fix: completely change voices for Dorothy, Harold, and Marcus

fix: run edge-tts locally on Space (removes dependency on droplet for audio)

fix: characters now know their identity; use real emotion tags; tune voices (Harold/Carlos to Edge, faster Dorothy)

feat: switch to Kokoro ONNX TTS; fix Carlos Spanish voice; diverse per-avatar voices

fix: retry Gradio connection every 5s on failure; remove IP from offline message

fix: return base64 audio data URI from lumi_api; tune per-avatar edge-tts voices

fix: mount Gradio before catch-all route so /gradio/config resolves correctly

fix: use positional array in predict() call — resolves 'data is not a valid keyword arg' error

Upload frontend + backend

style: remove sun symbol from header

fix: hide empty video drop-zone placeholder in sky-blue theme

style: clear-sky theme — white/sky-blue, large fonts, elderly-friendly UI

fix: replace streaming VAD with record→send voice chat; remove upload/share buttons from Chatbot and Image; fix portrait note

Upload pipeline/tts.py with huggingface_hub

Upload pipeline/profiles.py with huggingface_hub

Upload app.py with huggingface_hub

Upload pipeline/talking_head.py with huggingface_hub

Upload portraits/liam.png with huggingface_hub

Upload portraits/amara.png with huggingface_hub

Upload portraits/dorothy.png with huggingface_hub

Upload portraits/harold.png with huggingface_hub

Upload portraits/james.png with huggingface_hub

Upload portraits/marcus.png with huggingface_hub

Upload portraits/priya.png with huggingface_hub

Upload portraits/mei.png with huggingface_hub

Upload portraits/sophie.png with huggingface_hub

Upload portraits/carlos.png with huggingface_hub

Upload app.py with huggingface_hub

Upload app.py with huggingface_hub

Upload requirements.txt with huggingface_hub

Upload requirements.txt with huggingface_hub

Upload app.py with huggingface_hub

Upload app.py with huggingface_hub

Upload pipeline/stt.py with huggingface_hub

Upload pipeline/tts.py with huggingface_hub

Upload README.md with huggingface_hub

fix: add type=messages to both Chatbots to fix gradio 5.9 API info crash

fix: bump sdk_version to 5.9.1, remove duplicate gradio pin

fix: bump sdk_version to 5.9.1, remove duplicate gradio pin

fix: replace pyaudioop with audioop-lts (correct PyPI package for Python 3.13)

fix: add pyaudioop for Python 3.13 audioop backport

Upload avatar/avatar_smile.png with huggingface_hub

Upload avatar/avatar_nod.png with huggingface_hub

Upload avatar/avatar_laugh.png with huggingface_hub

Upload avatar/avatar_gentle.png with huggingface_hub

Upload avatar/avatar_concerned.png with huggingface_hub

Upload pipeline/tts.py with huggingface_hub

Upload pipeline/scam_filter.py with huggingface_hub

Upload pipeline/parser.py with huggingface_hub

Upload pipeline/__init__.py with huggingface_hub