Let word mode combine with the VOX preset

CURRENT.md

@@ -40,6 +40,12 @@ mobile draft at 390px.
   - [x] C11 preview caching: Discord-hosted image/GIF previews (WebP 320px, hash-addressed, 5 MiB
     cap, one attempt) run after every walk in both modes. Live run `...-20260916090722`:
     `previewed: 2`; thumbs served over `/api/media/<scrape>/<item>/thumb` (200).
+  - [x] C65 Word mode with the Half-Life preset. Word mode no longer yields to VOX prosody: with
+    a preset it synthesizes each word with the preset profile (per-word `\!R`/`\!%` tags), and the
+    VOX parser is skipped because its tags cannot survive a word split. The daemon always exports
+    `VESKSLOP_TTS_VOX` (0/1) so an explicit off is not overridden by the preset implication, and
+    the TTS panel allows preset + VOX + word mode together (word mode wins for delivery). Wrapper
+    tests cover preset+word mode with tags, the implied-VOX case, and the profile.
   - [x] C64 Correct preset application. The design's bug-fix section landed in the wrapper:
     Flextalk rate/pitch now ride as sticky `\!R` / `\!%` vendor tags (attributes only for MSTTS),
     neutral sliders emit neither, word mode tags every word, and presets write the audible profile

docs/voice.md

@@ -158,8 +158,10 @@ from the TTS panel:
   default 0), and word mode (`tts.words`) with a gap in ms (`tts.wordGapMs`) for the radio/robot
   character. **VOX prosody** (`tts.vox`) routes the line through the NetTTS-style encoder first -
   sentence breaks, lead-ins, weighted pauses, and Flextalk vendor tags - for Black Mesa
-  announcement energy; **clean** (`tts.voxClean`) drops the staccato wrapper. VOX outranks word
+  announcement energy; **clean** (`tts.voxClean`) drops the staccato wrapper. Word mode **combines
-  mode and needs `python3` on the host. Both travel to the command provider as
+  with the presets** and takes precedence over VOX prosody: each word is its own engine instance,
+  so the parser is skipped and the vendor tags go on every word (`\!R` rate, `\!%` pitch). VOX
+  needs `python3` on the host. Both travel to the command provider as
   `VESKSLOP_TTS_VOX` / `VESKSLOP_TTS_VOX_CLEAN` (or the `{vox}` / `{vox_clean}` placeholders).
   A **preset** select (`tts.preset`: factory, VOX, HEV) shifts the Keith Bell speaker profile to
   the Half-Life approximation; `vox` also implies VOX prosody unless VOX is explicitly off, and
@@ -168,8 +170,10 @@ from the TTS panel:
   both registry hives (plus the preset's Gender/Line Mode and a Default Speaker pointer), and for
   Flextalk it sends rate/pitch as the sticky `\!R` / `\!%` vendor tags rather than
   `ITTSAttributes`; attributes only apply to MSTTS voices. Neutral sliders emit neither, so the
-  preset is reproduced exactly. `sh microslopsam/setup/test-wrapper.sh` checks this contract with
+  preset is reproduced exactly. The daemon always exports `VESKSLOP_TTS_VOX` (0 or 1) so an
-  stubbed wine and engine binaries, no Wine install needed. The preset values also show up in the
+  explicit off is honored, and the panel lets a preset, VOX, and word mode be combined.
+  `sh microslopsam/setup/test-wrapper.sh` checks this contract with stubbed wine and engine
+  binaries, no Wine install needed. The preset values also show up in the
   voice metadata defaults, so the panel's "default" snap points match what the engine gets.
 - **Test lines** - Attention, Laboratory, Critical, Health, and Radiation, plus two custom lines
   kept in `localStorage`. Each button speaks the line with the current profile, so a change can be

microslopsam/setup/test-wrapper.sh

@@ -103,6 +103,22 @@ VESKSLOP_TTS_VOICE=keith-bell run_wrapper
 check "plain run writes both hives" "2" "$(grep -c 'Speakers\\{11260610' "$STUB_REG")"
 check "plain run has no preset extras" "0" "$(grep -cE 'Gender|Line Mode|Default Speaker' "$STUB_REG")"
 
+# Word mode combines with a preset and skips the VOX parser, with tags per word.
+VESKSLOP_TTS_VOICE=keith-bell VESKSLOP_TTS_PRESET=vox VESKSLOP_TTS_VOX=1 \
+    VESKSLOP_TTS_WORDS=1 VESKSLOP_TTS_WORD_GAP_MS=0 VESKSLOP_TTS_SPEED=1.25 run_wrapper
+check "word mode with preset synthesizes every word" "3" "$(grep -c '^ARGV:' "$STUB_LOG")"
+check "word mode skips the VOX parser" "0" "$(grep -c 'wH1' "$STUB_LOG")"
+check "word mode tags every word" "3" \
+    "$(sed -n 's/.*\[-t\] \[\(.*\)\] \[-o\].*/\1/p' "$STUB_LOG" | grep -c '^\\!R0.800 ')"
+check "word mode keeps the preset profile" "2" "$(grep -c '"Pitch"=dword:00000003' "$STUB_REG")"
+check "word mode emits no attributes" "0" "$(attr_count)"
+
+# The vox preset implies VOX when the variable is unset, but word mode still wins.
+VESKSLOP_TTS_VOICE=keith-bell VESKSLOP_TTS_PRESET=vox VESKSLOP_TTS_WORDS=1 \
+    VESKSLOP_TTS_WORD_GAP_MS=0 run_wrapper
+check "implied VOX yields to word mode" "3" "$(grep -c '^ARGV:' "$STUB_LOG")"
+check "implied VOX skipped for words" "0" "$(grep -c 'wH1' "$STUB_LOG")"
+
 # Voice metadata follows the active preset.
 check "voices reflect hev" "1" \
     "$(VESKSLOP_TTS_PRESET=hev sh "$WRAP" --voices | tr ',' '\n' | grep -c 'keith-bell|Keith Bell|male|en|expression=59+defaults=pitch:15:rate:24')"

microslopsam/setup/veskslop-tts.sh

@@ -16,7 +16,8 @@
 #                                             prominences, head_size, lowering
 #                   VESKSLOP_TTS_HIGHPASS     high-pass Hz (40-400, default 40)
 #                   VESKSLOP_TTS_ECHO         mono echo amount 0-100 (default 0)
-#                   VESKSLOP_TTS_WORDS        1 = synthesize each word, merge (Flextalk)
+#                   VESKSLOP_TTS_WORDS        1 = synthesize each word, merge (Flextalk);
+#                                             combines with a preset and wins over VOX prosody
 #                   VESKSLOP_TTS_WORD_GAP_MS  silence between words in ms (0-500)
 #                   VESKSLOP_TTS_PRESET       vox | hev (Keith Bell profile preset; vox implies VOX)
 #                   VESKSLOP_TTS_VOX          1 = VOX prosody, NetTTS-style (Flextalk)
@@ -305,9 +306,11 @@ fi
 
 # SAPI4 is ANSI/CP1252. VOX mode (Flextalk only) first transforms the text into
 # vendor tags (\!wH1/\!wH0, \!br, \!sfNN) with setup/vox_parser.py; the parser
-# works on UTF-8, so it runs before the CP1252 transliteration.
+# works on UTF-8, so it runs before the CP1252 transliteration. Word mode builds
+# its own cadence from single-word instances, so the parser is skipped there: the
+# VOX tags cannot survive a word split.
 TEXT_SRC="$TEXT_IN"
-if [ "$VOX" = "1" ] && [ -n "$SPEAKER_GUID" ]; then
+if [ "$VOX" = "1" ] && [ -n "$SPEAKER_GUID" ] && [ "$WORDS" != "1" ]; then
     if command -v "$MICROSLOP_PYTHON" >/dev/null 2>&1 && [ -f "$MICROSLOP_ROOT/setup/vox_parser.py" ]; then
         if [ "$VOX_CLEAN" = "1" ]; then
             "$MICROSLOP_PYTHON" "$MICROSLOP_ROOT/setup/vox_parser.py" --clean < "$TEXT_IN" > "$TMP/vox.txt"
@@ -356,8 +359,9 @@ try_run_host() {
 
 # Word mode (Flextalk): synthesize each word and merge with silence gaps, which
 # gives the staccato radio cadence the Half-Life VOX/HEV voices are known for.
-# VOX mode takes precedence: it produces its own cadence with vendor tags.
+# It combines with a preset (the profile is already written) and takes precedence
-if [ "$WORDS" = "1" ] && [ -n "$SPEAKER_GUID" ] && [ "$VOX" != "1" ]; then
+# over VOX prosody, which cannot survive a word split.
+if [ "$WORDS" = "1" ] && [ -n "$SPEAKER_GUID" ]; then
     GAP=$WORD_GAP
     case "$GAP" in *[!0-9]*|"") GAP=120 ;; esac
     [ "$GAP" -gt 500 ] && GAP=500

poc/public/app.js

@@ -2031,7 +2031,6 @@ function fillProfileControls() {
     $("#tts-vox").checked = state.speech.vox === true;
     $("#tts-vox-clean").checked = state.speech.voxClean === true;
     $("#tts-vox-clean").disabled = state.speech.vox !== true;
-    $("#tts-words").disabled = state.speech.vox === true;
     const gap = Number(state.speech.wordGapMs);
     $("#tts-word-gap").value = Number.isFinite(gap) && state.speech.wordGapMs !== null ? gap : 120;
     $("#tts-word-gap-value").textContent = `${$("#tts-word-gap").value}ms`;
@@ -2744,17 +2743,14 @@ function init() {
     $("#tts-words").addEventListener("change", () => void saveSpeech());
     $("#tts-vox").addEventListener("change", () => {
         $("#tts-vox-clean").disabled = !$("#tts-vox").checked;
-        $("#tts-words").disabled = $("#tts-vox").checked;
-        // The vox preset implies VOX; turning VOX off means "factory/custom" again.
-        if (!$("#tts-vox").checked && $("#tts-preset").value === "vox") $("#tts-preset").value = "";
         void saveSpeech();
     });
     $("#tts-preset").addEventListener("change", () => {
-        // Picking the VOX preset turns on VOX prosody so the panel matches the engine.
+        // Picking the VOX preset turns on VOX prosody as a default. Word mode still wins
+        // for word-by-word delivery, and turning VOX off keeps the preset.
         if ($("#tts-preset").value === "vox") {
             $("#tts-vox").checked = true;
             $("#tts-vox-clean").disabled = false;
-            $("#tts-words").disabled = true;
         }
         void saveSpeech();
     });

poc/src/tts.ts

@@ -381,7 +381,9 @@ export class Tts {
                     ...(options.wordGapMs !== undefined
                         ? { VESKSLOP_TTS_WORD_GAP_MS: String(options.wordGapMs) }
                         : {}),
-                    ...(options.vox ? { VESKSLOP_TTS_VOX: "1" } : {}),
+                    // Always defined, so an explicit "off" is not overridden by the wrapper's
+                    // preset implication (which only fires when the variable is unset).
+                    VESKSLOP_TTS_VOX: options.vox ? "1" : "0",
                     ...(options.voxClean ? { VESKSLOP_TTS_VOX_CLEAN: "1" } : {}),
                     ...(options.preset ? { VESKSLOP_TTS_PRESET: options.preset } : {})
                 },