I-Vector Representation Based on GMM and DNN for Audio Classification - 2015

Explore the state-of-the-art I-vector approach for audio classification tasks in this 56-minute lecture by Najim Dehak from MIT CSAIL. Delve into the modeling of variability in Gaussian Mixture Model (GMM) mean components and weights, and discover recent subspace techniques like Non-negative Factor Analysis (NFA) and Subspace Multinomial Model (SMM). Learn how these approaches can be applied to model hidden layer neuron activations in deep neural networks for sequential data recognition tasks such as language and dialect recognition. Gain insights from Dehak's extensive background in artificial intelligence, pattern recognition, and speech processing as he shares his research on extending the I-vector representation to various audio classification problems, including speaker diarization, language recognition, and emotion recognition.