Skip to content

Mihir-A/Cadence

Repository files navigation

Cadence

Cadence is a focused interview practice recorder built with Next.js + Tailwind CSS, Gemini, and TwelveLabs.

Cadence front page

Usage

  • Open Cadence and pick a question to practice.
  • Hit record, answer out loud, then stop when you're done.
  • Review the transcript and scores, then retry or download your clip.

What it does

  • Records interview responses and stores clips locally in the browser.
  • Uses Gemini to transcribe, score technical correctness, and count long pauses + filler words.
  • Uses TwelveLabs to score delivery confidence with concise feedback.
  • Lets you review results, retry a question, and download or delete clips.

Demo

Getting started

Install dependencies:

npm install

Create a .env file with:

GEMINI_API_KEY=your_key
TWELVE_LABS_API_KEY=your_key

Run locally:

npm run dev

Build:

npm run build

Configuration

Optional environment variables:

AI_CALLS_DISABLED=false

When NEXT_PUBLIC_AI_CALLS_DISABLED=true or AI_CALLS_DISABLED=true, Cadence returns placeholder responses and skips external API calls.

Deploy on Vercel

Deploy with the Vercel platform or your preferred host. Ensure the environment variables above are set in your deployment settings.

More screenshots

Cadence feedback view 1

Cadence feedback view 2

About

A interview prep website that gives AI generated feedback on technical accuracy, pauses, filler words, and confidence.

Resources

Stars

Watchers

Forks

Packages

 
 
 

Contributors

Languages