Commit Graph

14 Commits

Author SHA1 Message Date
c70f83d04a Cost monitoring, PTT fix, Devon tuning, WEIRD pool expansion, YT thumbnails, LLM SEO, publish ep37
- Add real-time LLM/TTS cost tracking with live status bar display and post-show reports
- Fix PTT bug where Devon suggestion layout shift stopped recording via mouseleave
- Devon: facts-only during calls, full personality between calls
- Double WEIRD topic pool (109→203), bump weight to 14-25%
- Auto-generate YouTube thumbnails with bold hook text in publish pipeline
- LLM SEO: llms.txt, robots.txt for LLM crawlers, structured data, BreadcrumbList schemas
- Publish episode 37

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-15 05:33:27 -06:00
3329cf9ac2 UI cleanup, Devon overhaul, bug fixes, publish ep36
- Fix Devon double messages, add conversation persistence, voice-to-Devon when no caller
- Devon personality: weird/lovable intern on first day, handles name misspellings
- Fix caller gender/avatar mismatch (avatar seed includes gender)
- Reserve Sebastian voice for Silas, ban "eating at me" phrase harder
- Callers now hear Devon's commentary in conversation context
- CSS cleanup: expand compressed blocks, remove inline styles, fix Devon color to warm tawny
- Reaper silence threshold 7s → 6s
- Publish episode 36

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-14 16:42:21 -06:00
6d4e490283 Caller generation overhaul, Devon intern, frontend redesign
Caller system: structured JSON backgrounds, voice-personality matching (68 profiles),
thematic inter-caller awareness, adaptive call shapes, show pacing, returning caller
memory with relationships/arcs, post-call quality signals, 95 comedy writer entries.

Devon the Intern: persistent show character with tool-calling LLM (web search, Wikipedia,
headlines, webpage fetch), auto-monitoring, 6 API endpoints, full frontend UI.

Frontend: wrap-up nudge button, caller info panel with shape/energy/emotion badges,
keyboard shortcuts (1-0/H/W/M/D), pinned SFX, visual polish, Devon panel.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-14 01:54:08 -06:00
0bdac16250 Upgrade Whisper to distil-large-v3, fix caller identity confusion, sort clips list
- Whisper base → distil-large-v3 for much better live transcription accuracy
- Add context hints to transcription (caller name, screening status)
- Increase beam_size 3→5 for better decoding
- Add explicit role clarification in caller system prompt so LLM knows Luke is the host
- Prefix host messages with [Host Luke] in LLM conversation
- Fix upload_clips episode list sorting (natural numeric order)
- Episodes 26-28 transcripts, data updates, misc fixes

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-05 12:46:51 -07:00
d611f60743 SFX emojis, non-blocking email view, deploy/git docs in CLAUDE.md
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-02-16 05:34:25 -07:00
d85a8d4511 Add listener email system with IMAP polling, TTS playback, and show awareness
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-02-16 05:22:56 -07:00
3164a70e48 Ep13 publish, MLX whisper, voicemail system, hero redesign, massive topic expansion
- Switch whisper transcription from faster-whisper (CPU) to lightning-whisper-mlx (GPU)
- Fix word_timestamps hanging, use ffprobe for accurate duration
- Add Cloudflare Pages Worker for SignalWire voicemail fallback when server offline
- Add voicemail sync on startup, delete tracking, save feature
- Add /feed RSS proxy to _worker.js (was broken by worker taking over routing)
- Redesign website hero section: ghost buttons, compact phone, plain text links
- Rewrite caller prompts for faster point-getting and host-following
- Expand TOPIC_CALLIN from ~250 to 547 entries across 34 categories
- Add new categories: biology, psychology, engineering, math, geology, animals,
  work, money, books, movies, relationships, health, language, true crime,
  drunk/high/unhinged callers
- Remove bad Inworld voices (Pixie, Dominus), reduce repeat caller frequency
- Add audio monitor device routing, uvicorn --reload-dir fix
- Publish episode 13

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-02-16 01:56:47 -07:00
28af0723c7 Ep12 publish, caller prompt overhaul, favicon, publish fixes, website updates
- Reworked caller prompt: edgy/flirty personality, play along with host bits
- Bumped caller token budget (200-550 range, was 150-450)
- Added 20 layered/morally ambiguous caller stories
- Valentine's Day awareness in seasonal context
- Default LLM model: claude-sonnet-4-5 (was claude-3-haiku)
- Publish: SCP-based SQL transfer (fixes base64 encoding on NAS)
- Favicons: added .ico, 48px, 192px PNGs for Google search results
- Website: button layout cleanup, privacy page, ep12 transcript
- Control panel: channel defaults match audio_settings.json
- Disabled OP3 permanently (YouTube ingest issues on large files)

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-02-14 22:53:34 -07:00
9fd977ad9f Postprod overhaul, control panel theme, caller names, website updates
- Fix denoise mangling host audio: strip aggressive afftdn/anlmdn, keep HPF only
- Add stem limiting for ads/SFX to prevent clipping
- Spoken-word compression on host (threshold -28dB, ratio 4:1)
- Add bus compressor on final stereo mix (LRA 7.9 → 5.7 LU)
- Drop SFX mix level from -6dB to -10dB
- De-esser fix: replace split-band with simple high-shelf EQ
- Pipeline now 15 steps (was 13)
- Control panel theme: match website warm brown/orange palette
- Expand caller names to 160 (80M/80F), fix duplicate name bug
- Update how-it-works page: returning callers, 15-step pipeline, remove busy diagram row

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-02-12 04:39:31 -07:00
7d88c76f90 Add post-production pipeline: stem recorder, postprod script, recording UI
New stem recording system captures 5 time-aligned WAV files (host, caller,
music, sfx, ads) during live shows. Standalone postprod.py processes stems
into broadcast-ready MP3 with gap removal, voice compression, music ducking,
and EBU R128 loudness normalization.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-02-08 17:53:32 -07:00
356bf145b8 Add show improvement features: crossfade, emotions, returning callers, transcripts, screening
- Music crossfade: smooth 3-second blend between tracks instead of hard stop/start
- Emotional detection: analyze host mood from recent messages so callers adapt tone
- AI caller summaries: generate call summaries with timestamps for show history
- Returning callers: persist regular callers across sessions with call history
- Session export: generate transcripts with speaker labels and chapter markers
- Caller screening: AI pre-screens phone callers to get name and topic while queued

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-02-07 02:43:01 -07:00
d5fd89fc9a Add on-air toggle for phone call routing
When off air, callers hear a message and get disconnected. When on
air, calls route normally. Toggle button added to frontend header
with pulsing red ON AIR indicator.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-02-06 14:03:38 -07:00
db134262fb Add frontend: call queue, active call indicator, three-party chat, three-way calls
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-02-05 13:46:19 -07:00
029ce6d689 Initial commit: AI Radio Show web application
- FastAPI backend with multiple TTS providers (Inworld, ElevenLabs, Kokoro, F5-TTS, etc.)
- Web frontend with caller management, music, and soundboard
- Whisper transcription integration
- OpenRouter/Ollama LLM support
- Castopod podcast publishing script

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-02-04 23:11:20 -07:00