- Unified speakAsync/speakAsyncWithRate into _speakAsyncCore with:
- 500ms startup check: bail immediately if speech won't start (mobile/no gesture)
- 8s safety timeout (down from 15s) to prevent blocking
- iOS-safe keepalive: only do Chrome pause/resume workaround on desktop Chrome
- Immediate bail when no voices loaded
- Sprite loading: 5s timeout per sprite prevents mobile hangs from stuck Image decodes
- Scene creation: 10s timeout in FightViewer so overlay/voice flow continues even if
canvas fails on mobile
- playRound: bail gracefully if fighter sprites missing instead of crashing
- Global audio unlock: first touch/click on site unlocks AudioContext + SpeechSynthesis
- Practice button pre-unlocks audio while still in user gesture context
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
TTS fixes:
- Remove 60-char truncation in FightViewer — let speakAnswer handle limits
- Smart truncation at sentence boundaries (period, comma, etc.) up to 200 chars
- Chrome keepalive: periodic pause/resume prevents silent 15s cutoff bug
- Deduplicated cleanup logic in speakAsync/speakAsyncWithRate
Retro mode:
- Simplified challenge prompt from verbose wall of text to clean 3-line format
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Fight Card page: dramatic poster background with cross-hatch, spotlights,
vignettes, corner brackets, scan lines; 3D VS orb with punch animation;
selectable undercard with main event always pinned at top
- PosterSprite: high-quality 480px poster frame with 6-pass renderer
(aura, glow, bevel, specular, particles); PixelGlove component
- 12-char bot name limit across all forms and server validation
- Mock bots: all 100 now have diverse archetypes (25 types), 25% human
fighters; seedMockBots updates existing bots on restart
- Leaderboard: inline SpritePreview next to each bot name
- Nostr auth: persistent login, nsec copy button
- Wallet: NWC + Lightning Address, ranked fight flow
- Server: payments, ranked queue, customization endpoint
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Disconnect and recreate sfxGain in stopAllAudio() so oscillators
and noise buffers stop immediately instead of playing to completion.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- texan profile pitch/rate adjusted (0.65/0.75) to differentiate from deep (0.7/0.7)
- Audited all 60 voice profiles: no remaining pitch+rate duplicates
- Category announces verified: all fire correctly via speak() pipeline
- Speech queue already capped at depth 2 with cancel, no debounce needed
- TTS fallback already graceful: speak() returns silently when unavailable
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- visibilitychange handler resumes speech + AudioContext on tab return
- speak() retry logic: re-attempts once on non-cancel errors after 200ms
- VOICE_VOLUME_SCALE (0.7) prevents voice from overpowering SFX/music
- Music crossfade: brief gain dip (40% -> 100%) on track switches
- Applied volume scale to all speech paths (speak, announce custom)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- FightViewer: track deferred speechSynthesis.cancel() timer, clear on unmount
- BotProfilePage: change isOwner from ref to computed for reactive login state
- sounds.ts: tighten speech queue depth cap from 3 to 2
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Remove stray duplicate template block outside TEMPLATES array that caused parse error preventing server startup (fixes /api/queue/status 500)
- Add choices field to Challenge/PromptEntry interfaces for multiple choice support
- Allow voice lines to overlap naturally instead of cancelling previous speech on every call, with safety flush at depth > 3
- Buffer challenges during cooldown in polling path, auto-submit random choice on timeout
- Block auto-battle for human fights, guard replay-done auto-transition
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Add choose-mode step: "I BUILD BOTS" vs "I FIGHT MYSELF" paths
- Human registration with baby avatar picker, no webhook required
- Live fight scene with SSE round streaming and real-time challenge UI
- 5-second timer per round, submit answers via browser
- Baby → toddler → kid → teen → adult → hero → super growth stages
- Huge sparkly baby eyes, diapers, pacifiers, bibs, rattles, rosy cheeks
- Speech bubble positioning fix (pushed to outside of sprite)
- Canvas text rendering via offscreen canvas to bypass kaplay color issues
- Voice timing improvements: await pauses between voice lines and hits
- 30 devastating announcement lines, 15 critical/hit word variants
- Orchestrator human player detection + waitForHumanResponse system
- Server endpoints: GET /challenge/:botId, POST /respond/:botId
- Human player auth: register-human route, isHuman flag on login
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Change speak() default to cancelPrevious=true so voices don't queue endlessly
- Add stopAllAudio() export and wire into FightViewer unmount + post-fight
- Flush speech queue at fight end with delayed cancel for clean cutoff
- Expand from 30 to 60+ voice profiles (robots, accents, game, characters)
- Add speech bubbles showing bot responses during rounds
- Wire announceFinishHim/Fatality/FlawlessVictory/Devastating to not cancel
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Add Dockerfile (multi-stage: build frontend + server, serve from single container)
- Add docker-compose.yml for Portainer stack deployment
- Server serves frontend SPA in production (static assets + SPA fallback)
- Auto-run migrations and seed mock bots on server startup
- DB path configurable via DB_PATH env var
- Add "Fight a Classic Bot" button for instant mock bot matches
- FIGHT button queues for real AI opponents
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
