Skip to content

Issue categorisation: how Sonalix groups mix problems

Send feedback

Last updated 3 min readTags:AnalysisAI Coaching

How Sonalix groups mix findings into seven issue categories and a general fallback, what these categories affect, and where keyword matching can go wrong.

Issue categorisation is how Sonalix assigns a type to a mix finding, such as low end, dynamics or stereo. It matches words in the finding’s title and description to a predefined category.

The category helps select the wording for the overall verdict and next step. It describes the kind of issue identified; it does not measure how serious that issue is.

What the categories mean

Sonalix uses seven specific categories and a general fallback. The labels are internal. You see the resulting guidance in your report.

Category Internal label What it covers
Clipping clipping Findings about clipping, true peak limits or distortion.
Low end low_end Bass balance, sub-bass, rumble and kick–bass overlap.
Dynamics dynamics Compression, loudness variation, transients and punch.
High end high_end Brightness, dullness, harshness and sibilance.
Midrange mid_range Mud, boxiness, nasal tone and midrange congestion.
Stereo stereo Width, imaging, phase relationships and mono compatibility.
Loudness loudness Overall level and loudness targets, including LUFS.
General general Findings whose wording does not match a specific category.

These are broad groups. A high-end finding might describe excessive brightness or a lack of it. The category alone does not tell you whether to boost or cut a frequency range.

How Sonalix assigns a category

The classifier combines the finding’s title and description, converts the text to lowercase, then checks it against keyword patterns in a fixed order. The first matching pattern determines the category. If none matches, it assigns general.

For example, words such as “compression” and “transient” can indicate dynamics. Terms such as “phase” and “mono compatibility” can indicate stereo. The result depends on the full text and which pattern matches first.

This step does not analyse the audio again. It classifies an existing finding. With the same text and rules, it returns the same category; a new analysis may produce different wording.

How categories affect your report

The category informs the templates used for the overall verdict and next-step guidance. A clipping finding can prompt guidance about peak levels; a stereo finding can prompt guidance about phase or width.

Categorisation is separate from severity and priority:

  • Category: what kind of issue the finding describes.
  • Severity: how serious a flagged condition is.
  • Priority: where a suggested fix appears in the order of work.

A stereo finding is not automatically more urgent than a loudness finding. Read the specific diagnosis and recommendation to understand what needs attention.

Where keyword matching can go wrong

Keyword matching relies on wording. It cannot, by itself, establish whether the diagnosis is correct.

Overlapping descriptions. A finding about “bass compression reducing punch” contains language associated with both low end and dynamics. The first matching pattern wins. Multiple possible matches do not automatically produce the general fallback.

Negation and context. A phrase such as “no clipping detected” still contains “clipping”. Rules need to account for context if they are to distinguish a problem from a statement that the problem is absent.

Unfamiliar wording. A finding may describe a relevant problem without using any recognised terms. It can then receive the general category. This means no specific pattern matched; it does not mean the mix has no problems.

How to use the guidance

Read the finding and its suggested adjustment together. Check the relevant measurement, then listen for the problem described before changing your mix.

If the verdict or next step seems unrelated to the detailed finding, the category match may be wrong. The diagnosis may also need review. Treat the category as a way to organise the report, rather than proof that a particular adjustment is required.

Was this page helpful?

    Type to search titles, headings and summaries.