This tool is a simple paper to speech player that leverages OpenAI's TTS and GPT 4o-mini to create a podcast-like audio for a given research paper.
- Five-sentence lookahead to avoid excessive api calling;
- Semantically segmentation of the text with sat models;
- Simplification for narration with a GPT model;
- TTS with OpenAI's endpoint;
The project was built with the help from Claude.