[juliusmarminge]

Local speech correction strips symbols from dictionary terms to build fuzzy-match keys, so C# (and C++) became the one-letter key "c", and every standalone "c" equalled that key and was rewritten. The review item was correct: the previous head dropped any key under 4 characters that lost a symbol, which also stopped A.I., I/O, Dr. and AT&T from correcting their spoken forms, a regression from main. This rework narrows the condition to key.length < 2 && lostSymbols, so only terms that collapse to a single letter are left out of fuzzy matching. Typing the term itself ("C#") is still handled by the unchanged exact-spelling pass, and cloud transcription is unaffected. The only regex involved is the existing /\s+/gu over dictionary terms (50 characters at most each), which runs in linear time.

Fixes #35686

Reproduction on main

On origin/main cc97303 I added a { term: "C#", aliases: [] } row and the sentence "vitamin c and plan c." to the existing dictionary test in apps/server/src/speech/SpeechService.test.ts. The test passes text through SpeechService.transcribe and startStream().finish; the model returns that text unchanged and has no vocabulary support, so local correction runs. Running vp test run src/speech/SpeechService.test.ts -t "honors exact aliases" returned "... vitamin C# and plan C#." where "... vitamin c and plan c." was expected. I also loaded customWords.ts directly with node --experimental-strip-types, as the reporter's script does. makeSpeechTextCorrector(['C#'], [{term:'C#',aliases:[]}]) turned "vitamin c and plan c" into "vitamin C# and plan C#" and "I write c sharp" into "I write C# sharp". applySpeechAliases, the cloud path, left both unchanged. I did not run a full audio transcription with a downloaded local model; the defect is in deterministic text cleanup after the model.

Evidence

In the real web app, select local Parakeet transcription, add C# without aliases to the Voice dictionary, and use Test model with the same synthetic recording: “vitamin c and plan c. I write c sharp.” Finish recording and inspect the transcript. Both runs used fresh state, Chromium, a 1280×800 viewport, and light theme. Recognition was real; only microphone audio was synthetic.

scenarios:prove exited 0: the scenario failed on base and passed on head.

	Before	After
Commit	cc97303381	201fa1ae96
Local transcript
Recording	Before recording	After recording

Scenario driver · Audio generator · Identical input audio

Set VOICE_EVIDENCE_AUDIO to the generated WAV when running the scenario. The recognizer varied in the final phrase, but the unrelated C letters gained # only on base.

Not shown:

• Non-streamed batch transcription, cloud transcription, and other dictionary terms were not exercised; this proof covers real local streamed dictation with C#.

Tests

vp test run src/speech/ in apps/server: 9 files, 82 tests passed, including the repro test from factory/repro-35686. A direct node run of makeSpeechTextCorrector([term], [{term, aliases: []}]): with C#, "vitamin c and plan c" and "I write c sharp" stay unchanged and "C#" is kept; with C++, "plan c" stays unchanged. A.I. turns "a i and ai" into "A.I. and A.I.", I/O turns "the i o layer and io" into "the I/O layer and I/O", Dr. turns "dr smith" into "Dr. smith", AT&T turns "at t" into "AT&T", and R&D turns "r and d" into "R&D", all the same as main. The pre-commit hook passed vp fmt and vp lint.

Not in this change

• Pre-existing typecheck errors in apps/server/src/workspace/WorkspaceSearchIndex.ts and WorkspaceSearchIndex.test.ts on origin/factory/repro-35686 (base cc97303).
• Existing over-matches on main that the review left for a follow-up: "us" becoming "U.S." (run by the reviewer) and ".NET" matching "net" (expected by the same logic, not run here).
• Existing typecheck errors in apps/server/src/workspace/WorkspaceSearchIndex.ts and its test, which this branch does not touch.

Who did the work

• Written by Claude Opus 5.5, in Claude Code
• Reviewed by GPT-6.1 Sol, in Codex, and by Claude Opus 5.5, each on its own
• Proof captured by GPT-6.1 Sol and checked by Claude Opus 5.5

Run by the t3code factory in T3 Code.

[macroscopeapp]

Approvability

Verdict: Approved at 201fa1a

Macroscope's review found this PR approvable — This narrows speech vocabulary matching so terms such as C# no longer rewrite standalone letters. Explicit dictionary aliases remain unchanged, and the adjustment is confined to transcript correction.

^{You can add or adjust custom eligibility rules. Learn more.}

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	5.0 KiB	5.0 KiB	0 B (0.0%)	6.8 KiB	✅
Codex	Thread snapshot wire	3.8 KiB	3.8 KiB	0 B (0.0%)	4.9 KiB	✅
Codex	Live turn WebSocket wire	1.2 KiB	1.2 KiB	0 B (0.0%)	2.0 KiB	✅
Codex	Live turn WebSocket decoded	21.5 KiB	21.5 KiB	0 B (0.0%)	24.4 KiB	✅
Codex	Live turn messages	1	1	0 (0.0%)	8	✅
Claude	Total thread wire	5.0 KiB	5.0 KiB	0 B (0.0%)	6.8 KiB	✅
Claude	Thread snapshot wire	3.8 KiB	3.8 KiB	0 B (0.0%)	4.9 KiB	✅
Claude	Live turn WebSocket wire	1.2 KiB	1.2 KiB	0 B (0.0%)	2.0 KiB	✅
Claude	Live turn WebSocket decoded	21.9 KiB	21.9 KiB	0 B (0.0%)	24.4 KiB	✅
Claude	Live turn messages	1	1	0 (0.0%)	8	✅

Other scenarios

All 64 scenario measurements match main.

All 64 scenario measurements

Provider	Scenario	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	short turn resume	WebSocket decoded	10.5 KiB	10.5 KiB	0 B (0.0%)	12.7 KiB	✅
Codex	short turn resume	WebSocket messages	2	2	0 (0.0%)	2	✅
Codex	long turn running, cold open	HTTP decoded	281.7 KiB	281.7 KiB	0 B (0.0%)	302.7 KiB	✅
Codex	long turn running, cold open	rows	75	75	0 (0.0%)	75	✅
Codex	long turn running, history page	HTTP decoded	92.4 KiB	92.4 KiB	0 B (0.0%)	107.4 KiB	✅
Codex	long turn running, history page	rows	75	75	0 (0.0%)	75	✅
Codex	long turn finished, cold open	HTTP decoded	381.5 KiB	381.5 KiB	0 B (0.0%)	410.2 KiB	✅
Codex	long turn finished, cold open	rows	151	151	0 (0.0%)	151	✅
Codex	long turn finished, socket cold open	WebSocket decoded	381.6 KiB	381.6 KiB	0 B (0.0%)	410.2 KiB	✅
Codex	long turn finished, socket cold open	WebSocket messages	2	2	0 (0.0%)	2	✅
Codex	long turn finished, socket cold open	rows	151	151	0 (0.0%)	151	✅
Codex	long turn finished, work page	HTTP decoded	186.2 KiB	186.2 KiB	0 B (0.0%)	200.2 KiB	✅
Codex	long turn finished, work page	rows	150	150	0 (0.0%)	150	✅
Codex	idle thread resume	WebSocket decoded	65 B	65 B	0 B (0.0%)	200 B	✅
Codex	idle thread resume	WebSocket messages	1	1	0 (0.0%)	1	✅
Codex	fleet shell snapshot	HTTP decoded	42.4 KiB	42.4 KiB	0 B (0.0%)	53.7 KiB	✅
Codex	fleet shell snapshot	rows	27	27	0 (0.0%)	27	✅
Codex	fleet shell subscribe	WebSocket decoded	65 B	65 B	0 B (0.0%)	200 B	✅
Codex	fleet shell subscribe	WebSocket messages	1	1	0 (0.0%)	1	✅
Codex	fleet shell resume check	HTTP decoded	18 B	18 B	0 B (0.0%)	100 B	✅
Codex	fleet shell resume check	rows	0	0	0	0	✅
Codex	stale shell reload	HTTP decoded	42.4 KiB	42.4 KiB	0 B (0.0%)	53.7 KiB	✅
Codex	stale shell reload	rows	27	27	0 (0.0%)	27	✅
Codex	stale shell resume	WebSocket decoded	140 B	140 B	0 B (0.0%)	400 B	✅
Codex	stale shell resume	WebSocket messages	2	2	0 (0.0%)	2	✅
Codex	stale todo list cold open	HTTP decoded	72.4 KiB	72.4 KiB	0 B (0.0%)	83.0 KiB	✅
Codex	stale todo list cold open	rows	75	75	0 (0.0%)	75	✅
Codex	stale todo list cold open	nodes	2	2	0 (0.0%)	4	✅
Codex	stale todo list cold open	subagents	2	2	0 (0.0%)	4	✅
Codex	stale todo list cold open	plans	1	1	0 (0.0%)	2	✅
Codex	stale todo list cold open	providerTurns	1	1	0 (0.0%)	2	✅
Codex	stale todo list cold open	checkpoints	1	1	0 (0.0%)	2	✅
Claude	short turn resume	WebSocket decoded	10.6 KiB	10.6 KiB	0 B (0.0%)	12.7 KiB	✅
Claude	short turn resume	WebSocket messages	2	2	0 (0.0%)	2	✅
Claude	long turn running, cold open	HTTP decoded	283.3 KiB	283.3 KiB	0 B (0.0%)	302.7 KiB	✅
Claude	long turn running, cold open	rows	75	75	0 (0.0%)	75	✅
Claude	long turn running, history page	HTTP decoded	93.4 KiB	93.4 KiB	0 B (0.0%)	107.4 KiB	✅
Claude	long turn running, history page	rows	75	75	0 (0.0%)	75	✅
Claude	long turn finished, cold open	HTTP decoded	384.2 KiB	384.2 KiB	0 B (0.0%)	410.2 KiB	✅
Claude	long turn finished, cold open	rows	151	151	0 (0.0%)	151	✅
Claude	long turn finished, socket cold open	WebSocket decoded	384.3 KiB	384.3 KiB	0 B (0.0%)	410.2 KiB	✅
Claude	long turn finished, socket cold open	WebSocket messages	2	2	0 (0.0%)	2	✅
Claude	long turn finished, socket cold open	rows	151	151	0 (0.0%)	151	✅
Claude	long turn finished, work page	HTTP decoded	188.1 KiB	188.1 KiB	0 B (0.0%)	200.2 KiB	✅
Claude	long turn finished, work page	rows	150	150	0 (0.0%)	150	✅
Claude	idle thread resume	WebSocket decoded	66 B	66 B	0 B (0.0%)	200 B	✅
Claude	idle thread resume	WebSocket messages	1	1	0 (0.0%)	1	✅
Claude	fleet shell snapshot	HTTP decoded	42.9 KiB	42.9 KiB	0 B (0.0%)	53.7 KiB	✅
Claude	fleet shell snapshot	rows	27	27	0 (0.0%)	27	✅
Claude	fleet shell subscribe	WebSocket decoded	66 B	66 B	0 B (0.0%)	200 B	✅
Claude	fleet shell subscribe	WebSocket messages	1	1	0 (0.0%)	1	✅
Claude	fleet shell resume check	HTTP decoded	18 B	18 B	0 B (0.0%)	100 B	✅
Claude	fleet shell resume check	rows	0	0	0	0	✅
Claude	stale shell reload	HTTP decoded	42.9 KiB	42.9 KiB	0 B (0.0%)	53.7 KiB	✅
Claude	stale shell reload	rows	27	27	0 (0.0%)	27	✅
Claude	stale shell resume	WebSocket decoded	140 B	140 B	0 B (0.0%)	400 B	✅
Claude	stale shell resume	WebSocket messages	2	2	0 (0.0%)	2	✅
Claude	stale todo list cold open	HTTP decoded	73.4 KiB	73.4 KiB	0 B (0.0%)	83.0 KiB	✅
Claude	stale todo list cold open	rows	75	75	0 (0.0%)	75	✅
Claude	stale todo list cold open	nodes	2	2	0 (0.0%)	4	✅
Claude	stale todo list cold open	subagents	2	2	0 (0.0%)	4	✅
Claude	stale todo list cold open	plans	1	1	0 (0.0%)	2	✅
Claude	stale todo list cold open	providerTurns	1	1	0 (0.0%)	2	✅
Claude	stale todo list cold open	checkpoints	1	1	0 (0.0%)	2	✅

Baseline: 7d51718 · PR result: 201fa1a · Source CI: success

Scenario and decoded snapshot size

10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.

• Codex decoded thread snapshot: 108.5 KiB
• Claude decoded thread snapshot: 108.8 KiB

Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed.

[coderabbitai]

No actionable comments were generated in the recent review.

Walkthrough

The speech corrector now filters certain short keys produced by symbol removal. A regression test checks that standalone “c” text remains unchanged when “C#” is a custom word.

Changes: Speech dictionary matching

Layer / File(s)	Summary
Filter short custom-word keys apps/server/src/speech/customWords.ts, apps/server/src/speech/SpeechService.test.ts	customWordKeys processes original terms and ampersand-expanded forms, and omits certain keys shorter than two characters. The test checks that “vitamin c and plan c” stays unchanged with “C#” in the custom-word options.

Estimated code review effort: 2 (Simple) | ~10 minutes

Change: Bug fix · Severity of issue fixed: Low

Suggested labels: bugfix

Merge Risk: Minimal · up to 201fa

The change stops dictionary terms such as C# from rewriting unrelated standalone letter c in local transcripts. It adds a regression test for that case. No merge-blocking risk is evident.

Pre-merge checks: 4

Check name	Status	Explanation
Title check		The title names the main change: a C# dictionary term no longer rewrites standalone letters. It uses the conventional commit format and matches the linked issue and the PR objectives. The title is spe…
Linked Issues check		Issue #35686 requires that adding C# to the Voice dictionary must not rewrite unrelated standalone letters in local transcripts. The diff to customWordKeys in apps/server/src/speech/customWords.ts…
Out of Scope Changes check		The whole-PR diff against the merge base cc97303 contains two changed files. apps/server/src/speech/customWords.ts contains the fuzzy-key fix for #35686. apps/server/src/speech/SpeechService.test…
Description check		The description is thorough and follows the template's intent. It gives a clear problem statement (C# and C++ turn standalone "c" into the one-letter key, rewriting unrelated letters), a change descri…

Full details: Title check

Explanation

The title names the main change: a C# dictionary term no longer rewrites standalone letters. It uses the conventional commit format and matches the linked issue and the PR objectives. The title is specific and clear, and it does not use vague terms.

Full details: Linked Issues check

Explanation

Issue #35686 requires that adding C# to the Voice dictionary must not rewrite unrelated standalone letters in local transcripts. The diff to customWordKeys in apps/server/src/speech/customWords.ts drops a fuzzy-match key when it has fewer than two characters and symbol removal caused the reduction. For C# and C++, the key c is dropped, so vitamin c and plan c stays unchanged. The diff keeps keys of two or more characters, so A.I., I/O, Dr., and AT&T keep their local correction. A single-letter term without symbols (for example C) is kept because lostSymbols is false. The test change in apps/server/src/speech/SpeechService.test.ts adds C# with the spoken text vitamin c and plan c, and it expects those words to remain unchanged. The cloud alias-only path is not changed. The PR author reports a passing vp test run src/speech/ run and direct corrector checks. I did not run these checks. The source diff supports the fix.

Full details: Out of Scope Changes check

Explanation

The whole-PR diff against the merge base cc97303 contains two changed files. apps/server/src/speech/customWords.ts contains the fuzzy-key fix for #35686. apps/server/src/speech/SpeechService.test.ts contains the regression test for the same defect. Both changes implement the linked issue. The PR description lists typecheck errors in WorkspaceSearchIndex files and existing over-matches (us, .NET) as pre-existing and not changed here. The reviewed diff shows no changes to those files, so they do not count as out-of-scope changes.

Full details: Description check

Explanation

The description is thorough and follows the template's intent. It gives a clear problem statement (C# and C++ turn standalone "c" into the one-letter key, rewriting unrelated letters), a change description (narrowing the skip condition to key.length < 2 && lostSymbols), and a scope tie to the linked issue #35686. Verification is detailed: focused test runs, direct corrector checks, regression checks for A.I., I/O, Dr., AT&T and R&D, and a real-app before/after scenario with evidence. It also states what was not checked. The template's title guidance is not shown in the description, but the scope and verification sections are substantive. Minor gap: the Scope section does not restate the bug-fix framing in the template's terms, and the agent attribution is present at the end as the template requests.

Finishing Touches

Generate unit tests (beta)

• Commit to this branch
• Create a new PR

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

Recent review info

Run configuration

• Source mode: Auto
• Repository: pingdotgg/t3code
• Branch: factory/issue-35686
• File: .coderabbit.config.ts
• Commit: 201fa1a
• Layers: Repository TS config
• Review profile: CHILL
• Plan: Team
• Run ID: b70362d9-017d-4de5-b757-2b55206fe1d9

Commits

Reviewing files that changed from the base of the PR and between 7d51718 and 201fa1a.

Files selected for processing: 2

• apps/server/src/speech/SpeechService.test.ts
• apps/server/src/speech/customWords.ts

Limit details: You’ve used all 10 included reviews currently available.