Project goals
- Stream article audio to browser from a TTS model
- Use an LLM to clean up the article. Remove content that doesn't flow with the main content (captions, footers)
- Use an LLM to annotate the article, allowing titles, blockquotes, and paragraphs to have pauses and emphasis
Features
- Convert your article to audio
- Streaming audio for speed
- Uses a deep learning model to convert audio
- Queues audio conversion tasks using Celery
- Depends on Torch, Flask, gunicorn, redis
- Uses Server Sent Events to stream audio
- Polls conversion API
- Currently just the starter Deno Fresh template.