsewhere on the site are organized around topics where the knowledge is settled enough to present systematically -- what scheduling latency is, how a benchmark tool works, what the LAD mailing list community produced. Evo is different. It is for the work that is still evolving: combinations of tools that are interesting but not yet fully understood, synthesis approaches that are promising but not yet documented as methodology, signal chain experiments where the results are real but the conclusions are still being worked out. The name is short for evolving, and that is the operative word. Expect this section to change over time as experiments mature, get abandoned, or get promoted into more complete reference material elsewhere on the site.

The framing here is intentional. Audio workflow experimentation is a legitimate form of practice alongside the more structured work of benchmarking and system evaluation. The process of trying things, noting what works and what does not, and building up a personal vocabulary of techniques that fit your actual way of working is how practical knowledge gets built. Evo is where that process is documented rather than tidied up. If you are looking for finished methodology, the sections linked at the end of this page are the right starting points. If you are looking for work in progress and are comfortable with the incompleteness that implies, read on.

What Evo Covers

The experiments documented in Evo fall into a few recurring areas. Signal chain construction -- the specific sequence of processors, routing decisions, and gain staging choices that produce a particular kind of audio output -- is a persistent thread. A signal chain that works well for one type of material often fails instructively for another, and the failure modes are as worth documenting as the successes. Understanding why a chain breaks down for a given input type is usually more informative than knowing that it works for the cases it was designed for.

Synthesis approaches are a second focus. The Linux audio ecosystem has a wide range of synthesis tools available, from the physically modeled to the purely abstract, and the relationship between synthesis architecture and musical result is often non-obvious. The experiments here explore what specific synthesis configurations actually sound like under extended use -- not just what they are theoretically capable of, but what kinds of material they are genuinely suited to and where they start to resist. For deeper technical background on synthesis engine architecture, the notes on the Audiality engine cover the design decisions that govern real-time synthesis behavior in a constrained environment.

Current Focus Areas

The current set of active experiments centers on three areas. First, lightweight modular routing in a JACK environment: how many modules can you run in a coherent signal graph before the cognitive overhead of managing the graph starts to outweigh the flexibility it provides. The answer depends on what you are building and how experienced you are with the modules involved, but there is a general threshold effect that is worth understanding and documenting. Second, the interaction between synthesis-generated audio and sampled material in the same session: the phase alignment and spectral interaction issues that arise when you mix synthesis output with tracked audio are real and not always solved by standard mixing approaches. Third, latency compensation in practice: not the theory, which is well documented elsewhere, but the actual behavior of latency compensation in different host configurations and the cases where it falls down.

These are not separate tracks that happen to be listed together. They interact. The routing complexity experiment affects how you can manage latency compensation when the signal graph is deep. The synthesis-versus-sampled interaction question has different answers depending on the routing architecture you are working in. Documenting them as separate experiments is a practical organization choice, not a reflection of how they actually relate. The connections between them are part of what Evo is here to work out.

Audio Quality in Experimental Work

One risk in workflow experimentation is losing track of signal quality while focused on getting a technique to work at all. The first pass through a new signal chain is often more concerned with whether the output sounds like the intended thing than with whether the signal path is clean and the dynamic range is being used well. Those are different questions, and the second question matters in proportion to how seriously you intend to use the results. Experiments documented in Evo include notes on signal quality where it is relevant, and where a technique has quality implications that are not obvious from the functional description, those implications are flagged. For the systematic reference on audio signal quality evaluation, the audio quality section covers the measurement and assessment methodology in detail.

How Evo Relates to the Rest of the Site

Evo feeds into the rest of the site rather than standing apart from it. When an experiment in Evo reaches a point where the conclusions are stable enough to be written up as methodology, that writeup goes into the relevant reference section. When an experiment reveals a question that the existing reference material does not answer, that question becomes a prompt for new material. The relationship is productive in both directions: the reference sections provide the foundation that makes experiments interpretable, and the experiments surface gaps and questions that the reference sections then address.

The longer-form writing in the notes section sometimes emerges from Evo material that has matured past the experiment stage but does not yet have a natural home in a specific reference section. Notes are where the analytical and contextual work happens -- the kind of writing that explains not just what a technique is but why it matters and what it connects to. If you follow this site over time, you may recognize ideas that first appeared in Evo reappearing in more developed form in notes, and eventually as settled methodology in reference sections. That is the intended trajectory for experiments that pan out.

Working With Experimental Material

The appropriate stance toward Evo material is one of critical engagement rather than direct adoption. The techniques documented here are things I have tried and found interesting, not things I am recommending as general best practice. Some will prove out and become best practice. Others will turn out to have limitations that make them useful only in specific circumstances. A few will turn out to be wrong in ways I have not yet discovered. That is how experimental work goes, and the documentation reflects the current state of the investigation rather than a final verdict.

If you try something from Evo and it behaves differently than documented, that is useful information. Different systems, different tool versions, different signal material, different monitoring environments -- all of these affect how a technique behaves, and variation in results is part of what the experimentation is exploring. The goal of Evo is not to produce a cookbook but to document a process of inquiry that others can learn from and build on.