feat: THE CREATOR god-tier character, bitcoin choreographies, omni-morph, cameos

- Guy Fawkes mask archetype with golden outline, bitcoin chest symbol, laptop, tier-gated effects
- 12 custom bitcoin-themed choreographies (8 regular + 4 exclusive ultimates)
- Omni-morph system: creator morphs into any of 80+ archetypes instead of cycling 3
- 6% per-round cameo in non-creator fights — drops golden ₿ gifts to fighters
- Custom golden code rain entrance with persistent orbiting particles
- pickChoreography: 50% ultimate chance (100% on crits), all tier ultimates + 4 exclusive
- Server auto-assigns the_creator archetype on login/register for creator pubkey
- FightViewer, payments, queue, orchestrator improvements

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

frontend/src/game/sounds.ts

@@ -104,6 +104,7 @@ const voiceProfiles: Record<string, VoiceProfile> = {
   robot:       { voice: null, pitch: 0.8, rate: 0.8, volume: 0.9 },    // Steady monotone
   screamer:    { voice: null, pitch: 1.3, rate: 1.6, volume: 1.0 },    // Frantic energy
   smooth:      { voice: null, pitch: 1.0, rate: 0.9, volume: 0.9 },    // Natural narrator
+  question_reader: { voice: null, pitch: 1.0, rate: 1.05, volume: 1.0 }, // Clear, brisk question announcer
   // 24 new profiles
   whisper:     { voice: null, pitch: 1.2, rate: 0.6, volume: 0.4 },    // Quiet dramatic whisper
   boomer:      { voice: null, pitch: 0.4, rate: 0.6, volume: 1.0 },    // Ultra deep booming
@@ -204,6 +205,12 @@ function loadVoices() {
   voiceProfiles.robot.voice = findVoice([/zarvox/i, /trinoids/i, /albert/i]) || anyVoices[0]
   voiceProfiles.screamer.voice = findVoice([/samantha/i, /karen/i, /bells/i]) || anyVoices[Math.min(2, anyVoices.length - 1)]
   voiceProfiles.smooth.voice = findVoice([/daniel/i, /oliver/i, /google.*uk/i]) || anyVoices[Math.min(1, anyVoices.length - 1)]
+  // Question reader: highest quality voice available — tries premium/neural first
+  voiceProfiles.question_reader.voice = preferPremium([
+    /evan.*premium/i, /samantha.*enhanced/i, /daniel.*premium/i,
+    /google.*us.*english/i, /google.*uk.*english/i,
+    /evan/i, /aaron/i, /daniel/i, /james/i,
+  ]) || anyVoices[0]
   // Assign voices to new profiles — spread across available voices for max variety
   const vLen = anyVoices.length
   const pick = (i: number) => anyVoices[i % vLen]
@@ -340,6 +347,52 @@ function speak(text: string, profileName: string, cancelPrevious: boolean = fals
   speechSynthesis.speak(utter)
 }
 
+/** Like speakAsync but with a forced minimum rate */
+function speakAsyncWithRate(text: string, profileName: string, minRate: number): Promise<void> {
+  return new Promise<void>((resolve) => {
+    if (typeof speechSynthesis === 'undefined' || masterMuted) { resolve(); return }
+    if (!voicesLoaded) loadVoices()
+    if (speechSynthesis.paused) speechSynthesis.resume()
+    const profile = voiceProfiles[profileName] || voiceProfiles.announcer
+    const utter = new SpeechSynthesisUtterance(text)
+    if (profile.voice) utter.voice = profile.voice
+    utter.pitch = profile.pitch
+    utter.rate = Math.max(minRate, profile.rate)
+    utter.volume = profile.volume * VOICE_VOLUME_SCALE
+    _speechQueueDepth++
+    const safetyTimeout = setTimeout(resolve, 15_000)
+    utter.onend = () => { clearTimeout(safetyTimeout); _speechQueueDepth = Math.max(0, _speechQueueDepth - 1); resolve() }
+    utter.onerror = () => { clearTimeout(safetyTimeout); _speechQueueDepth = Math.max(0, _speechQueueDepth - 1); resolve() }
+    speechSynthesis.speak(utter)
+  })
+}
+
+/** Like speak() but returns a promise that resolves when the utterance finishes */
+function speakAsync(text: string, profileName: string, cancelPrevious: boolean = false): Promise<void> {
+  return new Promise<void>((resolve) => {
+    if (typeof speechSynthesis === 'undefined' || masterMuted) { resolve(); return }
+    if (!voicesLoaded) loadVoices()
+    if (speechSynthesis.paused) speechSynthesis.resume()
+    if (cancelPrevious) { speechSynthesis.cancel(); _speechQueueDepth = 0 }
+    const profile = voiceProfiles[profileName] || voiceProfiles.announcer
+    const utter = new SpeechSynthesisUtterance(text)
+    if (profile.voice) utter.voice = profile.voice
+    utter.pitch = profile.pitch
+    utter.rate = profile.rate
+    utter.volume = profile.volume * VOICE_VOLUME_SCALE
+    _speechQueueDepth++
+    utter.onend = () => { _speechQueueDepth = Math.max(0, _speechQueueDepth - 1); resolve() }
+    utter.onerror = () => { _speechQueueDepth = Math.max(0, _speechQueueDepth - 1); resolve() }
+    // Safety timeout — never block forever (max 15s for any utterance)
+    const safetyTimeout = setTimeout(resolve, 15_000)
+    const origOnEnd = utter.onend
+    utter.onend = (ev) => { clearTimeout(safetyTimeout); (origOnEnd as any)(ev) }
+    const origOnError = utter.onerror
+    utter.onerror = (ev) => { clearTimeout(safetyTimeout); (origOnError as any)(ev) }
+    speechSynthesis.speak(utter)
+  })
+}
+
 export function stopAllAudio() {
   stopMusic()
   if (typeof speechSynthesis !== 'undefined') speechSynthesis.cancel()
@@ -512,6 +565,52 @@ export function announceRoundHype() {
   voices[Math.floor(Math.random() * voices.length)](line)
 }
 
+// === TTS: Questions, Answers & Narration ===
+// Distinct voice roles so players can always tell who's speaking.
+// Question = clear smooth reader, Answers = unique per-bot (intelligible subset),
+// Narration = dramatic judge.
+
+// Only voices with rate >= 0.6 and rate <= 1.3, and pitch between 0.5-1.5
+// so every voice is actually understandable by humans
+const INTELLIGIBLE_VOICES = [
+  'smooth', 'robot', 'drill', 'surfer', 'pirate_v', 'cowboy_v', 'ninja_v',
+  'posh', 'aussie', 'scottish', 'french', 'texan', 'wrestler_v', 'ai_core',
+  'android_v', 'siri', 'professor', 'npc', 'news', 'crotchety', 'mech',
+  'wizard_v', 'echo_v', 'boss_taunt', 'angel', 'punk', 'valley',
+]
+
+function hashName(name: string): number {
+  let h = 0
+  for (let i = 0; i < name.length; i++) h = ((h << 5) - h + name.charCodeAt(i)) | 0
+  return Math.abs(h)
+}
+function botVoiceKey(name: string): string {
+  return INTELLIGIBLE_VOICES[hashName(name) % INTELLIGIBLE_VOICES.length]
+}
+
+/** Speak the challenge question — premium voice, clear and brisk.
+ *  Does NOT cancel previous speech so intro/hype lines finish naturally.
+ *  Returns promise that resolves when the question finishes reading. */
+export function speakQuestion(text: string): Promise<void> {
+  const trimmed = text.length > 140 ? text.slice(0, 137) + '...' : text
+  return speakAsync(trimmed, 'question_reader', false)
+}
+
+/** Speak a bot's answer in their assigned character voice at brisk pace.
+ *  Returns promise that resolves when finished. */
+export function speakAnswer(botName: string, answer: string): Promise<void> {
+  const trimmed = answer.length > 80 ? answer.slice(0, 77) + '...' : answer
+  // Use speakAsync with a rate boost: override profile rate to min 1.1
+  return speakAsyncWithRate(trimmed, botVoiceKey(botName), 1.15)
+}
+
+/** Speak the judge's narration — brisk sportscaster energy.
+ *  Returns promise that resolves when finished. */
+export function speakNarration(text: string): Promise<void> {
+  const trimmed = text.length > 160 ? text.slice(0, 157) + '...' : text
+  return speakAsyncWithRate(trimmed, 'sportscaster', 1.2)
+}
+
 // === FANFARES (8-bit melodic announcements) ===
 
 export function fanfareRound(roundNum: number) {