ASR/NLP Data Labeling & Model Training (Research Project)
I designed and developed a voice recognition system for detecting criminals using Python-based ASR/NLP methods. I implemented machine learning pipelines for model training and performed feature extraction on speech data for improved recognition accuracy. My work included handling labeled audio datasets and training models to classify different speakers based on their vocal features. • Extracted informative features from raw speech recordings and labeled them for supervised learning. • Used TensorFlow, Librosa, and SpeechRecognition libraries to build, train, and evaluate ASR models. • Designed classification tasks to distinguish between criminals and non-criminals from audio samples. • Managed and preprocessed data, ensuring quality and consistency for machine learning analysis.