Advanced Sonic Operations Grid
Voice Recognition and Identification
- Read Audio file : DONE
- Convert to Features
- Extract Features {with MFCC and Mel Spectogram}
- Design a unsupervised neural network with RNN, CNN (for prediction of number of voices) and Auto-encoder with KMeans frontend (for feature enhancement for voice separation).
- Generate data for deep learning by mixing audio files by single actors
- Train network with partial fit
- Deploy with continual learning in android.