A new method for speaker adaptation using bilinear model

Abstract

In this paper, a novel method for speaker adaptation using bilinear model is proposed. Bilinear model can express both characteristics of speakers (style) and phonemes across speakers (content) independently in a training database. The mapping from each speaker and phoneme space to observation space is carried out using bilinear mapping matrix which is independent of speaker and phoneme space. We apply the bilinear model to speaker adaption. Using adaptation data from a new speaker, speaker-adapted model is built by estimating the style(speaker)-specific matrix. Experimental results showed that the proposed method outperformed eigenvoice and MLLR. In vocabulary-independent isolated word recognition for speaker adaptation, bilinear model reduced word error rate by about 38% and about 10% compared to eigenvoice and MLLR respectively using 50 words for adaptation.

A single-ensemble-based hybrid approach to clutter rejection combining bilinear Hankel with regression

“…; K, is established in Eq. (16). This closedform result is useful for designing clutter rejection filters.…”

Rapid feature space MLLR speaker adaptation with bilinear models

“…Bilinear model can factor out two independent variations for the underlying style and content factors of observations and express them into a model, which is useful in various applications [7,8]. Next, we will briefly introduce the basic concept of bilinear models and investigate the FMLLR estimation with bilinear models.…”

Adaptation of Acoustic Models in Joint Speaker and Noise Space Using Bilinear Models

“…Bilinear models have also been investigated in the speech processing area. Song et al [6] apply the bilinear model to speaker adaptation with two factors defined as speaker (style) and tied state (content), although only the style factor is estimated during adaptation. In Popa et al [7], the authors apply the bilinear model to voice conversion and report good results compared with the Gaussian mixture model (GMM)-based method using a small amount of training data.…”