I'm a data scientist with 6 years of experience in developing machine learning models, worked on both video and audio models and have done my masters in machine learning..
With my extensive background as a data scientist, I am confident in my ability to deliver a high-quality solution that meets your objectives.
My approach to this project involves training a deep learning network using audio characteristics extracted from call recordings. By leveraging the sound characteristics present in the pitch contours, energy envelopes, and MFCC coefficients, the lie detection system will be trained on a comprehensive dataset of deceitful and non-deceitful recordings.
Below is my proposal:
1. I will carefully curate a dataset of call recordings encompassing both deceitful and non-deceitful recordings
2. Utilizing signal processing techniques, will extract essential acoustic features from the call recordings.
3. I will design and implement a RNN architecture tailored for sequence classification.
4. Using the collected dataset, I will train the RNN model while continually evaluating its performance on validation sets.
5. Once the model achieves satisfactory results, I will deploy it into a real-time system, allowing for seamless classification of incoming audio as truthful or deceptive.
By entrusting this project to me, you can be assured of my commitment to delivering an effective lie detection system. Looking forward to the opportunity to collaborate on this important project.