The proof is in the groove.

I don’t process a master because analysis says it looks difficult. I test it physically, and only change the audio when the cut proves that it needs changing.

Digital analysis of a side can produce numbers that look alarming. A peak in the excursion model. A velocity event flagged in the high-frequency range. A warning attached to a particular passage. These are real measurements of real signal behaviour, and they are worth taking seriously.

But they describe the audio file. They do not describe what happens when that audio is cut into physical plastic at a specific level, with a specific groove pitch, on a specific cutter head.

The gap between those two things is where most of the actual engineering work happens.

What analysis can tell you — and what it cannot

Analysis is a detection tool. It helps identify passages in a side that are likely to be worth testing physically. A section with strong low-frequency excursion, or high-frequency velocity, or complex stereo geometry, is a section worth understanding before committing to a full cut.

What analysis cannot tell you is whether those passages will present a real problem at the chosen cutting conditions. Whether the groove will be clean. Whether the cut will track. Whether the physical record is actually a problem, or simply a measurement that looks large on a screen.

Those answers only come from the physical record itself.

This distinction matters for how processing decisions get made. A software flag is a reason to test. It is not an instruction to treat.

Isolating the problem rather than treating the side

When a passage is flagged — by analysis, or by a first test cut — the response is to isolate it. Not to apply a broad reduction to the whole side because one section looks difficult, but to identify the exact point where the difficulty exists and test that point directly.

If earlier tracks on a side pass cleanly at the desired cutting level, that level has been physically validated for those tracks. Applying a conservative reduction across the whole side in response to one later problem passage would mean cutting the tracks that are fine at a level they do not need.

The surgical approach is to build a physical test specifically around the passage in question. Cut it at the target level. Inspect the groove. Play back the cut. Assess whether the problem is real and whether it is audible. If it is, understand what is causing it and deal with that specific cause — not the rest of the side.

This takes more work than applying a global treatment and moving on. It also produces a better record.

A real example from development

A side in development contained multiple tracks. Earlier material passed cleanly at the target cutting level. Later in the side, a track containing spoken-word vocals created an audible physical problem — a specific passage where the vocal character was generating excessive stylus velocity in a way that showed up on playback.

The response was not to reduce the level of the whole side. It was to isolate that vocal passage and build a test around it — a physical test sequence covering just that section, at different treatments, assessed by groove inspection and playback. The earlier tracks that were already proven were left at their validated level. The solution was applied only to the passage that demonstrated a physical problem.

The result was a side cut at the correct level for the music, with a targeted fix applied only where the physical evidence required it.

That is the difference between solving an actual problem and solving a theoretical one.

The same logic applies to cutting level

Cutting level decisions follow the same principle. It is not desirable to cut unnecessarily quietly. A conservative level is not automatically a safe level — it is a level that may be leaving available quality behind.

If analysis and initial testing suggest there is room for a higher cutting level, that level is physically tested. If the higher level exposes a problem in one passage, the response is to understand and address that passage — not to immediately reduce the whole side back to a conservative starting point.

The aim is to find the highest level at which the entire side passes physically. That is not the same as the highest level the analysis suggests might be possible, and it is not the same as a level chosen to minimise risk without testing. It is the level established by cutting and listening to the actual record.

What the physical record proves

A clean cut that tracks well is a result. It represents the programme, the cutting conditions and the physical medium working together. That result belongs to the exact context in which it was produced — the specific level, pitch, preparation and cutter state on that session — but it is real evidence, not a prediction.

An untreated passage that passes physically is a valuable result. It means the material did not need treatment, and treatment would have changed the source without improving the record.

A passage that fails physically is also a valuable result. It identifies exactly where the actual problem is, which makes it possible to address that problem precisely rather than treating everything around it as a precaution.

Software does not produce these results. The physical record does.

Why this matters for the complete side

All of this applies at the level of individual passages and individual test cuts. But the final validation is always the complete side, played back in full.

A passage that passes in isolation may behave differently as part of a continuous groove. Inner groove conditions change through a side. Thermal state of the cutter changes. The relationship between adjacent groove turns changes. A section that cuts cleanly in a targeted test still has to perform within the full side before the order proceeds.

Surgical testing establishes what a passage needs. The final full-side cut and playback establishes whether the whole record works.

What this means when you send music to Diz Lathe Cuts

When your audio arrives, it is assessed: levels, frequency content, stereo geometry, programme length, format. Analysis identifies passages worth paying attention to. That assessment is preparation for testing, not a verdict on the master.

If something looks difficult, it gets tested physically. If it passes, it is left alone. If the test reveals a real problem, the response is as specific as the problem — not a precautionary reduction applied to the whole side.

If the cutting level can be higher than a conservative starting point, that will be established by testing and not assumed. If one passage limits what the rest of the side can achieve, that passage will be understood and addressed rather than used as a reason to pull the whole record back.

The goal is the best possible cut of the master you supplied. Not the safest cut. Not the most processed cut. The cut that is physically validated to be as good as the programme allows.