SpeakSharp

SpeakSharp is your AI-powered 2-minute speaking coach—helping you improve fluency, clarity, and grammar with real-time feedback, scores, and actionable tips after every practice round.

Comment

🎯 Inspiration English fluency is critical in today’s global world, yet many learners struggle to practice speaking without judgment or feedback. We noticed that while writing and reading tools have exploded with AI, there’s still a gap in accessible, personalized spoken English practice. SpeakSharp was born out of this need — a focused, judgment-free space to speak, reflect, and grow in just two minutes.

🔍 What We Learned Building SpeakSharp taught us how to:

Use the MediaRecorder API effectively for real-time audio recording in the browser.

Integrate OpenAI’s Whisper API for accurate speech-to-text transcription.

Harness the power of GPT-4 to deliver nuanced language feedback on clarity, fluency, grammar, and more.

Streamline user experience using ShadCN UI components and Tailwind CSS to create a calm, distraction-free UI.

Optionally add Text-to-Speech with ElevenLabs API for audible feedback, improving accessibility.

🛠️ How We Built It Frontend Stack: Next.js, Tailwind CSS, ShadCN UI

Core Features:

A single, distraction-free page with a 2-minute countdown

A "Start Speaking" button that records audio via the browser's MediaRecorder API

Auto-uploading the recording to OpenAI Whisper API

Submitting the transcript to GPT-4 with a detailed analysis prompt

Displaying feedback with scores and suggestions

Optional voice feedback via ElevenLabs TTS

Clear state transitions from Idle → Recording → Processing → Feedback

Error handling, retry flow, and a clean UX for focused language learning

⚙️ Challenges Faced Ensuring audio recording compatibility across browsers using MediaRecorder

Managing long audio uploads and accurate conversion to formats acceptable by the Whisper API

Optimizing GPT-4 prompts to ensure constructive, specific, and consistent feedback

Balancing simplicity with usefulness — keeping the interface clean while delivering impactful results

Tuning the user flow to minimize friction and encourage repeated practice

SpeakSharp is built for learners, professionals, and anyone looking to practice and polish their English — one bite-sized session at a time.

Built With

  • elevenlabs
  • netlify
  • next.js
  • openai
  • shadcn
  • tailwind-css
Share this project:

Updates