SpeakLens is a real-time AI assistant designed to enhance video meetings on platforms like Zoom, Google Meet, and Microsoft Teams. It transcribes conversations, identifies speakers, extracts insights, and visualizes meeting intelligence — all in one powerful dashboard.
-
📝 Real-Time Transcription
Converts spoken words to text using state-of-the-art STT models. -
🗣️ Speaker Diarization
Identifies who is speaking using Voice Activity Detection (VAD) and segmentation. -
📋 Intelligent Summarization & QA
Extracts key points, decisions, and action items using transformer-based NLP models. -
🎭 Sentiment & Emotion Analysis
Detects emotional tone using both audio and text-based classification. -
📊 Interactive Dashboard
Displays highlights, action items, and emotional trends in an easy-to-use interface.
| Layer | Technology |
|---|---|
| Frontend | React + TypeScript, Tailwind CSS |
| Audio Handling | WebRTC / MediaRecorder API (in-browser recording & streaming) |
| Backend API | Python + FastAPI |
| Worker Services | Go (for VAD, segmentation, and Redis queue consumers) |
| Model Serving | Hugging Face Inference API or local deployment using transformers + torch |
| Storage | PostgreSQL (logs, metadata), Redis (task queues), MinIO or AWS S3 (audio) |
| Deployment | Docker + Fly.io / Railway / Render |