Unsupervised Adaptation for Deep Neural Network using Linear Least Square Method

被引：0

作者：

Hsiao, Roger ^{[1
]}

Ng, Tim ^{[1
]}

Tsakalidis, Stavros ^{[1
]}

Nguyen, Long ^{[1
]}

Schwartz, Richard ^{[1
]}

机构：

[1] Raytheon BBN Technol, 10 Moulton St, Cambridge, MA 02138 USA

来源：

16TH ANNUAL CONFERENCE OF THE INTERNATIONAL SPEECH COMMUNICATION ASSOCIATION (INTERSPEECH 2015), VOLS 1-5 | 2015年

关键词：

deep neural network; unsupervised adaptation; keyword search; SPEAKER ADAPTATION;

D O I：

暂无

中图分类号：

O42 [声学];

学科分类号：

070206 ; 082403 ;

摘要：

In this paper, we propose a novel model based adaptation for deep neural networks based on a linear least Square method. Our proposed algorithm can perform unsupervised adaptation even if the auto transcripts may have 60-70% of word error rate. We evaluate our algorithm on low resource languages. from the the IARPA BABEL program, such as Assamese, Bengali. Haitian Creole, Lao and Zulu. Our experiments focus on unsupervised speaker, dialect and environment adaptation and we show that it can improve both speech recognition and keyword search performance.

引用

页码：2887 / 2891

页数：5

共 22 条

[11]

Karakos D., 2013, P IEEE WORKSH AUT SP

[12]

Karatifiat M., 2013, P INTERSPEECH

[13]

Kingsbury B., 2009, P IEEE INT C AC SPEE

[14] MAXIMUM-LIKELIHOOD LINEAR-REGRESSION FOR SPEAKER ADAPTATION OF CONTINUOUS DENSITY HIDDEN MARKOV-MODELS [J].

LEGGETTER, CJ ;

WOODLAND, PC .

COMPUTER SPEECH AND LANGUAGE, 1995, 9 (02) :171-185

[15]

Liao H, 2013, INT CONF ACOUST SPEE, P7947, DOI 10.1109/ICASSP.2013.6639212

[16]

Miao Yajie, 2014, P INTERSPEECH

[17]

Ng T., 2011, P INTERSPEECH

[18]

Saon G., 2013, P IEEE WORKSH AUT SP

[19] Language-independent and language-adaptive acoustic modeling for speech recognition [J].

Schultz, T ;

Waibel, A .

SPEECH COMMUNICATION, 2001, 35 (1-2) :31-51

[20]

Yao KS, 2012, IEEE W SP LANG TECH, P366, DOI 10.1109/SLT.2012.6424251

← 1 2 3 →