PodTalk

Inspiration

Hackathons are all about rapid innovation—PodTalk lets teams transform any text into a slick podcast episode in seconds, so you can demo audio-driven insights without building a full TTS pipeline.

What it does

Drag-and-drop your docs or paste URLs; PodTalk’s AI auto-summarizes, scripts, and generates high-quality audio, complete with voice options and music beds, ready for your demo.

How we built it

In 48 hours we wired together a Python Flask backend running an extractive-abstractive summarizer and a fine-tuned neural TTS model, all orchestrated via Docker for instant scalability.

Challenges we ran into

Cramming summarization and TTS steps into sub-5-second responses forced us to optimize batch processing and reduce model latency without sacrificing audio clarity.

Accomplishments that we're proud of

We nailed sub-5-second end-to-end generation, integrated live voice-style switching, and spun up a React front end with real-time progress indicators.

What we learned

Fast iteration beats perfect accuracy at hackathons—focus on core value (audio from text) and polish UX around loading states rather than chasing ideal summaries.

What's next for PodTalk

Post-hackathon we’ll open-source our summarization-TTS microservices, add plugin hooks for custom NLP features, and package a minimal CLI for developer adoption.

Built With

• fastapi
• flask
• javascript
• python
• react