Industry
EdTech / Speech Recognition
Core_Pipeline
Saamay AI model / Flutter / Python
Accuracy
98%
Latency
20ms
Reciters require frame-by-frame phonetic feedback. Saamay AI replaces the need for physical proximity to a tutor by deploying an edge-computing model capable of detecting minute Arabic phoneme variances in real-time.
Custom fine-tuning of OpenAI’s Whisper on 5,000+ hours of diverse Arabic dialects and Quranic recitation styles to ensure absolute phonetic accuracy.
A Fastapi bridge for real-time audio buffering, ensuring that visual feedback appears within 20ms of the user's vocalization.