live
LingoPT
AI-powered English speaking coach built as a Christmas gift for my mom
$ apt list --installed
AWSGenAIReactServerless
metrics.sh
$ metrics --project=lingopt
> AiModel: Claude 3.5 Sonnet
> AwsServices: 5 (Bedrock, Polly, Transcribe, S3, Lambda)
An AI-powered English speaking practice app I built as a Christmas 2024 gift for my mom. LingoPT combines speech recognition, Claude-powered feedback, and text-to-speech to create a personal English learning companion.
How It Works
- •Practice Mode — Pick a prompt (or create your own), record yourself speaking, and get AI-generated feedback with corrections, explanations, and a natural audio playback of the improved version via AWS Polly
- •Chat Mode — Ask an AI English teacher grammar questions, vocabulary help, or pronunciation tips, with context from recent practice sessions
Tech Stack
- •Frontend: React + Vite, deployed on AWS Amplify
- •Backend: Node.js/Express on AWS Lambda behind API Gateway
- •AI: Claude 3.5 Sonnet via AWS Bedrock for feedback generation and Q&A
- •Speech: AWS Transcribe (speech-to-text) + AWS Polly (text-to-speech)
- •Storage: S3 for temporary audio processing
Key Features
- •Cross-browser speech recording with multi-format support
- •Structured feedback: encouraging comments, specific corrections with explanations, and an improved version
- •Adjustable playback speed (normal & slow) for listening practice
- •Dark mode support
- •Fully serverless architecture
$ ls ./screenshots
practice-mode.png

Practice Mode — record speech, get AI feedback with corrections and a better version
chat-mode.png

Chat Mode — ask grammar, vocabulary, or pronunciation questions
$git clone [private repository]