SwiftSpeak

SwiftLearn
Arcchitecture Diagram

Inspiration

The articulation of spoken English is a major challenge for many people, both in English speaking countries, and globally. While many can write well, speaking fluently and confidently remains a struggle. SwiftSpeak was born to solve this problem. We also included Nigeria’s three major languages, Igbo, Hausa and Yoruba, to support both native speakers and foreigners who want to learn these languages with accurate pronunciation. Our goal is to make language learning more inclusive, accessible and culturally rich.

What it does

Swiftspeak is a real-time AI-powered voice learning platform hosted on Google Cloud Run, designed to help users practice speech, improve pronunciation, and get instant feedback using serverless and scalable cloud infrastructure. SwiftSpeak helps users learn and practice in four languages: English, Igbo, Yoruba and Hausa. Users can: *Select a language. *Hear native pronunciation of common phrases. *Practice speaking and compare their pronunciation.

How we built it

We built SwiftSpeak using: *Frontend: [Next.js / TypeScript / Tailwind] - UI served from Cloud Run; responsive and fast. *Backend API: Node.js (Express / Nest) - Handles audio uploads, transcription, and feedback logic. *Containerization: Docker + Cloud Build - Multi-stage Dockerfile for lightweight deploys. Hosting: Google Cloud Run - Scalable, serverless deployment. Database: Firestore - Stores user data & feedback scores. *Storage: Cloud Storage (for audio files) - Stores audio securely and cheaply.

Cloud run allowed us deploy our backend quickly and scale it automatically without managing servers.

Challenges we ran into

Multi-stage Docker build issues (COPY paths, tsconfig resolution). Adapting monorepo structure for Cloud Build. Ensuring low latency between front-end and API services. Managing authentication for multiple microservices.

Accomplishments that we're proud of

Fully deployed and live on Google Cloud Run. Integrated CI/CD pipeline using Cloud Build. Auto-scaling infrastructure with zero downtime. Clean, modern UI and working speech feedback system.

What we learned

How to optimize Docker images for Node.js Cloud Run services. How to use multi-service deployments (web + API) under one project. How Cloud Run cold starts and concurrency affect latency. How to debug gcb-internal vs. docker builds in Cloud Build.

What's next for SwiftSpeak

We plan to: •Add gamification features like badges and levels. •Include speech recognition for pronunciation feedback. •Expand to more languages. •Build offline access for low connectivity areas. Add multilingual support and gamified learning.

SwiftSpeak is just getting started. We’re excited to keep growing and improving it.