feat: OpenWhispr-inspired transcription polish pass

Major quality pass on top of Phase 2. Five substantive changes plus
cross-cutting touches across audio, hotkey, transcription, and Tauri
command layers.

  Transcription quality

  - Long-audio chunking in commands/transcription.rs: Parakeet and large
    file transcription now chunk-and-recompose with overlap trimming, so
    the live-path chunking advantage extends to file-based workflows.
  - Stateful live speech gate in commands/live.rs on top of the earlier
    duplicate-boundary filtering — distinguishes start-of-speech from
    mid-speech and holds state across chunks.

  Auto-learning corrections

  - New crates/ai-formatting/src/correction_learning.rs: extracts user
    text corrections from viewer edits and proposes additions to the
    active profile's vocabulary.
  - src-tauri/src/commands/profiles.rs bridge for frontend-driven
    confirmation of learned terms.
  - src/routes/viewer/+page.svelte hooks the learning path into the
    segment-edit flow so corrections feed profile_terms without a
    separate 'train this profile' UX.

  Transcript profile provenance

  - Migration v8 (crates/storage/src/migrations.rs) adds profile_id to
    transcripts, defaulting to DEFAULT_PROFILE_ID so existing rows stay
    valid.
  - crates/storage/src/database.rs: TranscriptRow + CRUD carry profile_id.
  - src-tauri/src/commands/transcripts.rs: add_transcript accepts and
    persists profile_id.
  - DictationPage.svelte + FilesPage.svelte send activeProfileId on
    capture so learned corrections are attributed to the right profile.

  Cleanup prompt contract

  - crates/ai-formatting/src/llm_client.rs hardened: the CLEANUP_PROMPT
    now specifies concrete do/do-not rules, ready for a real model-backed
    cleanup pass. The llm_client is still a stub — kon-llm remains unwired
    — but the prompt shape is final.

  Cross-cutting polish

  - Minor touches in audio (capture/decode/resample), hotkey (lib/linux/stub),
    core, transcription (concurrency/model_manager/local_engine/whisper_rs),
    and the rest of src-tauri/src/commands/*: error-path tightening, log
    clarity, TS-migration follow-ups (@ts-nocheck additions for incremental
    typing).

Verified locally: npm run check, cargo test -p kon-ai-formatting,
cargo test -p kon-storage, cargo test -p kon --lib commands::live::tests,
cargo check — all green.

Scope boundary: kon-llm crate is still a stub; task extraction remains
rule-based. Bundled local-LLM runtime is the next clean step and is not
in this commit.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

src/lib/pages/DictationPage.svelte

@@ -421,10 +421,13 @@
         id: historyId,
         date: new Date().toLocaleString("en-GB"),
         source: "live",
+        profileId: profilesStore.activeProfileId,
         preview: transcript.slice(0, 120),
         text: transcript,
         segments: segments,
         duration: (Date.now() - startTime) / 1000,
+        engine: settings.engine,
+        modelId: selectedModelId(),
         language: effectiveLanguage(),
         template: activeTemplate || undefined,
         audioPath,
@@ -537,6 +540,7 @@
       id: crypto.randomUUID(),
       date: new Date().toLocaleString("en-GB"),
       source: "typed",
+      profileId: profilesStore.activeProfileId,
       preview: transcript.slice(0, 120),
       text: transcript,
       segments: [],

src/lib/pages/FilesPage.svelte

@@ -75,6 +75,7 @@
       try {
         const result = await invoke("transcribe_file", {
           path,
+          engine: settings.engine,
           language: settings.language,
           initialPrompt: "",
           profileId: profilesStore.activeProfileId,
@@ -98,11 +99,14 @@
           id: crypto.randomUUID(),
           date: new Date().toLocaleString("en-GB"),
           source: fileName,
+          profileId: profilesStore.activeProfileId,
           preview: text.slice(0, 120),
           text,
           segments: result.segments,
           duration: result.duration,
           language: result.language,
+          engine: result.engine ?? settings.engine,
+          modelId: result.modelId ?? null,
         });
       } catch (err) {
         error = `Failed: ${fileName} — ${typeof err === "string" ? err : err.message}`;

src/lib/stores/page.svelte.ts

@@ -114,6 +114,7 @@ function mapTranscriptRow(row: TranscriptDto): TranscriptEntry {
     id: row.id,
     text,
     source: row.source ?? "",
+    profileId: row.profileId ?? "00000000-0000-0000-0000-000000000001",
     title: row.title ?? "",
     audioPath: row.audioPath ?? null,
     duration: Number(row.duration ?? 0),
@@ -136,6 +137,7 @@ function normaliseTranscriptEntry(entry: TranscriptWriteEntry): TranscriptEntry
     id: String(entry.id),
     text: entry.text ?? "",
     source: entry.source ?? "microphone",
+    profileId: entry.profileId ?? "00000000-0000-0000-0000-000000000001",
     title: entry.title ?? "",
     audioPath: entry.audioPath ?? null,
     duration: Number(entry.duration ?? 0),
@@ -187,6 +189,7 @@ export async function addToHistory(entry: TranscriptWriteEntry) {
         id: normalised.id,
         text: normalised.text,
         source: normalised.source,
+        profileId: normalised.profileId,
         title: normalised.title || null,
         audioPath: normalised.audioPath ?? null,
         duration: normalised.duration,

src/lib/types/app.ts

@@ -78,6 +78,7 @@ export interface TranscriptDto {
   id: string;
   text: string;
   source: string;
+  profileId: string;
   title: string | null;
   audioPath: string | null;
   duration: number;
@@ -95,6 +96,7 @@ export interface TranscriptEntry {
   id: string;
   text: string;
   source: string;
+  profileId: string;
   title: string;
   audioPath: string | null;
   duration: number;

src/routes/viewer/+page.svelte

@@ -8,6 +8,8 @@
   import { errorMessage } from "$lib/utils/errors.js";
   import { parseStoredJson } from "$lib/utils/storage.js";
   import { saveTranscriptMeta } from "$lib/stores/page.svelte.js";
+  import { toasts } from "$lib/stores/toasts.svelte.ts";
+  import { DEFAULT_PROFILE_ID } from "$lib/stores/profiles.svelte.ts";
 
   let item = $state<TranscriptEntry | null>(null);
   let audioEl = $state<HTMLAudioElement | null>(null);
@@ -28,6 +30,7 @@
   let textDirty = $state(false);
   let textSaveTimer: ReturnType<typeof setTimeout> | null = null;
   let editingTextareaEl = $state<HTMLTextAreaElement | null>(null);
+  let textLearnBase = $state("");
 
   function stopAudio() {
     if (audioEl) audioEl.pause();
@@ -59,6 +62,7 @@
   function loadViewerItem(nextItem: TranscriptEntry | null) {
     item = nextItem;
     textDraft = nextItem?.text ?? "";
+    textLearnBase = nextItem?.text ?? "";
     activeSegmentIdx = -1;
     editingIdx = -1;
     editingText = "";
@@ -83,9 +87,36 @@
       clearTimeout(textSaveTimer);
       textSaveTimer = null;
     }
-    if (textDirty) commitTextEdit();
+    if (viewerMode === "edit") {
+      commitTextEdit(true);
+    } else if (textDirty) {
+      commitTextEdit();
+    }
   });
 
+  async function maybeLearnCorrections(originalText: string, editedText: string) {
+    if (!item || !originalText.trim() || !editedText.trim() || originalText === editedText) {
+      return;
+    }
+
+    try {
+      const learned = await invoke<Array<{ term: string }>>("learn_profile_terms_from_edit_cmd", {
+        profileId: item.profileId || DEFAULT_PROFILE_ID,
+        originalText,
+        editedText,
+      });
+      if (learned.length > 0) {
+        const count = learned.length;
+        toasts.success(
+          count === 1 ? "Learned 1 profile term" : `Learned ${count} profile terms`,
+          learned.map((entry) => entry.term).join(", "),
+        );
+      }
+    } catch (err) {
+      console.warn("viewer maybeLearnCorrections failed", errorMessage(err));
+    }
+  }
+
   function handleStorageChange(e: StorageEvent) {
     if (e.key === "kon_viewer_item" && e.newValue) {
       loadViewerItem(parseStoredJson<TranscriptEntry>(e.newValue));
@@ -193,14 +224,18 @@
 
   function finishEditing() {
     if (editingIdx >= 0 && item?.segments) {
+      const previousText = item.text;
       item.segments[editingIdx].text = editingText.trim();
       // Update full text
       item.text = item.segments.map((s) => s.text).join(" ").trim();
+      textDraft = item.text;
       saveItemToHistory();
       // Task 2.5 — persist segment boundaries (text + starred flags) to
       // SQLite via segments_json. update_transcript above only covers text
       // and title; the segment array needs the dedicated meta command.
       persistSegments();
+      void maybeLearnCorrections(previousText, item.text);
+      textLearnBase = item.text;
     }
     editingIdx = -1;
     editingText = "";
@@ -281,18 +316,24 @@
     }, 400);
   }
 
-  function commitTextEdit() {
+  function commitTextEdit(learnCorrections = false) {
     if (!item) return;
     const next = textDraft;
+    if (next !== item.text) {
+      item.text = next;
+      // The compact preview in History is derived from `preview`; keep it
+      // roughly in sync so the list does not show stale copy after an edit.
+      item.preview = next.slice(0, 120);
+      saveItemToHistory();
+    }
+    if (learnCorrections && next !== textLearnBase) {
+      void maybeLearnCorrections(textLearnBase, next);
+      textLearnBase = next;
+    }
     if (next === item.text) {
       textDirty = false;
       return;
     }
-    item.text = next;
-    // The compact preview in History is derived from `preview`; keep it
-    // roughly in sync so the list does not show stale copy after an edit.
-    item.preview = next.slice(0, 120);
-    saveItemToHistory();
     textDirty = false;
   }
 
@@ -441,6 +482,7 @@
                  text-text leading-relaxed resize-none focus:outline-none focus:border-accent"
           bind:value={textDraft}
           oninput={scheduleTextSave}
+          onblur={() => commitTextEdit(true)}
           data-no-transition
           placeholder="Edit the transcript..."
         ></textarea>