Edinburgh Research Explorer

Differentiable Pooling for Unsupervised Acoustic Model Adaptation

Research output: Contribution to journalArticle

Original languageEnglish
Pages (from-to)1773-1784
Number of pages12
Journal IEEE/ACM Transactions on Audio, Speech, and Language Processing
Volume24
Issue number10
Early online date24 Jun 2016
DOIs
Publication statusPublished - Aug 2016

Abstract

We present a deep neural network (DNN) acoustic model that includes parametrised and differentiable pooling operators. Unsupervised acoustic model adaptation is cast as the problem of updating the decision boundaries implemented by each pooling operator. In particular, we experiment with two types of pooling parametrisations: learned Lp-norm pooling and weighted Gaussian pooling, in which the weights of both operators are treated as speaker-dependent. We perform investigations using three different large vocabulary speech recognition corpora: AMI meetings, TED talks and Switchboard conversational telephone speech. We demonstrate that differentiable pooling operators provide a robust and relatively low-dimensional way to adapt acoustic models, with relative word error rates reductions ranging from 5–20% with respect to unadapted systems, which themselves are better than the baseline fully-connected DNNbased acoustic models. We also investigate how the proposed techniques work under various adaptation conditions including the quality of adaptation data and complementarity to other feature- and model-space adaptation methods, as well as providing an analysis of the characteristics of each of the proposed approaches.

Download statistics

No data available

ID: 26403247