Nonlinear ISA with Auxiliary Variables for Learning Speech Representations
This work addresses the challenge of disentangling speech signals into independent subspaces for applications like speaker verification and phoneme recognition, representing an incremental advancement in unsupervised speech representation learning.
The paper tackles the problem of learning unsupervised speech representations by extending nonlinear Independent Component Analysis (ICA) to nonlinear Independent Subspace Analysis (ISA) with auxiliary variables, resulting in an algorithm that shows empirical gains in speaker verification and phoneme recognition tasks.
This paper extends recent work on nonlinear Independent Component Analysis (ICA) by introducing a theoretical framework for nonlinear Independent Subspace Analysis (ISA) in the presence of auxiliary variables. Observed high dimensional acoustic features like log Mel spectrograms can be considered as surface level manifestations of nonlinear transformations over individual multivariate sources of information like speaker characteristics, phonological content etc. Under assumptions of energy based models we use the theory of nonlinear ISA to propose an algorithm that learns unsupervised speech representations whose subspaces are independent and potentially highly correlated with the original non-stationary multivariate sources. We show how nonlinear ICA with auxiliary variables can be extended to a generic identifiable model for subspaces as well while also providing sufficient conditions for the identifiability of these high dimensional subspaces. Our proposed methodology is generic and can be integrated with standard unsupervised approaches to learn speech representations with subspaces that can theoretically capture independent higher order speech signals. We evaluate the gains of our algorithm when integrated with the Autoregressive Predictive Decoding (APC) model by showing empirical results on the speaker verification and phoneme recognition tasks.