Saranga: Carnatic Raga AI

Saranga is an advanced AI-powered audio engine designed for real-time Carnatic raga identification directly from live microphone input. Saranga provides continuous, real-time melodic analysis to help you recognize and understand ragas as the music unfolds.

* Carnatic Raga, Raga Identification, Hindustani Raga, Indian Classical Music, Alapana, Melakarta, Swara.

🎵 Key Features

* Real-Time Identification: Continuously analyzes live vocal melody and displays raga predictions with confidence metrics as you listen or perform.
* 200+ Supported Ragas: Trained on an extensive library of Carnatic ragas, covering both major foundational ragas and rare Melakarta / Janya ragas.
* 100% Offline & Private: All neural network processing, pitch extraction, and separation models run entirely on-device. No internet connection required, zero audio data uploaded, and zero privacy concerns.
* Shruti-Independent Analysis: Powered by neural networks trained to focus strictly on relative melodic structures and swara transitions, working seamlessly across all vocal registers and pitch keys.

⚡ Dual Analysis Engine Modes

Saranga features two specialized processing engines optimized for different musical contexts:

Alapana Mode (Fast Engine)

* Optimized for rapid, real-time prediction with minimal battery usage.
* Ideal for solo vocal Alapanas, Viruthams, Slokas, and solo instrumental renditions where percussion is minimal or absent.

Full Mode (Enhanced Engine)

* Designed for full concert compositions, Kritis, and performances accompanied by percussion (Mridangam, Ghatam, Kanjira, Morsing).
* More accurate than Alapana mode but slower and more battery intensive.