source selection

Constrain podcast script length via model prompt and rewrite pass

Fix podcast playback and enforce TTS length limit

Fix podcast playback: valid MP3 output, Gradio file route, and TTS input cap

Simplify podcast TTS: save OpenAI MP3 directly, fix audio player

Fix OpenAI TTS by chunking text to respect 4096 char limit

Switch podcast TTS from HF Inference API to OpenAI TTS

Test HF Inference API endpoint URL

Test TTS model: use facebook/mms-tts-eng instead of kakao-enterprise

Update services/podcast_service.py

Update services/podcast_service.py

Update services/podcast_service.py

Update services/podcast_service.py

Update services/podcast_service.py

Update services/summary_service.py

Update services/quiz_service.py

Update services/quiz_service.py

Update services/quiz_service.py

Update services/summary_service.py

Update services/quiz_service.py

Update services/podcast_service.py

Create summary_service.py

Update services/podcast_service.py

Update services/quiz_service.py

Update services/quiz_service.py

Update services/quiz_service.py

Update services/quiz_service.py

Update services/quiz_service.py

Update services/quiz_service.py

Update services/quiz_service.py

Rename services/quiz_Service.py to services/quiz_service.py

Rename services/Podcastt.py to services/podcast_service.py

Rename services/Quizz.py to services/quiz_Service.py

Update services/Podcastt.py

Create Podcastt.py

Create Quizz.py

refactor: move all prompts to prompt.poml, fix greeting handling

add chat history

Switch RAG generation model to Qwen2.5-7B-Instruct

fix(rag): switch HF generation from text_generation to chat_completion for Mistral

added RAG retrieval/generation pipeline and fix source processing flow

RAG: wire chat + prompt + services module