VoxMind
VoxMind is an AI-powered public speaking trainer for people who freeze, spiral, or lose their train of thought while speaking. It gives you a private place to practice, transcribes your speech in real time, and turns each session into structured feedback you can use for the next rep.
The product is built around a simple belief from the landing page: your voice is a muscle, and it can be trained.
What the Product Does
VoxMind helps users rebuild speaking confidence through short, repeatable practice sessions:
- Pick a mode like Extempore, Technical Explainer, or Group Discussion
- Get an AI-generated prompt or question
- Speak your answer out loud in the browser
- See your words appear in real time
- Receive feedback on structure, clarity, and completeness
The app is designed for students, professionals, interview candidates, and anyone who has ever gone blank mid-sentence.
Core Practice Modes
Extempore: random topics for thinking on your feetTechnical Explainer: interview-style technical speaking practiceGroup Discussion: argument-building and structured opinion practice
The codebase also includes upcoming modes like Debate, Story Narration, and Sales Pitch.
Product Philosophy
VoxMind presents itself as more than a speech tool. The landing page frames it as confidence training based on:
- Graduated exposure therapy
- Cognitive behavioral therapy principles
- Deliberate practice with high-feedback repetition
Each session is meant to be one more rep toward calmer, clearer speaking.
Key Features
- Real-time speech transcription
- AI-generated prompts and feedback
- Session history and streak tracking
- Resume-aware technical interview support
- Browser-based experience with optional authentication
- First-session flow that can work without a required account
Tech Stack
React 19TypeScriptViteTanStack RouterTanStack StartTailwind CSSClerk for authenticationNVIDIA NIM for question generation and feedback
Local Development
Install dependencies and start the app:
npm install
npm run dev
Environment
The project currently includes a placeholder Clerk key in .env:
VITE_CLERK_PUBLISHABLE_KEY=pk_test_REPLACE_ME
Authentication is optional in the current landing flow. If Clerk is not configured, the primary CTA still routes users into the app.
For AI features, VoxMind expects access to NVIDIA NIM. If a server-side key is not available, the app asks the user to paste their own NVIDIA API key and keeps it on the device.
Why This Project Exists
The landing page centers the founder story around presentation anxiety, blanking out mid-talk, and the lack of a safe place to practice every day. VoxMind exists to make speaking practice private, repeatable, and honest, so improvement comes from reps instead of pressure.
Status
This is an active browser app with a polished landing page, a guided speaking workflow, and feedback-driven practice loops already in place.
