Speeches hosted in Google Cloud SQL and app served from Google App Engine. App Link.
This repo is continuation of DJT-Speech-Generator repo. The model hosted in Google Cloud Run uses CPU to generate speech in real-time, but it takes up to 60 sec because it's being served by CPU.
To serve the speech faster, I batch generated 10,000+ speeches from the finetuned GPT-2 model. The speeches are stored in a database in Google Cloud SQL. I created a Google Cloud App to take to display a randomly selected speech.
- Flask
- Jinja
- Bootstrap
- Postgre
- Python