ayush23chaudhary/SpeakWise-Speech-AI-Coach

SpeakWise is a thoughtful full-stack project with a clear point of view. Instead of offering generic speech statistics, it focuses on how an interviewer, investor, or examiner may perceive the speaker. The different evaluation modes, recorded-response analysis, practice scenarios, progress tracking, achievements, leaderboards, and guest access make it feel closer to a real product than a basic AI demo. The React interface has organized routing and protected account areas, while the Express backend separates its routes, models, controllers, and services. Integrating MongoDB, Google Speech-to-Text, Gemini, OAuth, JWT authentication, analytics, and production hosting shows strong practical development experience. The repository also keeps credentials out of source control and provides useful environment setup guidance.
The main improvement area is consistency between the documentation and the code. The README’s clone command points to a different GitHub account and repository. Its project structure also lists files that do not exist under those names, while leaving out many features that are present in the current application. The licensing language conflicts as well. The repository includes an MIT license, which permits commercial use, but the README tells users to contact the author before using it commercially. That should be resolved. Testing is currently described as manual, and neither package includes a real automated test suite. Adding API, authentication, speech-processing, and frontend component tests would make future changes safer. The server should also remove its hardcoded fallback session secret and fail clearly when required production secrets are missing. Overall, SpeakWise has a strong concept, substantial working scope, and a solid foundation. More accurate documentation, automated testing, and tighter production security would make the project much easier to trust, run, and contribute to.