Add episode ending chime with silence padding

Update README with setup gotchas and IAM policy

Fix X Articles with empty body silently using preview text

Fix Telegram bot not showing summary after progress completes

Add Google Gemini as LLM backend with TTS reliability fixes

Fix Ollama backend for thinking models and large inputs

Add Ollama model picker to Telegram /model command

Add interactive model picker with --model flag

Add qwen3.5:9b as documented local model option

Rename project from article2audio to a2pod

Add MIT license

Add preventive .gitignore patterns for secrets

Switch to single-provider architecture (local or S3)

Make local feed the default — no AWS needed out of the box

Add /speed Telegram command to control TTS speech rate