EE 302: AI-based Digital Signal Processing
This course explores the intersection of digital signal processing (DSP) and artificial intelligence.
Students learn how traditional DSP concepts such as filtering, sampling, and spectral analysis can be
augmented by machine learning and deep learning models. Applications include real-time audio analysis, denoising, and feature extraction. The course includes both theoretical foundations and hands-on coding assignments using Python-based toolkits.
EE 422 / 522: Speech Processing
A comprehensive course covering modern speech signal processing from Fourier analysis to transformer-based ASR systems. Students work hands-on with tools like PyTorch and Huggingface to build and analyze models such as Wav2Vec2 and Whisper. Topics include STFT, mel-spectrograms, LPC synthesis, noise suppression with RNNs, and CTC decoding strategies. The course also introduces HiFi-GAN, a state-of-the-art neural vocoder used for high-quality waveform reconstruction from mel-spectrograms, enabling students to understand speech synthesis pipelines end-to-end.
EE 423: Deep Learning for Signal Processing
This course focuses on applying deep learning models—such as CNNs, RNNs, and attention-based architectures—to audio and signal processing. Students learn to build, train, and optimize deep models for tasks such as classification, generation, and enhancement of signals.
AI 422 / 525: Large Language Models
A graduate-level deep dive into modern transformer-based architectures and their use in text and speech applications. Topics include training and fine-tuning large language models, embedding extraction, prompting, quantization, and Retrieval-Augmented Generation (RAG). Emphasis is placed on practical experimentation using PyTorch and Hugging Face tools.