What the polish pass changes, and what it never touches: the 31-case golden corpus

After the speech model returns text, and before Lven types it into the focused app, a single-pass engine — one engine on every platform, held to one contract by a shared golden corpus — applies seven rules in a fixed order.

Hyphen-fragment pruning; prefix-stutter pruning; n-gram repetition collapse for n = 1, 2, 3; filler purge (uh, um, eh, er, ah, uh-huh); contraction, proper-noun and phonetic-spelling restoration from a fixed table; sentence casing; terminal punctuation, where the last sentence decides ? versus period.

It is not a language model. It cannot rephrase, restructure, change tense, match tone to an app, or decide what you meant. It has no network access. Given the same transcript it produces the same text every time, on every platform.

The corpus is the polisher's own golden regression test — the file CI runs on every commit, so every platform's implementation is held to one contract. We ran it against the shipped build on 2026-08-15, when it held these 31 cases; it has grown since, and every pair shown is its current answer.

Word counts and change classes were then computed over the 31 pairs with a token diff: case- and apostrophe-insensitive alignment, then classification of each difference.

Every word that survives is a word the speaker said. The only substitution in the corpus is a spelling restoration from a fixed table; nothing is paraphrased and nothing is added.

Note what is preserved: 'you know like you know' in the first case stays, because it is not in the filler list and the engine does not judge whether a phrase is worth keeping. 'had had' and 'that that' stay, because the collapse pass guards legitimate doubles.

A generative cleanup layer optimizes for text that reads well, so every sentence has to be re-read for meaning: the layer is allowed to decide what you meant.

In July 2026 one cloud dictation vendor traced accuracy complaints to its cleanup default having changed words users had not asked it to touch.

A deterministic pass optimizes for the text being what you said: every word that survives is yours. This corpus is what that choice looks like in practice.

Every pair in the corpus is on this page, and the inspector on the Lven Instant page runs the same token diff in your browser, so the counts in the results table are recomputed from the pairs. Any of these inputs, given to the shipped build, produces exactly the output shown.