PracticePerfect - Improve your speaking skills locally
This is a submission for the Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle What I Built I sometimes start speaking before I’ve figured out how to organize what I want to say. I know the point I’m trying to make, but halfway through an explanation, I’m repeating myself or losing the thread. I wanted somewhere to practice that felt private and easy to return to. So I built…
PracticePerfect is a speaking-practice application that enables users to hone their communication skills privately and conveniently. The app, built by an individual developer, employs open-source AI models running locally on a laptop. Upon launching, users select a daily topic, record their response via phone, and transfer the audio to their computer.
The app then processes the recording into a text transcript using Whisper, an open-source speech recognition model. Afterward, Gemma, another open model running locally via Ollama, assesses the transcript against the chosen topic and provides structured feedback on areas for improvement. Users may try again, applying the insights gained from the feedback, and continue refining their speaking abilities.
The app's architecture separates the transcription and feedback processes, allowing each AI model to focus on its designated task. The user interface is crafted with React and TypeScript, styled using Tailwind CSS, while Python and FastAPI power the backend. FFmpeg facilitates audio file conversion, and SQLite and local file storage manage practice records and audio files.
The app's creators emphasize that feedback scores are AI-generated, not clinical assessments, and are intended for personal skill development rather than objective evaluation. The project is currently in a pre-revenue testing phase, exploring potential paid features such as guided practice programs and enhanced progress tracking.
The open-source nature of the application allows users to maintain control over their data, utilizing local inference models without relying on third-party inference APIs.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
