Real-Time Deepfake Audio Detection System Development

I am looking for a skilled freelancer with expertise in machine learning, audio processing, and real-time system development to enhance my existing deepfake audio detection project. The current model is functional and achieves good accuracy but operates in an offline mode, processing entire audio files. I need the solution to be optimized for real-time detection, capable of analyzing audio streams and providing instant feedback.

Project Goals:
Convert the existing deepfake audio detection system to handle real-time audio streams.
Maintain or improve the detection accuracy of the current model.
Minimize latency to ensure seamless real-time performance.
Current System Overview:
Model: The existing system uses a LSTM-based model for feature extraction and temporal dependency learning, combined with a KNN classifier for final prediction.
Features: Audio features like MFCC, Chroma, Spectral Contrast, and Tonnetz are used for classification.
Dataset: Built and trained on a balanced dataset of real and deepfake audio, containing speech generated using Retrieval-based Voice Conversion techniques.
Performance: The model provides high accuracy but is designed for batch processing, not streaming.