MyArxivPodcast orchestrates an end-to-end AI pipeline to transform scholarly content into engaging audio shows. First, it polls arXiv APIs for new research submissions in user-defined categories and retrieves metadata and abstracts. Next, it invokes OpenAI's GPT-4 model to craft clear and concise summaries, highlighting key contributions and results. Summaries are fed into a TTS engine such as Amazon Polly or Google Cloud Text-to-Speech, producing natural-sounding narration. The agent automatically tags and organizes the generated audio, compiles episodes, updates an RSS feed, and handles file hosting integration. Advanced settings allow custom voice selection, summary length control, publication schedules, and distribution via popular podcast platforms, providing researchers and listeners with seamless, up-to-date scientific audio briefings.
