*The technical approach:* Instead of basic word counting, Orator analyzes five dimensions based on communication research: - *Pitch:* Time & frequency-domain analysis for vocal variety detection and monotone identification - *Pause:* Silence detection with timing analysis - *Pace:* Real-time WPM calculation with engagement optimization - *Power:* Volume dynamics and emphasis pattern recognition - *Prose:* Readability scoring and linguistic complexity analysis
*Architecture:* AI-based audio processing, real-time visualization with pitch movements and power variations, pause timing charts, and comprehensive analytics dashboard. Frontend using Next.js with Typescript, React and Tailwind, backend Python Flask API and various LLM/AI services
*Why free:* Wanted to democratize access to professional-level speech coaching. No signup required, all advanced features included.
*Demo:* You can analyze famous speeches (e.g. Ellen DeGeneres, Steve Jobs) or upload your own recordings. The visual analytics are surprisingly insightful – you can see quantitatively what you’re good at and what you need to improve.
Try it: https://orator.feynman.ai/
Looking for feedback on the technical approach and whether these analytics actually help people improve. What speech analysis features would be most valuable to the HN community?