Phoneme Analyzer
Daily practice sentences
System Architecture
Daily activity tracker
Progress chart
Use history

Inspiration

We thought about international students, we know people who face language barrier. Learning English pronunciation is hard, especially for people who don’t speak it as their first language. Old ways of learning don’t give quick feedback or help that’s made just for you. That’s why we made naccent. We use AI to fix this problem, making pronunciation practice easier, fun, and way better for everyone.

What it does

Voice recording capabilities
Playback comparison with native speakers
Adjustable difficulty levels
Comprehensive phoneme analysis

How we built it

Frontend: Next.js 14+
UI Framework: Tailwind CSS with shadcn/ui
Speech Processing with TTS1-Nova:
- Whisper API for transcription
- SpeechAce API for pronunciation assessment
- Claude AI for content generation for daily quotes
Authentication: Built-in Next.js authentication
Styling: Tailwind CSS with custom theming
State Management: React hooks
Charts: Chart.js for analytics visualization -Flask: Phonemizer library

Naccent - AI-Powered English Pronunciation Trainer

Naccent is an innovative web application designed to help users improve their English pronunciation using advanced AI technology. The platform combines speech recognition, real-time feedback, and daily practice exercises to enhance users' speaking skills.

Features

1. Real-time Pronunciation Assessment

Advanced speech recognition technology
Phoneme-level pronunciation feedback
Quality scores for each sound
Visual feedback with color-coded results
Comparison with native speaker pronunciation

2. Daily Practice Materials

Tongue twisters for pronunciation practice
Common English phrases
Famous movie quotes
Regular updates to keep content fresh

3. Progress Tracking

Detailed pronunciation analytics
Historical performance tracking
Progress visualization through charts
Personalized feedback on areas for improvement

4. Interactive Learning Tools

Voice recording capabilities
Playback comparison with native speakers
Adjustable difficulty levels
Comprehensive phoneme analysis

Technology Stack

Frontend: Next.js 14+
UI Framework: Tailwind CSS with shadcn/ui
Speech Processing:
- OpenAI Whisper API for transcription
- SpeechAce API for pronunciation assessment
- Claude AI for content generation
Authentication: Built-in Next.js authentication
Styling: Tailwind CSS with custom theming
State Management: React hooks
Charts: Chart.js for analytics visualization

Getting Started

ScreenRecording2025-01-12at6 00 40AM-ezgif com-video-to-gif-converter

Clone the repository at https://github.com/enkhbold470/naccent.git

Challenges we ran into

So, we took a step back and looked at what we were building. Along the way, we ran into some challenges—especially with managing phonetic tones. At first, we thought this project would be as simple as calling a bunch of LLM APIs and calling it a day. But we quickly realized that LLMs have their limitations, especially when it comes to handling accents.

Accomplishments that we're proud of

Despite these hurdles, our team pushed through, put in the effort, and managed to deliver an MVP we’re proud of. It wasn’t easy, but every challenge taught us something new and made the final product even stronger.