Auto-rule configuration

Both Auto-rules share a few configuration pieces (Enabled/Disabled toggle, When to Apply filters, Assigned Property, Re-evaluate Session button). This page is the side-by-side reference for what’s unique to each rule’s settings. For the step-by-step how-tos, see Configure a Mismatch rule and Configure a Non-Verbal rule.

Shared structure

Every Auto-rule panel exposes the same skeleton:

Section	Purpose
Master Enabled / Disabled toggle	Pauses the rule without losing the configuration.
When to Apply	Optional Session-level filters that scope the rule to a subset of Sessions.
Assigned Property	The Property the rule applies to flagged Session items.
Save Changes / Reset	Save staged edits, or revert to the saved configuration.
Re-evaluate Session	Apply the saved rule to existing items in the current Session.

New items run through the active rules as they arrive. Re-evaluation is only needed when you change the rule or change the underlying data (the Scripts, mostly) after items were already evaluated.

Mismatch rule — the Comparison Pipeline

The Mismatch rule compares each Session item’s Script against its transcription. Before that comparison, both texts pass through a normalization pipeline of five toggles.

Toggle	Default	What it does
Convert numbers to words	Off	`25` → `twenty-five` on both sides.
Convert ordinals to words	Off	`1st` → `first` on both sides.
Ignore case	On	`Hello` matches `hello`.
Ignore punctuation	On	`Hello, there!` matches `Hello there`.
Normalize whitespace	On	Multiple spaces, tabs, and newlines collapse to single spaces.

The three “On” defaults are almost always what you want. Flip the number-conversion toggles on when your scripts mix digit and word forms — for example, scripts written it cost $25 but ASR transcribing it as it cost twenty-five dollars.

Synonyms

The Mismatch rule also has a Synonyms section. Each synonym group has a canonical form and one or more variants; before comparison, every variant is replaced with the canonical form on both sides. Use this for filler words, contractions, or studio-specific abbreviations:

Canonical	Variants
`okay`	`ok`, `OK`, `o.k.`
`cannot`	`can't`, `cant`
`going to`	`gonna`

Synonyms apply after the comparison pipeline runs.

Test Your Configuration

A scratchpad in the panel: paste a script line and a transcript, run them through the current settings, and see whether they’d match. Faster than re-running the whole Session every time you tweak a toggle.

Non-Verbal rule — Speech Detection Sensitivity

The Non-Verbal rule runs on the audio (not the transcript) and flags items that don’t contain enough real speech. A single numeric threshold (0.00–1.00) controls how selective the rule is.

Threshold range	Label	Behavior
`0.00`–`0.30`	Very Sensitive	Almost everything passes through to transcription.
`0.31`–`0.50`	Sensitive	Most audio transcribed.
`0.51`–`0.70`	Balanced	The middle. `0.60`–`0.70` is recommended.
`0.71`–`0.90`	Strict	Quieter or partial takes flagged non-verbal.
`0.91`–`1.00`	Most Strict	Only confidently-spoken audio transcribed.

Lower = more audio sent to transcription. Higher = more audio tagged as non-verbal.

Default: 0.70.

Re-evaluation

Re-evaluation is the same in both panels:

Make sure the rule is enabled.
Click Re-evaluate Session at the bottom-left of the panel.
A blue toast appears in the bottom right — Evaluating session items… — while the run is in flight.
A green toast replaces it when done — Evaluation complete! with N added, M removed counts.

Re-evaluation runs in the background. You can close the panel or navigate around; the toast still surfaces the result.